nach oben

Erschienen in:

Open Access 12.05.2023 | Imaging Informatics and Artificial Intelligence

Performance and clinical applicability of machine learning in liver computed tomography imaging: a systematic review

verfasst von: Keyur Radiya, Henrik Lykke Joakimsen, Karl Øyvind Mikalsen, Eirik Kjus Aahlin, Rolv-Ole Lindsetmo, Kim Erlend Mortensen

Erschienen in: European Radiology | Ausgabe 10/2023

Abstract

Objectives

Machine learning (ML) for medical imaging is emerging for several organs and image modalities. Our objectives were to provide clinicians with an overview of this field by answering the following questions: (1) How is ML applied in liver computed tomography (CT) imaging? (2) How well do ML systems perform in liver CT imaging? (3) What are the clinical applications of ML in liver CT imaging?

Methods

A systematic review was carried out according to the guidelines from the PRISMA-P statement. The search string focused on studies containing content relating to artificial intelligence, liver, and computed tomography.

Results

One hundred ninety-one studies were included in the study. ML was applied to CT liver imaging by image analysis without clinicians’ intervention in majority of studies while in newer studies the fusion of ML method with clinical intervention have been identified. Several were documented to perform very accurately on reliable but small data. Most models identified were deep learning-based, mainly using convolutional neural networks. Potentially many clinical applications of ML to CT liver imaging have been identified through our review including liver and its lesion segmentation and classification, segmentation of vascular structure inside the liver, fibrosis and cirrhosis staging, metastasis prediction, and evaluation of chemotherapy.

Conclusion

Several studies attempted to provide transparent result of the model. To make the model convenient for a clinical application, prospective clinical validation studies are in urgent call. Computer scientists and engineers should seek to cooperate with health professionals to ensure this.

Key Points

• ML shows great potential for CT liver image tasks such as pixel-wise segmentation and classification of liver and liver lesions, fibrosis staging, metastasis prediction, and retrieval of relevant liver lesions from similar cases of other patients.

• Despite presenting the result is not standardized, many studies have attempted to provide transparent results to interpret the machine learning method performance in the literature.

• Prospective studies are in urgent call for clinical validation of ML method, preferably carried out by cooperation between clinicians and computer scientists.

Supplementary file1 (PDF 176 KB)

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00330-023-09609-w.

Keyur Radiya and Henrik Lykke Joakimsen share first authorship of this review.

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

3D RA U-Net

3D hybrid residual attention U-shaped neural network

Article

ACM

Auto-context model

AHC Blocks

Attention hybrid connection blocks

ANN

Artificial neural network

ASM

Active shape model

BPSO

Binary particle swarm optimization

CDNN

Convolutional—deconvolutional neural network

CEDCNN

Cascade encoder-decoder CNN

CENet

Contour embedded neural network

CNN

Convolutional neural network

CRF

Conditional random field

Computed tomography

DBN-DNN

Deep belief network-deep neural network

DCT

Discrete cosine transforms

Deep learning

DLA

Deep learning algorithm

DLO

Dice loss

DResU-Net

Deep residual U-net

DRL

Deep reinforcement learning

ELM

Extreme learning machine

FCMC

Fuzzy C-means clustering

FCN

Fully convolutional neural network

FCNN

Fully convolutional neural network

GAN

Generative adversarial network

GDL

Generalized dice loss

GLC U-Net

Global and local contexts composition U-shaped neural network

GTL

Generalized Teverskry loss

GWO

Grey wolf optimization

HCC

Hepatocellular carcinom

HCC

Hepatocellular carcinoma

HDCNN

Hybridized fully convolutional neural network

k-NN

k-nearest neighbor

Machine learning

MOGA

Multi objective genetic algorithm

MPNet

Message passing neural network

MRF

Markov random field

MSCA

Mean-shift clustering algorithm

MW-U-Net

Modality weighted U-net

PCA

Principal component analysis

PNN

Probabilistic neural network

Proceeding paper from conference

R-CNN

Region based convolutional neural network

RES-U-Net

Residual U-net

RFC

Random forest classifier

Reinforcement learning

RPN

Region proposal network

SSD

Single-shot multibox detector

SSD

Support vector machine

TDP

Three-dimensional dual path multiscale convolutional neural network

Teverskry loss

U-NET

U-shaped neural network (referes to the model architecture)

U-RES-Net

U-shaped residual neural network

VGG 16

Visual Geometry group 16 (Personal name of model named after a research group)

VOE

Volume overlap error

Introduction

For several tasks related to medical imaging, ML is emerging as a new reliable tool due to its high performance and a superior capacity to build complex models for making predictions [1]. More than 220 medical devices using ML have been approved in the USA and Europe [2]. This development has increased steadily since 2014. Today, ML software can be considered a medical device [3].

Computer tomography (CT) imaging plays an essential role in diagnostics and post-treatment follow-up in liver diseases [4]. Applying ML-based tools to CT images has shown promising results [5]. It has been tested theoretically for tasks including identification and segmentation of the liver, lesions, blood vessels, and bile ducts in the liver [6], quantification of liver tissue characteristics [7], evaluation of cancer treatment, and prediction of liver disease [8, 9].

A recently published systematic review and meta-analysis demonstrated the diagnostic accuracy of deep learning (DL) in ophthalmology, respiratory medicine, and breast surgery [10]. In addition, a limited literature review has been published in the subfield of ML applied to liver imaging [11‐13]. However, the performance and clinical applicability of ML in liver imaging are not comprehensively addressed in the literature.

A search in PROSPERO—a database of prospectively registered systematic reviews in health and social care [14]—did not reveal any forthcoming publication in this rapidly developing field. We, therefore, conducted a systematic review from a clinical perspective.

This review aims to answer the following questions: (1) How is ML applied in CT liver imaging? (2) How well do ML systems perform in liver CT imaging? (3) What are the clinical applications of ML in liver CT imaging?

Some important part of this article is given in the electronic supplementary material due to length of the article.

Methods

This systematic review was conducted in accordance with the guidelines for the “Preferred Reporting Items for Systematic Reviews and Meta-Analyses” extension for diagnostic accuracy studies statement [15]. A selection and retrieval of studies from the literature was done in accordance with Cochrane handbook for systematic review [16]. A search was conducted in Medline, EMBASE, and Web of Science and included studies published between January 1, 2011, and October 31, 2021. The search string consisted of exploded MeSH-terms, Emtree-terms, and free text to find all studies containing the terms “Artificial intelligence” AND “Computed tomography” AND “liver” (or containing all possible synonyms of all three) in the title, abstract, or keywords. The exact search string was given in the electronic supplementary material.

When considering study quality, we identified characteristics as important given in the electronic supplementary material. The suggested list is comprehensive, and studies might be quite informative with minimal risk of bias, without meeting all requirements [17]. Yet, if a study followed only few of the characteristics, it was not considered well-documented for clinical use.

Results

The search was conducted in two phases, one in October 2020 and one in October 2021. There were 191 studies included for review. The selection process is illustrated in the PRISMA flow diagram in Fig. 1 [18]. The selected studies are summarized in Table 1 and details given in the electronic supplementary material.

Table 1

Description of included studies with detail about included in group, document type A = article and PP = proceeding paper, type of journal – medical or non-medical, AI method used, amount of test set, external validity status, ML to clinician, using of publicly available datasets

Study	Group	Type of paper	Type of journal	Type of ML method used	Test Set	External validity	ML to clinician	Public dataset
Mubashir et al 2019 [19]	Liver segmentation	A	Non-medical	DBN-DNN	15 + 15 (2 open datasets)	No	No	Yes
Mubashir et al 2019 [19]	Liver segmentation	PP	Non-medical	CNN	5	No	No	Yes
Ahn et al 2020 [20]	Liver segmentation	A	Medical	3d U-net, DLA DeepLabV3	20 CT series and 60 CT series	Yes	Yes	Yes
Bhavya et al 2018 [21]	Liver segmentation	PP	Medical	Real AdaBoost classifier	70	Yes	No	Yes
Albishri 2019 [22]	Liver segmentation	PP	Non-medical	Cascade U-net	32 patient’s data (unclear about the total number of data)	No	No	No
Ali 2017 [23]	Liver segmentation	PP	Non-medical	SVM	50	No	Yes	Yes
Alirr 2020 [24]	Liver segmentation	A	Non-medical	U NET + level set	20	Yes	Yes	Yes
Astono 2018 [25]	Liver segmentation	PP	Non-medical	CNN-adjacent net	10 scans	Yes	No	Yes
Ben-Cohen 2016 [26]	Liver segmentation	PP	Non-medical	FCN-VGG 16-layer net	70 CT sets	Yes	No	Yes
Bevilacqua et al 2017 [27]	Liver segmentation	PP	Non-medical	ANN classifier by using mono-objective genetic algorithm (GA)	Not mentioned	Yes	No reference	No
Bhole 2011 [28]	Liver segmentation	PP	Non-medical	MRF	10 series of 10 patients	No	No	Yes
Budak et al 2020 [29]	Liver segmentation	A	Medical	CEDCNN	5 sets (589 slices)	No	No	Yes
Cai 2019 [30]	Liver segmentation	A	Medical	Adaptive scale-kernel fuzzy clustering models	They have created 3 model from different dataset, and the fourth model for fine tuning. Difficult to give in number of patients used in for training. As they have used transfer learning from one model to another model where adding some more data	Yes	No	Yes
Chen 2019 [31]	Liver segmentation	PP	Non-medical	MPNet, adversarial densely connected network and a deep FCNN	10	Yes	No	Yes
Chen et al 2019 [31]	Liver segmentation	A	Medical	Channel-U-net, a spatial channel wise convolutional neural network		Yes	No	Yes
Chlebus 2018 [32]	Liver segmentation	A	Non-medical	FCNN and object based postprocessing	Not mentioned	Yes	Yes	No
Choi et al 2018 [33]	Liver segmentation	A	Medical	CNN	150 images	Yes	Yes	No
Chung 2020 [34]	Liver segmentation	A	Non-medical	CENet	28 volumes	No	No	Yes
Danciu 2013 [35]	Liver segmentation	A	Medical	SVM	76 patients (20–40 images of the liver per patient)	No	Yes	Yes
Danciu 2012 [36]	Liver segmentation	PP	Non-medical	3D DCT and SVM	26,608 images of 70 CT scans from 40 patients	No	No	Yes
Delmoral 2019 [37]	Liver segmentation	PP	Non-medical	CNN	31CT	Yes	No	Yes
Dong 2020 [38]	Liver segmentation	A	Non-medical	HDCNN	50 patients, 1272 images	No	No	Yes
Dou et al 2016 [39]	Liver segmentation	A	Medical	3D deeply supervised network	5 patients for testing, 5 patients for validation	Yes	No	No
Guo 2019 [40]	Liver segmentation	A	Medical	FCNN	–	Yes	No reference	No
He et al 2016 [41]	Liver segmentation	A	Medical	Ada Boost guided active shape model	(1) 46 lesions for validation, 46 lesions for testing; (2) not specified	Yes	No	Yes
Heker 2019 [42]	Liver segmentation	PP	Non-medical	Cascade U-net	Not specified	Yes	No	No
Hu 2016 [43]	Liver segmentation	A	Medical	3D-CNN	10 patients	Yes	No	Yes
Huang et al 2012 [44]	Liver segmentation	PP	Non-medical	ELM	Not specified	Yes	No	No
Ji 2013 [45]	Liver segmentation	A	Non-medical	ACM	Not specified	Yes	No	No
Jiang 2018 [46]	Liver segmentation	A	Medical	Registration based organ positioning, FCMC, ELM, ACM	Not specified	No	No	Yes
Jiang 2019 [47]	Liver segmentation	A	Non-medical	3D FCN, AHCBlocks	12 images for validation, 1 for testing	Yes	No	Yes
Jin 2017 [48]	Liver segmentation	PP	Non-medical	FCN-U-net	25 patients for testing, 25 patients for validation	Yes	No	Yes
Kavur et al 2020 [49]	Liver segmentation	A	Medical	CNN	20 patients	No	No	Yes
Kumar 2016 [50]	Liver segmentation	A	Medical	Feedforward neural network	Not mentioned	No	No reference	No
Chung 2020 [34]	Liver segmentation	A	Non-medical	CNN (CENet)	150 images	No	No	No
Zheng et al 2019 [51]	Liver segmentation	PP	Non-medical	GAN + deep atlas prior	28 volumes	No	No	Yes
Zhang, Y. 2018 [52]	Liver segmentation	PP	Non-medical	FCN + CRF	5 scans/patients	No	No	Yes
Zhang, L. 2018 [53]	Liver segmentation	PP	Non-medical	U-net	76 patients (20–40 images of the liver per patient)	No	No	Yes
Xu 2019 [54]	Liver segmentation	PP	Non-medical	RES-U-Net, connected components analyzing and CRF	26,608 images of 70 CT scans from 40 patients	No	No	Yes
Xi 2020 [55]	Liver segmentation	A	Non-medical	Cascade U-RES-Net (CNN + Dlo + TL + GDL + GTL)	70 image sets	No	No	Yes
Xin 2020 [56]	Liver segmentation	A	Non-medical	CNN	32 patients, 643 slices containing lesions	No	No	No
Xia 2019 [57]	Liver segmentation	A	Non-medical	CNN Deep Adversarial Networks (DeepLab-v3) + weighted loss function	8800 images	No	No	Yes
Winkel et al 2020 [58]	Liver segmentation	A	Medical	DRL (CNN + RL)	20 sets, 6 sets per patient	No	Yes	No
Wang et al 2019 [59]	Liver segmentation	PP	Non-medical	CNN	28 patients	No	No	Yes
Tian 2019 [60]	Liver segmentation	PP	Non-medical	U-net (GLC-U-net, CNN)	50 patients, 1272 images	No	No	Yes
Tang 2017 [61]	Liver segmentation	PP	Non-medical	FCN (+ level set)	5 patients for testing, 5 patients for validation	No	No	No
Seo et al 2020 [62]	Liver segmentation	A	Non-medical	CNN (modified U-Net)	(1) validation: 5 patients; 2550 images; testing: 35 patients; 16,125 images; (2) 5 patients, 525 images	Yes	No	Yes
Selvi 2014 [63]	Liver segmentation	PP	Non-medical	High-order neural network	Not specified	No	No	No
Selvathi et al 2013 [64]	Liver segmentation	PP	Non-medical	ELM + FCMC	Not specified	No	No	No
Sayed 2016 [65]	Liver segmentation	PP	Non-medical	Fuzzy clustering + GWO (Liver and liver lesion segmentation); SVM (liver disease classification: benign/malignant)	Not provided	No	No	No
Sakboonyara 2019 [66]	Liver segmentation	PP	Non-medical	U-Net, 2D (CNN/ FCN)	5 images	No	No	Yes
K S et al 2018 [67]	Liver segmentation	PP	Non-medical	U-Net and 3D CRF	–	No	No	No
Raj 2016 [68]	Liver segmentation	PP	Non-medical	SVM	Not specified	No	No	No
Rafiei 2018 [69]	Liver segmentation	PP	Non-medical	FCN + CRF	10 patients	No	No	Yes
Qin et al 2018 [70]	Liver segmentation	A	Medical	CNN (SBBS-CNN, based on CifarNet)	Not specified	No	No	Yes
Ponnoprat et al 2020 [71]	Liver segmentation	A	Non-medical	U-Net for segmentation + CRF for post-processing + SVM for classification	17 patients, 2042 images	No	No	No
Ouhmich 2019 [72]	Liver segmentation	A	Medical	U-Net	Not specified	No	No	No
Ng et al 2020 [73]	Liver segmentation	A	Medical	Gaussian mixture model and U-Net	6 patients (fivefold cross validation)	No	No	No
Nayak et al 2019 [74]	Liver segmentation	A	Medical	Segmentation: region-growing; classification: SVM	Not specified	No	No	Yes
Mukherjee et al 2013 [75]	Liver segmentation	PP	Non-medical	SVM + PCA	Not specified	No	No	No
Morshid et al 2019 [76]	Liver segmentation	A	Medical	Segmentation: U-Net, 2D; prediction: RFC	Not specified	Yes	Yes	Yes
Mohagheghi and Foruzan 2020 [77]	Liver segmentation	A	Medical	U-Net	12 images for validation, 1 for testing	No	No	Yes
Mofrad 2014 [78]	Liver segmentation	A	Medical	Classification: SVM, k-NN	1 patient	No	No	No
Meng L 2020 [79]	Liver segmentation	A	Non-medical	TDP-CNN + CRF (post-processing)	25 patients for testing, 25 patients for validation	No	No	Yes
Luo and Li 2014 [80]	Liver segmentation	PP	Non-medical	SVM	1 image, 1 patient	No	No	Yes
Lu et al 2017 [81]	Liver segmentation	A	Medical	CNN + graph cut	SLiver07: 10 patients; 3D-IRCADb: 20 patients	Yes	Yes	Yes
Selvaraj 2013 [82]	Liver segmentation	PP	Non-medical	Lesion segmentation: FCM; feature selection: BPSO; classification: PNN	15 images	No	No	No
Li 2014 [83]	Liver segmentation	PP	Non-medical	PCA + ASM + k-NN	5 whole body scans, 5 abdominal contrast-enhanced scans	No	No	Yes
Li et al 2018 [84]	Liver segmentation	A	Non-medical	H-Dense U-Net	LiTS 2017: 70 patients; 3D-IRCADb: cross-validation	Yes	No	Yes
Liu et al 2019 [85]	Liver segmentation	A	Non-medical	U-Net + graph cut	20 patients	No	No	Yes
Linguraru et al 2012 [86]	Liver segmentation	A	Non-medical	SVM	LiTS 2008: 4 patients; SLiver07: 10 patients	Yes	No	Yes
Astono et al 2018 [25]	Liver segmentation	A	Non-medical	Adjacent Net	Validation: 2 patients; test: 2 × 10 patients	No	No	Yes
Afifi and Nakaguchi 2015 [87]	Liver segmentation	PP	Non-medical	MSCA + graph cut in detection	Not specified	No	No reference	No
Roth 2020 [88]	Liver segmentation	PP	Non-medical	U-net	70	Yes	No	Yes
Tran 2021 [89]	Liver segmentation	A	Non-medical	U-Net multilayer	30 scan (15 CT from each datasets)	Yes	No	Yes
Xu et al 2020 [90]	Liver segmentation	PP	Non-medical	pyramidal U-net	fourfold cross-validation	No	No reference	0
Yu et al 2021 [91]	Liver segmentation	A	Non-medical	DResU-Net	25	Yes	No	Yes
Zhang, Y et al 2021 [92]	Liver segmentation	A	Non-medical	RECIST NET	46	No	No	0
Zhang, Yao 2021 [92]	Liver segmentation	PP	Non-medical	CNN (deep attentive refinement network)	70	Yes	No	Yes
Ayalew 2021 [93]	Liver segmentation	A	Non-medical	U-net	392 images	No	No	Yes
Chen et al 2020 [94]	Liver segmentation	A	Medical	U-net	300 images	Yes	No	Yes
Chung 2021 [95]	Liver segmentation	A	Medical	CNN	80 patients	No	No	Yes
Elmenabawy et al 2020 [96]	Liver segmentation	PP	Non-medical	CDNN	33 patients	No	No	Yes
Fan 2020 [97]	Liver segmentation	A	Non-medical	U-net multi-scale attention net	70 patients	No	No	Yes
He et al 2021 [98]	Liver segmentation	A	Medical	U-net (3D RA-U-Net)	252 images, 63 patients	Yes	Yes	Yes
Kwon 2020 [99]	Liver segmentation	PP	Non-medical	U-net	70 patients	No	No	Yes
Lei 2020 [100]	Liver segmentation	PP	Non-medical	U-Net / V-Net	31 patients	No	No	Yes
Afifi 2015 [87]	Lesion detection	PP	Non-medical	Mean-shift segmentation algorithm	15 patients 169 lesions	No	No	No
Ali et al 2017 [23]	Lesion detection	PP	Non-medical	SVM	50	No	Yes	Yes
Ben-Cohen 2016 [26]	Lesion detection	PP	Non-medical	FCN-VGG 16 layer net	70 CT sets	Yes	No	Yes
Ben-Cohen 2018 [101]	Lesion detection	A	Non-medical	FCN8 net-VGG 16 layer net, and sparsity-based dictionary learning (localized patch level analysis usin superpixel sparse based classification	7 data sets	No	No	Yes
Bevilacqua et al 2017 [102]	Lesion detection	PP	Non-medical	ANN clssifier by using mono-objective GA	Not mentioned	Yes	No reference	No
Bevilacqua et al 2017 [102]	Lesion detection	PP	Non-medical	ANN clssifier by using MOGA	Not mentioned	No	No reference	No
Chen et al 2019 [103]	Lesion detection	PP	Non-medical	Dual-attention dilated residual network-weakly supervised localization	10 + 10 dataset from Sliver	No	No	Yes
Frid-Adar 2017 [104]	Lesion detection	PP	Non-medical	Multi-class patch based CNN system	(1) Validation: 5 patients, 2550 images, testing: 35 patients, 16,125 images; (2) 5 patients, 525 images	No	No	Yes
Furuzuki et al 2019 [105]	Lesion detection	PP	Non-medical	Faster R-CNN	Not specified	No	No	No
Gong et al 2019 [106]	Lesion detection	A	Medical	R-CNN, partial least square regression discriminant analysis model	5images	No	Yes	Yes
Huang et al 2013 [107]	Lesion detection	PP	Non-medical	Kernel-based ELM with classifier	17 patients, 2042 images	No	No	No
Jin 2017 [108]	Lesion detection	PP	Non-medical	CNN + ensemble learning	SLiver07: 10 patients; 3D-IRCADb: 20 patients	No	No	Yes
Jin 2015 [109]	Lesion detection	PP	Non-medical	Improved back propagation neural network	15 images	No	No	No
Kim 2019 [110]	Lesion detection	PP	Non-medical	Cycle-Consistent CNN	Not specified	No	No	No
Vivanti 2017 [111]	Lesion detection	A	Medical	CNN + RFC	Not specified	No	No	No
Tao et al 2019 [112]	Lesion detection	PP	Non-medical	FCN + RPN	∼ 5000 images for testing and ∼ 5000 images for validation	No	No	Yes
Liang et al 2019 [113]	Lesion detection	PP	Non-medical	CNN (recurrant with long short-term memory)	(1) validation: 175; test: 153; (2) validation: 175; test: 153	No	No	No
Lee 2018 [114]	Lesion detection	PP	Non-medical	SSD	fivefold cross-validation	No	No	No
Afifi 2015 [87]	Lesion detection	PP	Non-medical	MSCA (+ graph cut in detection)	Not specified	No	No reference	No
Yang et al 2021 [115]	Lesion detection	A	Non-medical	CNN	337	Yes	No reference	0
Zhou et al 2021 [116]	Lesion detection	A	Medical	CNN	1/4 of lesion was used for testset	No	No	0
Albishri 2019 [22]	Lesion segmentation	PP	Non-medical	Cascade U-net	32 patients data (unclear about the total number of data)	No	No	No
Alirr 2020 [24]	Lesion segmentation	A	Non-medical	U NET + level set	20	Yes	Yes	Yes
Almotairi 2020 [117]	Lesion segmentation	A	Non-medical	Modified Seg Net	20 CT from local hospital	No	No	Yes
Anter 2019 [118]	Lesion segmentation	A	Medical	Fast fuzzy C-means and adaptive watershed algorithm	30	Yes	No	Yes
Budak 2020 [29]	Lesion segmentation	A	Medical	CEDCNN	5 sets (589 slices)	No	No	Yes
Chen, L. 2019 [103]	Lesion segmentation	PP	Non-medical	MPNet, adversarial densely connected network and a deep FCNN	10	Yes	No	Yes
Chen, X. et al 2019 [119]	Lesion segmentation	PP	Non-medical	FED-Net	10 CT series	No	No	Yes
Chen, Y. et al 2019 [31]	Lesion segmentation	A	Medical	Channel-U -net		Yes	No	Yes
Chlebus 2018 [32]	Lesion segmentation	A	Non-medical	FCNN- and object-based postprocessing	Not mentioned	Yes	Yes	No
Delmoral 2019 [37]	Lesion segmentation	PP	Non-medical	CNN	31CT	Yes	No	Yes
Deng 2019 [120]	Lesion segmentation	A	Medical	Dynamic regulation to functional parameters over iterations using the 3D CNN	20 sets, 6 sets per patient	Yes	No	No
Dong 2020 [38]	Lesion segmentation	A	Non-medical	HDCNN	50 patients, 1272 images	No	No	Yes
Heker 2019 [42]	Lesion segmentation	PP	Non-medical	Cascade U-net	Not specified	Yes	No	No
Huang et al 2013 [107]	Lesion segmentation	PP	Non-medical	Kernel-based ELM with classifier	17 patients, 2042 images	No	No	No
Huang et al 2014 [121]	Lesion segmentation	PP	Non-medical	Random feature subspace ensemble–based ELM	6 patients (fivefold cross-validation)	No	No	No
Jiang 2018 [46]	Lesion segmentation	A	Medical	Registration based organ positioning, fuzzy C means clustering and ELM, ACM	Not specified	No	No	Yes
Jiang 2019 [47]	Lesion segmentation	A	Non-medical	3D FCN composed of multiple AHCBlocks	12 images for validation, 1 for testing	Yes	No	Yes
Kadoury 2015 [122]	Lesion segmentation	A	Medical	Grassmanian kernels and discriminant manifold, CRF	5 whole body scans, 5 abdominal contrast-enhanced scans	Yes	No	Yes
Almotairi 2020 [117]	Lesion segmentation	A	Non-medical	Modified SegNet	3 patients, 454 images for testing and 45 for validation	No	No	Yes
Zhou 2013 [123]	Lesion segmentation	PP	Non-medical	CNN	16 patients	Yes	No	No
Zhang, Yue et al 2020 [124]	Lesion segmentation	A	Non-medical	2D U-net + 3D FCN and unsupervised fuzzy c-means clustering	(1) 36 images; (2) 70 images	Yes	No	Yes
Zhang, Yi 2020 [125]	Lesion segmentation	A	Non-medical	CNN	(1) 9 sets for testing, 20 for verification/validation; (2) 5 for testing, 5 for verification	Yes	No	Yes
Zhang, Xing 2011 [126]	Lesion segmentation	PP	Non-medical	SVM + traditional feature extraction	Not specified	No	No	Yes
Xi 2020 [55]	Lesion segmentation	A	Non-medical	Cascade U-RES-Net (CNN + Dlo + TL + GDL + GTL)	70 image sets	No	No	Yes
Xin 2020 [56]	Lesion segmentation	A	Non-medical	CNN	32 patients, 643 slices containing lesions	No	No	No
Wu 2019 [127]	Lesion segmentation	PP	Non-medical	MW-U-net	15 patients, 100–135 images per patient	No	No	Yes
Wei et al 2019 [128]	Lesion segmentation	PP	Non-medical	CNN (HMMMNet)	(1) LiTS 2017: 26 patients; (2) decathlon: 70 (not specified in the article—found at medicaldecathlon.com)	No	No	Yes
Vorontsov et al 2018 [129]	Lesion segmentation	PP	Non-medical	CNN (FCN)	15 patients	No	No	Yes
Vorontsov et al 2017 [130]	Lesion segmentation	A	Non-medical	MLP	5 patients	No	No	No
Vivanti 2017 [111]	Lesion segmentation	A	Medical	CNN + RFC	Not specified	No	No	No
Vivanti 2018 [129]	Lesion segmentation	A	Non-medical	CNN (× 2: global and individual)	Not specified	No	No	No
Todoroki 2019 [131]	Lesion segmentation	PP	Non-medical	CNN	266,000, 282,000, and 215,000 patch images (tested once each)	No	No	No
Sun 2017 [132]	Lesion segmentation	PP	Non-medical	FCN	(1) 3D-IRCADb: 40 images; (2) JDRD: 36 images	Yes	No	Yes
Shimizu 2013 [133]	Lesion segmentation	A	Non-medical	U-Boost	Not specified	No	No	No
Seo 2020 [62]	Lesion segmentation	A	Non-medical	Modified U-Net	(1) Validation: 5 patients; 2550 images; testing: 35 patients, 16,125 images; (2) 5 patients, 525 images	Yes	No	Yes
Selvathi et al 2013 [64]	Lesion segmentation	PP	Non-medical	ELM + FCMC	Not specified	No	No	No
Sayed 2016 [65]	Lesion segmentation	PP	Non-medical	Segmentation: fuzzy clustering + GWO; classification: SVM	Not provided	No	No	No
Raj 2016 [68]	Lesion segmentation	PP	Non-medical	SVM	Not specified	No	No	No
Ouhmich 2019 [72]	Lesion segmentation	A	Medical	U-Net	Not specified	No	No	No
Morshid 2019 [76]	Lesion segmentation	A	Medical	Segmentation: U-Net; prediction: RFC	Not specified	Yes	Yes	Yes
Moawad et al 2020 [134]	Lesion segmentation	A	Medical	U-Net	Not specified	No	Yes	No
Meng et al 2020 [79]	Lesion segmentation	A	Non-medical	TDP-CNN + CRF	25 patients for testing, 25 patients for validation	No	No	Yes
Selvaraj 2013 [82]	Lesion segmentation	PP	Non-medical	Segmentaion: FCM; feature selection: BPSO; classification: PNN	15 images	No	No	No
Li et al 2018 [84]	Lesion segmentation	A	Non-medical	H-DenseU-Net	LiTS 2017: 70 patients; 3D-IRCADb: cross-validation	Yes	No	Yes
Radu et al 2020 [135]	Lesion segmentation	A	Medical	CNN	30 CT for testing	Internal	No	0
Roth 2020 [88]	Lesion segmentation	PP	Non-medical	U-net	70	External	No	Yes
Xin 2020 [56]	Lesion segmentation	A	Medical	CNN	643 slice for test	No	No	0
Tran 2021 [89]	Lesion segmentation	A	Non-medical	U-Net multilayer	30 scan (15 ct from each datasets)	Yes	No	Yes
Haq et al 2021 [136]	Lesion segmentation	PP	Non-medical	Resnet R-CNN	70	Yes	No	Yes
Yang et al 2021 [115]	Lesion segmentation	A	Non-medical	CNN	337	Yes	No reference	0
Yu et al 2021 [91]	Lesion segmentation	A	Non-medical	DResU-Net	25	Yes	No	Yes
Zhang, Yao 2021 [92]	Lesion segmentation	PP	Non-medical	CNN (deep attentive refinement network)	70	Yes	No	Yes
Anil 2021 [137]	Lesion segmentation	A	Non-medical	MDCN + FRN	NA	No	No	Yes
Aslam et al 2021 [138]	Lesion segmentation	A	Non-medical	ResU-Net	NA	No	No	Yes
Ayalew 2021 [93]	Lesion segmentation	A	Non-medical	U-net	392 images	No	No	Yes
Chen et al 2021 [94]	Lesion segmentation	A	Medical	U-net	300 images	Yes	No	Yes
Dey 2020 [139]	Lesion segmentation	PP	Non-medical	CNN	70 patients	No	No	Yes
Elmenabawy et al 2020 [96]	Lesion segmentation	PP	Non-medical	CDNN (conv-deconv neural net)	33 patients	No	No	Yes
Fan 2020 [97]	Lesion segmentation	A	Non-medical	U-net (multi-scale attention net)	70 patients	No	No	Yes
Hamard et al 2020 [140]	Lesion segmentation	A	Medical	NA (off the shelf product)	44	Yes	Yes	No
He et al 2021 [98]	Lesion segmentation	A	Medical	U-net (3D RA-U-Net)	252 images, 63 patients	Yes	Yes	Yes
Kwon 2020 [99]	Lesion segmentation	PP	Non-medical	U-net	70 patients	No	No	Yes
Adcock 2014 [18]	Classification	A	Non-medical	SVM-LibSVM (multidimensional scaling (CMDS)	Not mentioned	No	No	No
AmirHosseini 2019 [141]	Classification	A	Non-medical	Fuzzy inference system	7 patients for HCC segmentation, 20 patients for liver segmentation	No	No	Yes
Balagourouchetty et al 2020 [142]	Classification	A	Non-medical	GoogLeNet based Ensemble FCNet Classifier	Not mentioned exactly number but they have 10% data to test set and have used tenfold cross-validation	No	No	Yes
Bevilacqua et al 2017 [27]	Classification	PP	Non-medical	ANN classifier by using mono-objective genetic algorithm (GA)	Not mentioned	Yes	No reference	No
Cao et al 2020 [143]	Classification	A	Medical	Multiphase convolutional dense network	42CT (12 from local and 20 + 10 from Sliver07)	No	Yes	Yes
Chen et al 2019 [103]	Classification	PP	Non-medical	Dual-attention dilated residual network—weakly supervised localization	10 + 10 dataset from Sliver	No	No	Yes
Das 2019 [144]	Classification	A	Medical	Watershed Gaussian–based deep learning, DNN	32 patients, 643 slices containing lesions	No	No	No
Devi 2020 [145]	Classification	A	Non-medical	Region growing process for liver segmentation = > kernalized fuzzy C-means algorithm for lesion extraction, SVM-based classifier for classification of tumor	28 patients	No	No	Yes
Jiang 2013 [146]	Classification	A	Non-medical	SVM-multi instance learning	1 patient	No	No	No
Jin 2016 [147]	Classification	PP	Non-medical	Improved random forest	1 image, 1 patient	No	No	Yes
Kashala 2020 [148]	Classification	A	Non-medical	FireNet module in SqueezeNet and obtained FCN as well-developed new particle swarm optimization called NPSO	LiTS 2017: 70 patients; 3D-IRCADb: cross-validation	No	No	Yes
Khalili et al 2020 [149]	Classification	A	Non-medical	CNN	Validation: 2 patients; test: 2 × 10 patients	No	Yes	Yes
Kumar 2013 [150]	Classification	A	Non-medical	Probabilistic neural network	150 images	No	No	No
Kutlu 2019 [151]	Classification	A	Non-medical	CNN with alexnet architecture, DWT (Discrete Wavelet Transform) and Long short-terms memory networks	30% of data for test	No	No	No
Yasaka et al 2018 [152]	Classification	A	Medical	CNN	100 patients/image sets	Yes	Yes	No
Xin et al 2020 [56]	Classification	A	Non-medical	CNN	32 patients, 643 slices containing lesions	No	No	No
Sreeja and Hariharan 2017 [153]	Classification	PP	Non-medical	SVM + Naive Bayes classifier	Not specified	No	No	No
Shi et al 2020 [154]	Classification	A	Medical	CNN	One per lesion	No	No	No
Selvathi et al 2013 [64]	Classification	PP	Non-medical	ELM + FCMC	Not specified	No	No	No
Sayed 2016 [65]	Classification	PP	Non-medical	Fuzzy clustering + GWO (liver and liver lesion segmentation); SVM (liver disease classification: benign/malignant)	Not provided	No	No	No
Romero et al 2019 [155]	Classification	PP	Non-medical	CNN (FCN × 2)	(1) 46 lesions for validation, 46 lesions for testing; (2) not specified	No	No	Yes
Renukadevi and Karunakaran 2020 [156]	Classification	A	Non-medical	DBN + GOA	Not specified	Yes	No	Yes
Rajathi 2019 [157]	Classification	A	Non-medical	WOA-SA + SVM + k-NN + RFC	21 patients	No	No	No
Raj 2016 [68]	Classification	PP	Non-medical	SVM	Not specified	No	No	No
Ponnoprat et al 2020 [71]	Classification	A	Non-medical	U-Net for segmentation + CRF for post-processing + SVM for classification (w GHI kernel)	17 patients, 2042 images	No	No	No
Peng et al 2020 [158]	Classification	A	Medical	CNN (ResNet50)	ZHHAJU: 89; SYUCC: 138 patients	Yes	No	No
Özyurt et al 2019 [159]	Classification	A	Non-medical	CNN	34	No	No	No
Ouhmich et al 2019 [72]	Classification	A	Medical	U-Net	Not specified	No	No	No
Nayak et al 2019 [74]	Classification	A	Medical	Segmentation: region-growing; classification: SVM	Not specified	No	No	Yes
Mukherjee et al 2013 [75]	Classification	PP	Non-medical	SVM + PCA	Not specified	No	No	No
Mofrad et al 2014 [78]	Classification	A	Medical	SVM (classification), k-NN (classification)	1 patient	No	No	No
Mala et al 2015 [160]	Classification	A	Non-medical	PNN, LVQ, BPN	20 patients, ca. 20 images per patient	No	No	No
Maaref et al 2020 [161]	Classification	A	Medical	2D CNN (Inception-Net, modified)	CLASSIFICATION: 20 patients for validation, 41 for testing; PREDICTION: 12 patients for validation, 24 for testing	No	No	No
Selvaraj 2013 [82]	Classification	PP	Non-medical	FCM (lesion segmentation) + BPSO (feature selection) + PNN (classification)	15 images	No	No	No
Li et al 2019 [162]	Classification	PP	Non-medical	BPN (+ PCA preprocessing)	57 (tenfold cross-validation)	No	No	No
Liang et al 2018 [163]	Classification	PP	Non-medical	CNN (ResNet w/ global and local pathways—for segmentation) + SVM (classification)	(1) Validation: 115, test: 96; (2) validation: 93, test: 110	No	No	No
Liang et al 2018 [163]	Classification	PP	Non-medical	CNN (ResNet w/ global and local pathways w/ bi-directional long short-term memory—for segmentation) + SVM (classification)	(1) Validation: 115, test: 96; (2) validation: 93, test: 110	No	No	No
Xin et al 2020 [56]	Classification	A	Medical	CNN	643 slices for test	No	No	0
Thuring et al 2020 [164]	Classification	A	Medical	Random forest and CNN	70 patients	No	Yes	Yes
Wang et al 2021 [165]	Classification	A	Medical	Nodule Net and HCCNet	385 from same hospital, external test set with 556 patients	Yes	Yes	0
Wang et al 2020 [166]	Classification	A	Non-medical	CNN (Siamese cross contrast neural network)	67 patients	No	No	0
Xu et al 2021 [167]	Classification	A	Medical	Random forest	tenfold cross-validation	No	No reference	0
Zhang et al 2020 [168]	Classification	A	Medical	GLM	57	No	No	0
Zhou et al 2021 [116]	Classification	A	Medical	CNN	1/4 of lesion was used for test set	No	No	0
Giannini et al 2020 [169]	Classification	A	Medical	Gaussian Naive Bayes classifier	10 patients, 33 tumors/metastases	No	No	No
Homayounieh et al 2020 [170]	Classification	A	Medical	Random forest	103 patients w benign (60/103) or malignant (43/103) tumors	No	No	No
Mao et al 2020 [171]	Classification	A	Medical	Gradient boosting (XGBoost)	60 patients	No	No	No
Mokrane et al 2020 [172]	Classification	A	Medical	Random forest	36 patients	Yes	No reference	No
Budai et al 2020 [173]	Miscellaneous	A	Medical	RF and SVM, K-means clustering	Independent validation dataset from > Sliver07(20 dataset), > MICCAI 2017 (LiTS) 131 scans	No	No reference	Yes
Choi et al 2018 [33]	Miscellaneous	A	Medical	CNN	150 images	Yes	Yes	No
Huo et al 2019 [174]	Miscellaneous	A	Medical	DCNN and morphological operation for attenuation and SS-Net (a DCNN model)	Not specified	Yes	Yes	Yes
Kayaalti et al 2014 [175]	Miscellaneous	A	Non-medical	SVM and K-nearest neighbors for classifying the images		No	No	No
Yasaka et al 2018 [176]	Miscellaneous	A	Medical	CNN	100 portal phase images from 100 patients	No	Yes	No
Son et al 2020 [177]	Miscellaneous	A	Medical	U-net	Not specified	No	Yes	No
Yin et al 2021 [178]	Miscellaneous	A	Medical	CNN	fivefold cross-validation	No	No	0
Ahmadi et al 2016 [179]	Miscellaneous	A	Medical	FCM and GA	Test dataset 1: 150 patients, test dataset 2: 50 patients	No	No	No
Ben-Cohen et al 2018 [180]	Miscellaneous	PP	Non-medical	U-net base—using unlabeled data features in supervised network	Test set 1: 421 patients. Test set 2: 298 (other institutions). Test set 3: 172 patients (from tertiary referral hospitals	No	No	No
Bevilacqua et al 2017 [27]	Miscellaneous	PP	Non-medical	ANN classifier by using mono-objective genetic algorithm (GA)	Not mentioned	Yes	No reference	No
Conze et al 2017 [181]	Miscellaneous	A	Medical	Scale adaptive super voxel-based random forests	Not specified	No	No reference	No
Fu et al 2019 [6]	Miscellaneous	A	Non-medical	U net-with multi stream feature fusion and multi scale dilated convolution, author called it M-Net	Not specified	No	No	No
Gensure et al 2012 [182]	Miscellaneous	A	Medical	SVM	Not provided	No	No	No
Huang et al 2018 [183]	Miscellaneous	A	Medical	3d U-Net	Not specified	Yes	Yes	Yes
Kumar et al 2016 [50]	Miscellaneous	A	Medical	SVM, weighted nearest neighbor	50 CT images were used from ImageCLEF 2014 with tenfold cross-validation	No	No	Yes
Zhang et al 2018 [52]	Miscellaneous	A	Non-medical	Fuzzy connectedness (fuzzy logic)	(1) VascuSynth: not eligible; (2) 3D-IRCADb: not provided; (3) Sliver07: 10 patients	Yes	No	Yes
Zeng et al 2016 [184]	Miscellaneous	A	Medical	ELM	100,000 images in total (training + testing data)	No	No	No
Yu et al 2019 [185]	Miscellaneous	PP	Non-medical	CNN	6 cases (+ 3 for validation); slices per case range: 135–500	No	No	No
Yang et al 2012 [186]	Miscellaneous	A	Medical	k-means	Not specified	No	No	No
Xin et al 2020 [56]	Miscellaneous	A	Non-medical	CNN	32 patients, 643 slices containing lesions	No	No	No
Wang et al 2018 [59]	Miscellaneous	PP	Non-medical	BoVW (K-CP with multilinear OMP, K-nearest neighbor)	Leave-on-out cross-validation is used for testing	No	No	No
Taghavi et al 2021 [9]	Miscellaneous	A	Medical	Random forest	21 patients	No	Yes	No
Ponnoprat et al 2020 [71]	Miscellaneous	A	Non-medical	U-Net for segmentation + CRF for post-processing + SVM for classification (w GHI kernel)	17 patients, 2042 images	No	No	No
Maaref et al 2020 [161]	Miscellaneous	A	Medical	2D CNN (Inception-Net, modified)	CLASSIFICATION: 20 patients for validation, 41 for testing; PREDICTION: 12 patients for validation, 24 for testing	No	No	No
Wang et al 2017 [187]	Miscellaneous	A	Non-medical	BoVW (sparse codebook-based feature representation)	(leave-one-out cross validation)	No	No	No
Li et al 2020 [188]	Miscellaneous	A	Medical	ResNet	69 patients, 3 images per patient (fivefold cross-validation)	No	No	No
Lee et al 2020 [8]	Miscellaneous	A	Non-medical	CNN + RFC and CNN + LRC	606 patients	No	No	No
Sun et al 2020 [189]	Miscellaneous	PP	Non-medical	SVM	34 labeled CT	No	No	0
Thuring et al 2020 [164]	Miscellaneous	A	Medical	Random Forrest & CNN	70 patients	No	Yes	Yes
Wang et al 2020 [166]	Miscellaneous	A	Non-medical	CNN (residual CNN)	70slices (17 patients)	No	No	0
Xu et al 2020 [190]	Miscellaneous	PP	Non-medical	CNN (Deep neural network)	20 from 3dIRCADb	Yes	No reference	Yes
Yang et al 2021 [191]	Miscellaneous	A	Non-medical	CNN (v-net)	8 CT	No	No	Yes
Yoshinobu et al 2020 [192]	Miscellaneous	PP	Non-medical	CNN (Deep CNN)	32 cases	No	No	0
Zhang et al 2020 [124]	Miscellaneous	A	Medical	CNN (DenseNet)	From multicenter data from 3 hospitals	Yes	Yes	Yes
Gu et al 2020 [193]	Miscellaneous	PP	Non-medical	CNN + ResNet	1 patient	No	No	No
Kobe et al 2021 [194]	Miscellaneous	A	Medical	ANN	21 metastases/lesions	No	No reference	No
Li et al 2022 [195]	Miscellaneous	A	Medical	CNN (DenseNet)	244 patients	No	Yes	Yes

We encountered studies with 19 different aims. To make comparison and discussion more feasible, we divided these studies into five groups according to study aim: (1) liver segmentation; (2) lesion segmentation; (3) lesion detection; (4) classification of liver or liver lesions; (5) miscellaneous/other. Aims are illustrated in electronic supplementary material. There is some overlap in the groups due to several studies having multiple aims. Detailed characteristics of included studies are given in supplementary tables.

Liver segmentation

The aim of liver segmentation was the primary or secondary study aim in eighty-four of the included studies. Of those, fifty-one are journal articles [20, 24, 29‐35, 38‐41, 43‐47, 49, 55‐58, 62, 63, 65, 68, 70‐79, 81, 84‐87, 89, 91, 93‐95, 97, 98, 196, 197], and 33 are proceeding papers [19, 21‐23, 25, 26, 36, 37, 42, 48, 51, 53, 54, 59‐61, 64, 66, 67, 69, 80, 82, 83, 88, 90, 92, 96, 99, 100, 103, 198]. The liver segmentation was done from the CT as a whole liver, not the clinical segmentation, e.g., Couinaud segments of the liver. Overall, this group of studies has contributed considerably with technically sound methods and experimented with various subdomains of ML, especially DL.

The quality of many recent studies has improved using external validation method to provide better generalizability. Though comparing directly with human experts is preferred, only eleven studies were found to do so.

The study group gives insinuation of obtaining labeled medical data which is challenging, as two-thirds of studies used datasets open for public use for training or testing their ML model. The dataset from LiTS 2017, which was the most frequently used, included 131 patients in their test set [199].

The attempt of transparency in reporting models’ performance was seen in many studies, though out of eighty-seven studies, only 11 reported their results with confidence interval or standard error; thus, further analyses of the result were not feasible in the group.

DICE score was used in most studies in this group to describe the model’s ability to predict which pixel contains the liver. The highest DICE reported was a score of 0.9851 [41], and the lowest score was 0.75 [94]. Other measures to describe the model’s performance were scattered, including AUC-ROC and accuracy (Table 2). Dong et al also reported a DICE of 0.92 and an accuracy of 0.9722 from their study, and the AUC of 0.96. References of studies in the group are in Table 3.

Table 2

Definition of performance and outcome measures

Segmentation	refers to a pixel-wise classification of images throughout this review. This is the standard meaning of “segmentation” of images in data science and engineering. It is not to be confused with anatomical segmentation like the Coineaud segmentation of liver lobes, commonly used for clinical segmentation of the liver according to the portal blood supply (19)
DICE	describes the percentage of overlap between the predicted and the observed/”correct” labeled area in an image (often labeled by a human radiologist), where 1.0/100% means a perfect overlap between predicted and correct segmentation
Accuracy	related to image segmentation in engineering is a measure describing how many pixels are correctly classified—1.0/100% being perfect. However, accuracy can be misleading in cases where a class is in very few pixels; for instance, a small tumor could be only in 2% of the image—and a model predicting that there are 0% tumors would still have an accuracy of 98%. Therefore, if only accuracy is reported for performance, a measure of class balance might be relevant to the readers' understanding
Precision and Recall	Precision is the number of relevant observations by a model divided by the total number of observations made by the model. For instance, if a model marks 100 pixels as tumor tissue and 40 are tumor tissue, the precision is 40%/0.4. Precision is the same as positive predictive value (PPV). Recall is the number of relevant observations divided by the total number of actual cases, e.g., if an image contains 100 pixels with actual tumor tissue, and the model observes 80 of them, the model has a recall of 80%/0.8. In binary classification cases, recall is the same as sensitivity, hit rate, and true positive rate
Volume Overlap Error (VOE)	gives a measure of the difference between actual area and predicted area. It functions as a combined score of both false positives and negatives \({\varvec{V}}{\varvec{O}}{\varvec{E}}\left({{\varvec{U}}}_{1},{{\varvec{U}}}_{2}\right)=100\times \boldsymbol{ }(1-\boldsymbol{ }\frac{{{\varvec{U}}}_{1}\cap \boldsymbol{ }{{\varvec{U}}}_{2}}{{{\varvec{U}}}_{1}\cup \boldsymbol{ }{{\varvec{U}}}_{2}})\) where U₁ and U₂ are true and predicted values, respectively. Optimal scores are as low as possible, 0 being the perfect score (20)
IoU / Jaccard Index	The intersection over union (IoU), is a measure that quantifies the percentage of overlap between prediction and observed/true output, much like the DICE coefficient. IoU measures the overlapping pixels between true and predicted segmentation and divides it by the total number of pixels either of them has marked as a pixel of interest. A perfect score would be 100%/1.0. This measure is also referred to as the Jaccard Index
Ground truth	refers to the label for anatomical structures in CT images given by a clinician or radiologist. What kind of expert and level of experience is often specified in each specific study
CNN	refers to Convolutional Neural Network – a deep learning model based on vector calculations used in image recognition and processing pixel data

Table 3

References of studies in each category according to characteristics

Characteristics of studies	Liver segmentation	Lesion segmentation	Lesion detection	Classification of liver or lesions	Miscellaneous
Journal article	51 studies [20, 24, 29‐35, 38‐41, 43‐47, 49, 55‐58, 62, 63, 65, 68, 70‐79, 81, 84‐87, 89, 91, 93‐95, 97, 98, 196, 197]	36 studies [24, 29, 31, 32, 38, 46, 47, 55, 56, 62, 72, 78, 84, 91, 93, 94, 97, 98, 111, 115, 117, 118, 120, 122, 124, 125, 130, 133‐135, 137, 138, 140, 199]	5 studies [101, 106, 111, 115, 202]	34 studies [56, 71, 72, 74, 78, 141‐146, 148‐152, 154, 156‐161, 164‐172, 202, 203]	29 studies [6, 8, 9, 33, 50, 52, 56, 71, 161, 164, 173‐179, 181‐184, 186‐188, 191, 194, 195, 205, 206]
Proceeding papers	33 studies [19, 21‐23, 25, 26, 36, 37, 42, 48, 51, 53, 54, 59‐61, 64, 66, 67, 69, 80, 82, 83, 88, 90, 92, 96, 99, 100, 103, 198]	24 studies [22, 37, 42, 64, 65, 68, 82, 88, 92, 96, 99, 103, 108, 121, 124, 126‐129, 131, 132, 136, 139, 200]	15 studies [23, 26, 27, 87, 102, 104, 105, 107‐110, 112‐114, 119]	13 studies [27, 64, 65, 68, 75, 82, 119, 147, 153, 155, 162, 163, 204]	8 studies [27, 180, 185, 189, 190, 192, 193, 207]
ML to human expert	10 studies [20, 23, 24, 32, 33, 35, 58, 76, 81, 98]	6 studies [24, 32, 76, 98, 134, 140]	2 studies [23, 106]	5 studies [143, 149, 152, 164, 165]	9 studies [9, 33, 164, 174, 176, 177, 183, 195, 206]
Using public datasets	57 studies [19‐21, 23‐26, 28‐31, 34‐37, 39, 41, 43, 46, 49, 51, 54, 55, 57, 59, 60, 62, 66, 69, 70, 74, 76, 77, 79‐81, 83‐86, 88, 89, 91‐96, 98‐100, 103, 196, 198]	38 studies [24, 29, 31, 37, 38, 46, 47, 55, 62, 76, 79, 84, 88, 89, 91‐94, 96‐99, 118, 122, 124‐129, 132, 136‐139, 200]	8 studies [23, 26, 101, 104, 106, 108, 112, 119]	12 studies [74, 119, 141‐143, 145, 147‐149, 155, 156, 164]	10 studies [20, 50, 52, 173, 174, 183, 190, 191, 195, 206]
Reporting of standard error	11 studies [20, 21, 27, 31, 34, 39, 49, 88, 90, 196, 197]	7 studies [29, 31, 47, 88, 115, 120, 122]	2 studies [27, 115]	3 studies [27, 143, 165]	8 studies [33, 50, 173, 178, 179, 181, 190, 205]
Reporting of DICE score	55 studies [20‐22, 24‐26, 29‐34, 37, 38, 42, 46‐48, 51, 54‐57, 59‐62, 65, 66, 69‐74, 77, 79, 85, 88‐92, 94‐96, 98‐100, 103, 196, 198]	42 studies [22, 24, 29, 31, 32, 37, 38, 42, 46, 47, 55, 56, 62, 72, 76, 79, 84, 88, 89, 91, 92, 94, 96‐99, 103, 115, 117, 120, 122, 124, 125, 127, 129, 131, 137, 139, 200]	4 studies [27, 106, 115, 202]	10 studies [56, 65, 72, 74, 144, 145, 155, 165‐167]	12 studies [52, 56, 71, 179‐183, 185, 189‐191]
Reporting of accuracy	13 studies [19, 35, 36, 38, 53, 63‐65, 74, 76, 82, 93, 94]	8 studies [42, 64, 65, 94, 118, 130, 138]	4 studies [23, 102, 108, 119]	31 studies [27, 64, 65, 71, 74, 75, 82, 119, 142‐145, 151, 153‐165, 168, 202‐204]	19 studies [6, 27, 33, 50, 52, 71, 161, 164, 174, 175, 178, 182, 184, 187‐191, 193]
Reporting of AUC	3 studies [23, 38, 94]	4 studies [38, 94, 111, 115]	3 studies [23, 111, 115]	16 studies [56, 74, 75, 143, 152, 154‐156, 158, 161, 164, 167, 168, 170‐172]	12 studies [8, 9, 33, 161, 173, 176‐178, 188, 194, 195, 205]
Reporting of precision	8 studies [23, 26, 34, 63, 74, 87, 95, 98]	5 studies [32, 56, 98, 111, 128]	9 studies [23, 87, 105, 108, 111, 113, 114, 202]	13 studies [56, 65, 74, 119, 143, 150, 153‐156, 160, 165, 170]	5 studies [6, 173, 186, 192, 207]
Reporting of VOE	17 studies [21, 30‐32, 35, 36, 39, 46‐48, 89, 91, 97, 100, 103, 196, 197]	24 studies [31, 32, 46, 47, 55, 62, 89, 91, 97, 103, 111, 120, 122, 123, 125‐127, 129, 132, 136, 137, 201]	Not available	Not available	1 study [27]
External validation	32 studies [20, 21, 24‐27, 30‐33, 37, 39‐45, 47, 48, 62, 76, 81, 84, 86, 88, 89, 91, 92, 94, 98, 103]	26 studies [24, 31, 32, 37, 42, 47, 62, 76, 84, 88, 89, 91, 92, 94, 98, 103, 115, 118, 120, 122‐125, 132, 136, 140]	4 studies [26, 27, 111, 115]	8 studies [27, 146, 151, 152, 156, 158, 165, 172]	7 studies [27, 33, 52, 174, 183, 190, 206]

Lesion segmentation

This group of studies performed segmentation of liver lesions from CT images with ML. The model’s goal was the highest possible truthfulness of segmented lesions compared to ground truth. Sixty studies had lesion segmentation as a primary or secondary study aim. Thirty-six are journal articles [24, 29, 31, 32, 38, 46, 47, 55, 56, 62, 72, 78, 84, 91, 93, 94, 97, 98, 102, 111, 115, 117, 118, 122, 124, 125, 130, 133‐135, 137, 138, 140, 201], and twenty-four [22, 37, 42, 64, 65, 68, 82, 88, 92, 96, 99, 103, 108, 121, 124, 126‐129, 131, 132, 136, 139, 200] are proceedings papers.

Several models have shown remarkable segmenting ability for predicted lesions larger than 2 cm in diameter, while almost every model is still struggling to segment lesion size less than 1 cm in diameter. However, this is comparable with clinicians in the clinical setting. Another limitation for the model to predict the lesion was quality of CT images. Several more recent studies used voxel-wise (3D pixels) classification. This could use more available information and give output in 3D to improve performance.

Validation of the model with external validation and ML to humans is improving for this group, and twenty-six studies have used external validation. Only six studies have compared their model with human experts.

More than half of the studies have reported performance in a DICE score in this group. The score range was seen skewed in different studies with the range of 0.44–0.96; a selection of lesion size played a key role here for higher performance or higher DICE score. Another informative measure called Volume Overlap Error (VOE) gives the difference between predicted and ground truth in an area. Thus, 0 is the optimal score. Twenty-two studies reported VOE, with a 0.01–0.46 mm range. Other measures were dispersed in different studies, including accuracy, AUC, precision, or PPV. Few studies have reported their performance with confidence intervals or standard errors—references of studies in the group in Table 3.

Lesion detection

Twenty studies had lesion detection as a primary or secondary study aim. This involves simply detecting if lesions are present in a CT image. Fifteen of them are proceedings papers [23, 26, 27, 87, 102, 104, 105, 107‐110, 112‐114, 119], and five are journal articles [101, 106, 111, 115, 202].

Several newer studies have detected lesions before segmentation of the lesions or diagnosis of the lesions with ML from CT liver images but have not reported performance of the lesion detection task of the model; thus, this group is smaller.

External validation was reported only in four studies. Most studies acquired their training data from local hospitals, and only eight studies have used data sets open for public use. DL was the choice of a subdomain of ML for this group.

Reporting of performance was seen as transparent and detailed in newer studies in all groups. In this group, performance was primarily reported in accuracy and precision, but five studies reported only false positives and true positive rate [26, 87, 101, 104, 115]. Two studies presented its result with a confidence interval or standard error. It is worth mentioning that the study reporting the best precision only performed internal validation on the relatively small, public dataset 3D-IRCADb—references of studies in the group in Table 3.

Classification of liver or lesions

Included studies in this group classifying the type and severity of lesions or tumors, grading hepatocellular carcinoma (HCC), and differentiating between HCC, hemangioma, and metastases. Most studies differed only between two categories, such as classifying tumors as either benign or malign. Forty-seven studies had the classification of liver or liver lesions as a study aim. Thirty-four of them journal articles [56, 71, 72, 74, 78, 141‐146, 148‐152, 154, 156‐161, 164‐172, 202, 203], and thirteen are proceedings papers [27, 64, 65, 68, 75, 82, 119, 147, 153, 155, 162, 163, 204]. For classification of liver or liver lesions, traditional machine learning, e.g., support vector machines and random forest models, and deep learning models were commonly used.

Nine studies compared their model performance directly to one or more clinicians in a competition-based comparison. Only 12 studies have used datasets open for public validation, and even fewer are needed for training purposes.

Accuracy was a method of choice to present the performance in this group; thirty-one studies reported accuracy, with a range of 0.76–0.99. Sixteen studies reported AUC, with a range of 0.68–0.97. Precision was reported in fourteen studies. The precision range was 0.82–1.00. Note that both Sreeja et al and Romero et al reported a perfect precision of 1.0, which Sreeja et al commented was possible due to the small size of their data set [153, 155]. Only three studies presented their result with a confidence interval—references of studies in the group are in Table 3.

Other/miscellaneous

The last and most diverse category we found eligible to compare was miscellaneous, including 29 journal article [6, 8, 9, 33, 50, 52, 56, 71, 161, 164, 173‐179, 181‐184, 186‐188, 191, 194, 195, 205, 206] and 8 proceeding paper [27, 180, 185, 189, 190, 192, 193, 207] total thirty-seven studies. The aims of the studies are clinical-oriented.

Seven studies have performed liver fibrosis staging [33, 173‐178] according to “Metavir” or “Fibrosis-4” classification [208, 209]. Four compared algorithms performance with human expert while two studies performed external validation. Only two studies used public dataset for liver segmenting purpose; however, private datasets were used for fibrosis staging training and validation purpose in all the seven studies. ML method like SVM, k-nearest neighbor were used traditionally but in the recent studies, CNN-based systems using different classifier to extract the feature from the liver image are gaining more attention. Jung et al used liver and spleen volumetric indices and perform the pathologic liver fibrosis staging with CNN [177]. Comparison of ML algorithm to 3 radiologists’ assessment of liver fibrosis staging was performed with more accurate result in ML group [33].

Six studies segmented blood vessels in the liver from CT images, including portal and liver veins [52, 179, 183‐185, 191]. Twelve studies reported a DICE score with a range of 0.68–0.98. The four studies reported accuracy with a range of 0.91–0.98, with a mean of 0.96 and a median of 0.97. Five studies stated that they externally validated their model.

Five retrieved focal liver lesion images as a study aim [50, 186, 187, 192, 206]. These studies showed how models could improve clinical workflow by retrieving similar cases in medical records, including earlier expert opinions.

Two studies, published as journal articles, predicted liver metastases within colorectal cancer patients [8, 9]. They reported AUC equal to 0.86 ± 0.01(12) and 0.747 ± 0.036.

One study focused on the segmentation of bile ducts and stones in the intrahepatic bile duct—hepatolith and reported DICE of 0.90 and 0.71 for bile duct and hepatolith segmentation, respectively [6].

Three study focused on response evaluation after chemotherapy or radio-embolization of malignant liver lesions using texture analysis [161, 181, 182]. They compared texture analysis predictions with survival and serologic response and reported an accuracy of 0.97, sensitivity of 0.93, and specificity of 1.0. This was after training on sixty-two patients and testing using cross-validation.

Two recent studies have predicted liver reserve function using Child–Pugh classification [164, 189] and Thuring et al have compared the results from their ML model with results from clinicians. Prediction of Child–Pugh accuracy was 53%, classification of Child–Pugh A vs B: accuracy was 78%, sensitivity 81%, specificity 70%, and AUC 0.80. Wang et al had preoperatively predicted early recurrence in HCC. One study has predicted overall survival of patients with unresectable HCC treated by transarterial chemoembolization [176]. This study also presented fusion of clinical data with ML model. References of studies in the group in Table 3.

Discussion

We found that ML is applied to liver CT imaging for various clinical oriented aims and covering a broad spectrum of applications.

At least one-third of studies were documented to perform very accurately on reliable, but small data. Unfortunately, reporting of performance was seldom appropriate due to lack of details. To our knowledge, there exists no standardized form of presenting results for machine learning models applied to medical imaging.

Several studies reported models that were close to clinical application. However, clinical validation with thorough documentation of both model and data (training and validation) to assess quality and generalizability were lacking. Evaluation of the model by only analysis of a result parameters would be imperfect [210].

Almost all studies that performed segmentation of liver structures from the CT images of the abdomen used deep learning models, mainly the subtype CNN. Open-access datasets and competitions like LiTS 2017 contribute substantially to the development of ML applied to liver imaging, as more than 280 studies report their model performance in a standardized format, and the competition is still ongoing with cumulative comparison. U-Net a sub domain of CNN is used by many participants and have shown promising result. The distribution of sources of dataset used by studies included in this review is illustrated in Fig. 2. The use of complex models and targeting for complex aims like automatic liver fibrosis staging, treatment response evaluation, prediction of occurrence of liver metastases, and liver blood vessels segmentation for traditional anatomical landmarks, e.g., Coineaud classification, are getting more common and may herald a maturing process in the field.

ML systems showed promising results on retrospective data for several tasks related to CT imaging, as some segmentation studies reported models with more than 98% ability to predict which pixels or voxels contained liver in abdominal CT scans. Further, several studies reported 95% performance compared to ground truth for liver or liver lesions classification. In recent years, identified studies have used ML for prediction of occurrence or treatment effect of liver metastases, liver vessel segmentation, and evaluation of treatment effect on liver malignancy. These showed results around 70–80% of ground truth.

Other applications such as classification of liver fibrosis stage and prediction of benign or malign lesions showed promising results and potential for the high impact of ML in future routine clinical practice.

Reporting of model performance should give in the state-of-the-art visualization methods, e.g., confusion matrix. In the studies like segmentation task, measuring parameter like mean surface distance with standard error should be reported to get overall transparency of the model performance [116]. Sixty-two studies identified in this review have such breach in reporting of model performance. This makes it difficult to get a good overall understanding of the field, especially for clinicians. We encourage the readers to assess such results with caution.

Further, reporting of standard error and confidence intervals was often lacking. We recommend that it should be reported by default. This problem was also seen in other applications of ML to medical images, and we concur with the need for reporting standards for medical application as stated by Aggarwal et al [10].

There are potentially many applications of ML in liver CT imaging have been identified thorough this review, especially in the miscellaneous group aims are clinically derived, while segmenting of liver and its lesions could implement as diagnostic and treatment planning tool. Studies in classification group could serve diagnosis of different lesions, e.g., different types of malign and benign tumors, or severity of the liver cirrhosis. Despite the promising performance reported in many studies, clinical applications of ML in liver CT imaging have to pass through the corridor of clinical validation and clinical trials [210].

The main issues identified in the literature were limited access to high-quality data and lack of clinical validation. External validation is becoming more popular among developers, illustrated in Fig. 3, but it is insufficient to qualify for medical application. There is an urgent need for a shift in focus towards clinical validation in this field. Scholars should perform feasibility studies in clinical routine, and design and carry out prospective studies to validate the performance of ML tools in realistic clinical settings. Developers should seek to collaborate with clinicians in this process. Strength and weakness of the study as well future perspective is given in the supplementary material.

Conclusion

We found reports of many ML applications to liver CT images in the literature, including automatic liver and lesion segmentation, lesion detection, liver or lesion classification, liver vessel segmentation including bile ducts, fibrosis staging, metastasis prediction, and evaluation of chemotherapy as treatment of hepatocellular carcinoma and retrieval of relevant liver lesions from other similar cases. Several were documented to perform very accurately on reliable but small data. Deep learning models and classification models of ML were commonly used. However, presenting the result of studies is not standardized in the literature. Some studies were close to reporting sufficient details on clinical relevance, data characteristics and quality, algorithm characteristics and bias, and performance measures on external data to be considered ready for clinical use. Further prospective, clinical studies are recommended, and the need for a more interactive technological and medical research is evident to achieve a secure clinical use of ML methodology in this field.

Acknowledgements

Infrastructure support for this research was provided by the University Hospital of North Norway and The Arctic University of Norway (UiT).

Guidance and support while writing this manuscript from Professor Arthur Revhaug MD PhD at the Arctic University of Norway (UiT). Arthur.revhaug@uit.no .

Declarations

Guarantor

The scientific guarantor of this publication is Keyur Radiya.

Conflict of interest

The authors of this manuscript declare no relationships with any companies, whose products or services may be related to the subject matter of the article.

Statistics and biometry

No complex statistical methods were necessary for this paper.

Ethical approval

Institutional Review Board approval was not required because systematic review article and not an experiment.

Methodology

• Systematic review

Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Publisher's note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Unsere Produktempfehlungen

e.Med Interdisziplinär

Kombi-Abonnement

Für Ihren Erfolg in Klinik und Praxis - Die beste Hilfe in Ihrem Arbeitsalltag

Mit e.Med Interdisziplinär erhalten Sie Zugang zu allen CME-Fortbildungen und Fachzeitschriften auf SpringerMedizin.de.

Jetzt testen ¹

e.Med Radiologie

Kombi-Abonnement

Mit e.Med Radiologie erhalten Sie Zugang zu CME-Fortbildungen des Fachgebietes Radiologie, den Premium-Inhalten der radiologischen Fachzeitschriften, inklusive einer gedruckten Radiologie-Zeitschrift Ihrer Wahl.

Jetzt testen ²

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (PDF 176 KB)

Baştanlar Y, Özuysal M (2014) Introduction to machine learning. In: Yousef M, Allmer J (eds) miRNomics: MicroRNA Biology and Computational Analysis. Humana Press, Totowa, NJ, pp 105–128

Muehlematter UJ, Daniore P, Vokinger KN (2021) Approval of artificial intelligence and machine learning-based medical devices in the USA and Europe (2015–20): a comparative analysis. Lancet Digit Health 3:e195–e203PubMed

FDA (2021) Artificial Intelligence and Machine Learning (AI/ML) Software as a Medical Device Action Plan 2021. FDA. Available via https://www.fda.gov/medical-devices/software-medical-device-samd/artificial-intelligence-and-machine-learning-software-medical-device. Accessed 24.01.2023

Rubin GD (2014) Computed tomography: revolutionizing the practice of medicine for 40 years. Radiology 273:S45–S74PubMed

Cardobi N, Dal Palu A, Pedrini F et al (2021) An overview of artificial intelligence applications in liver and pancreatic imaging. Cancers 13:11

Fu X, Cai N, Huang K et al (2019) M-Net: a novel U-Net with multi-stream feature fusion and multi-scale dilated convolutions for bile ducts and hepatolith segmentation. IEEE Access 7:148645–148657

Decharatanachart P, Chaiteerakij R, Tiyarattanachai T, Treeprasertsuk S (2021) Application of artificial intelligence in chronic liver diseases: a systematic review and meta-analysis. BMC Gastroenterol 21:10PubMedPubMedCentral

Lee S, Choe EK, Kim SY, Kim HS, Park KJ, Kim D (2020) Liver imaging features by convolutional neural network to predict the metachronous liver metastasis in stage I-III colorectal cancer patients based on preoperative abdominal CT scan. BMC Bioinf 21:382

Taghavi M, Trebeschi S, Simões R et al (2021) Machine learning-based analysis of CT radiomics model for prediction of colorectal metachronous liver metastases. Abdom Radiol (NY) 46:249–256PubMed

10.

Aggarwal R, Sounderajah V, Martin G et al (2021) Diagnostic accuracy of deep learning in medical imaging: a systematic review and meta-analysis. NPJ Digital Med 4:65

11.

Zhou LQ, Wang JY, Yu SY et al (2019) Artificial intelligence in medical imaging of the liver. World J Gastroenterol 25:672–682PubMedPubMedCentral

12.

Park HJ, Park B, Lee SS (2020) Radiomics and deep learning: hepatic applications. Korean J Radiol 21:387–401PubMedPubMedCentral

13.

Azer SA (2019) Deep learning with convolutional neural networks for identification of liver masses and hepatocellular carcinoma: a systematic review. World J Gastrointest Oncol 11:1218–1230PubMedPubMedCentral

14.

Schiavo JH (2019) PROSPERO: an international register of systematic review protocols. Med Ref Serv Q 38:171–180PubMed

15.

McInnes MDF, Moher D, Thombs BD et al (2018) Preferred Reporting Items for a Systematic Review and Meta-analysis of diagnostic test accuracy studies: the PRISMA-DTA statement. JAMA 319:388–396PubMed

16.

Cumpston M, Li T, Page MJ et al (2019) Updated guidance for trusted systematic reviews: a new edition of the Cochrane Handbook for Systematic Reviews of Interventions. Cochrane Database Syst Rev 10:Ed000142PubMed

17.

de Hond AAH, Leeuwenberg AM, Hooft L et al (2022) Guidelines and quality criteria for artificial intelligence-based prediction models in healthcare: a scoping review. NPJ Digit Med 5:2PubMedPubMedCentral

18.

Haddaway NR, Page MJ, Pritchard CC, McGuinness LA (2022) PRISMA2020: an R package and Shiny app for producing PRISMA 2020-compliant flow diagrams, with interactivity for optimised digital transparency and Open Synthesis. Campbell Syst Rev 18:e1230PubMedPubMedCentral

19.

Mubashir A, Yuan D, Syed Furqan Q, Jian Y (2019) Convolutional-neural-network-based feature extraction for liver segmentation from CT imagesProcSPIE, pp 1117934

20.

Ahn Y, Yoon JS, Lee SS et al (2020) Deep learning algorithm for automated segmentation and volume measurement of the liver and spleen using portal venous phase computed tomography images. Korean J Radiol 21:987–997PubMedPubMedCentral

21.

Bhavya A, Aditya B, Karthik K (2018) Automatic and fast CT liver segmentation using sparse ensemble with machine learned contextsProcSPIE, pp 105740L

22.

Albishri AA, Shah SJH, Lee Y (2019) CU-Net: cascaded U-Net model for automated liver and lesion segmentation and summarization. In: IEEE International Conference on Bioinformatics and Biomedicine (BIBM), San Diego, pp 1416–1423

23.

Ali L, Khelil K, Wajid SK et al (2017) Machine learning based computer-aided diagnosis of liver tumour. In: IEEE 16^th International Conference on Cognitive Informatics and Cognitive Computing (ICCI*CC), Oxford, pp 139–114

24.

Alirr OI (2020) Deep learning and level set approach for liver and tumor segmentation from CT scans. J Appl Clin Med Phys 21:200–209PubMedPubMedCentral

25.

Astono I, Welsh JS, Chalup S (2018) Adjacent network for semantic segmentation of liver CT scans. In: 18th IEEE International Conference on Bioinformatics and Bioengineering, Taichung, pp 35–40

26.

Ben-Cohen A, Diamant I, Klang E, Amitai M, Greenspan H (2016) Fully convolutional network for liver segmentation and lesions detection. In: 2nd International Workshop on Deep Learning in Medical Image Analysis (DLMIA) / 1st International Workshop on Large-Scale Annotation of Biomedical Data and Expert Label Synthesis (LABELS). Springer International Publishing Ag, Athens, pp 77–85

27.

Bevilacqua V, Brunetti A, Trotta GF et al (2017) A novel approach for hepatocellular carcinoma detection and classification based on triphasic CT Protocol2017 IEEE Congress on Evolutionary Computation (CEC), pp 1856–1863

28.

Bhole C, Morsillo N, Pal C (2011) 3D segmentation in CT imagery with conditional random fields and histograms of oriented gradients. In: Suzuki K, Wang F, Shen D, Yan P (eds) Machine Learning in Medical Imaging. Springer, Berlin Heidelberg, Berlin, Heidelberg, pp 326–334

29.

Budak U, Guo Y, Tanyildizi E, Sengur A (2020) Cascaded deep convolutional encoder-decoder neural networks for efficient liver tumor segmentation. Med Hypotheses 134:8

30.

Cai J (2019) Segmentation and diagnosis of liver carcinoma based on adaptive scale-kernel fuzzy clustering model for CT images. J Med Syst 43:322PubMed

31.

Chen Y, Wang K, Liao X et al (2019) Channel-Unet: a spatial channel-wise convolutional neural network for liver and tumors segmentation. Front Gen 10

32.

Chlebus G, Schenk A, Moltz JH, van Ginneken B, Hahn HK, Meine H (2018) Automatic liver tumor segmentation in CT with fully convolutional neural networks and object-based postprocessing. Sci Rep 8:15497PubMedPubMedCentral

33.

Choi KJ, Jang JK, Lee SS et al (2018) Development and validation of a deep learning system for staging liver fibrosis by using contrast agent-enhanced CT images in the liver. Radiol 289:688–697

34.

Chung M, Lee J, Lee M, Lee J, Shin Y-G (2020) Deeply self-supervised contour embedded neural network applied to liver segmentation. Comput Methods Programs Biomed 192:105447PubMed

35.

Danciu M, Gordan M, Florea C, Orghidan R, Sorantin E, Vlaicu A (2013) A hybrid 3D learning-and-interaction-based segmentation approach applied on CT liver volumes. Radioeng 22:100–113

36.

Danciu M, Gordan M, Florea C, Vlaicu A (2012) 3D DCT supervised segmentation applied on liver volumes 2012. 35th International Conference on Telecommunications and Signal Processing (TSP), pp 779–783

37.

Delmoral JC, Costa DC, Borges D, Tavares JMRS (2019) Segmentation of pathological liver tissue with dilated fully convolutional networks: a preliminary study2019 IEEE 6th Portuguese Meeting on Bioengineering (ENBENG), pp 1–4

38.

Dong X, Zhou Y, Wang L, Peng J, Lou Y, Fan Y (2020) Liver cancer detection using hybridized fully convolutional neural network based on deep learning framework. IEEE Access 8:129889–129898

39.

Dou Q, Yu LQ, Chen H et al (2017) 3D deeply supervised network for automated segmentation of volumetric medical images. Med Image Anal 41:40–54

40.

Guo X, Schwartz LH, Zhao B (2019) Automatic liver segmentation by integrating fully convolutional networks into active contour models. Med Phys 46:4455–4469PubMed

41.

He B, Huang C, Sharp G et al (2016) Fast automatic 3D liver segmentation based on a three-level AdaBoost-guided active shape model. Med Phys 43:2421PubMed

42.

Heker M, Ben-Cohen A, Greenspan H (2019) Hierarchical fine-tuning for joint liver lesion segmentation and lesion classification in CT2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp 895–898

43.

Hu P, Wu F, Peng J, Liang P, Kong D (2016) Automatic 3D liver segmentation based on deep learning and globally optimized surface evolution. Phys Med Biol 61:8676–8698PubMed

44.

Huang W, Tan ZM, Lin Z et al (2012) A semi-automatic approach to the segmentation of liver parenchyma from 3D CT images with extreme learning machine. In: 34th Annual International Conference of the IEEE Engineering-in-Medicine-and-Biology-Society (EMBS). IEEE, San Diego, pp 3752–3755

45.

Ji H, He J, Yang X, Deklerck R, Cornelis J (2013) ACM-based automatic liver segmentation from 3-D CT images by combining multiple atlases and improved mean-shift techniques. IEEE J Biomed Health Inform 17:690–698PubMed

46.

Jiang H, Li S, Li S (2018) Registration-based organ positioning and joint segmentation method for liver and tumor segmentation. Biomed Res Int 2018:8536854PubMedPubMedCentral

47.

Jiang H, Shi T, Bai Z, Huang L (2019) AHCNet: an application of attention mechanism and hybrid connection for liver tumor segmentation in CT volumes. IEEE Access 7:24898–24909

48.

Jin X, Ye H, Li L, Xia Q (2017) Image segmentation of liver CT based on fully convolutional network2017 10th International Symposium on Computational Intelligence and Design (ISCID), pp 210–213

49.

Kavur AE, Gezer NS, Barış M et al (2020) Comparison of semi-automatic and deep learning-based automatic methods for liver segmentation in living liver transplant donors. Diagn Interv Radiol 26:11–21PubMed

50.

Kumar A, Dyer S, Kim J et al (2016) Adapting content-based image retrieval techniques for the semantic annotation of medical images. Comput Med Imaging Graph 49:37–45PubMed

51.

Zheng H, Lin L, Hu H et al (2019) Semi-supervised segmentation of liver using adversarial learning with deep atlas prior. In: Shen D, Liu T, Peters TM et al (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019. Springer International Publishing, Cham, pp 148–156

52.

Zhang R, Zhou Z, Wu W, Lin CC, Tsui PH, Wu S (2018) An improved fuzzy connectedness method for automatic three-dimensional liver vessel segmentation in CT images. J Healthc Eng 2018:2376317PubMedPubMedCentral

53.

Zhang L, Xu L (2018) An automatic liver segmentation algorithm for CT images U-net with separated paths of feature extraction2018 IEEE 3rd International Conference on Image, Vision and Computing (ICIVC), pp 294–298

54.

Xu W, Liu H, Wang X, Qian Y (2019) Liver segmentation in CT based on ResUNet with 3D probabilistic and geometric post process2019 IEEE 4th International Conference on Signal and Image Processing (ICSIP), pp 685–689

55.

Xi XF, Wang L, Sheng VS, Cui Z, Fu B, Hu F (2020) Cascade U-ResNets for simultaneous liver and lesion segmentation. IEEE Access 8:68944–68952

56.

Xin S, Shi H, Jide A, Zhu M, Ma C, Liao H (2020) Automatic lesion segmentation and classification of hepatic echinococcosis using a multiscale-feature convolutional neural network. Med Biol Eng Comput 58:659–668PubMed

57.

Xia K, Yin H, Qian P, Jiang Y, Wang S (2019) Liver semantic segmentation algorithm based on improved deep adversarial networks in combination of weighted loss function on abdominal CT images. IEEE Access 7:96349–96358

58.

Winkel DJ, Weikert TJ, Breit H-C et al (2020) Validation of a fully automated liver segmentation algorithm using multi-scale deep reinforcement learning and comparison versus manual segmentation. Eur J Radiol 126:108918PubMed

59.

Wang C, Song H, Chen L et al (2018) Automatic liver segmentation using multi-plane integrated fully convolutional neural networks2018 IEEE International Conference on Bioinformatics and Biomedicine (BIBM), pp 1–6

60.

Tian J, Liu L, Shi Z, Xu F (2019) Automatic Couinaud segmentation from CT volumes on liver using GLC-UNet. In: Suk H-I, Liu M, Yan P, Lian C (eds) Machine Learning in Medical Imaging. Springer International Publishing, Cham, pp 274–282

61.

Tang M, Valipour S, Zhang Z, Cobzas D, Jagersand M (2017) A deep level set method for image segmentation. In: Cardoso MJ, Arbel T, Carneiro G et al (eds) Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. Springer International Publishing, Cham, pp 126–134

62.

Seo H, Huang C, Bassenne M, Xiao R, Xing L (2020) Modified U-Net (mU-Net) with incorporation of object-dependent high level features for improved liver and liver-tumor segmentation in CT images. IEEE Trans Med Imaging 39:1316–1325PubMed

63.

Selvi E, Selver MA, Güzeliş C, Dicle O (2014) A higher-order neural network design for improving segmentation performance in medical image series. J Phys: Conf Ser 490:012079

64.

Selvathi D, Malini C, Shanmugavalli P (2013) Automatic segmentation and classification of liver tumor in CT images using adaptive hybrid technique and Contourlet based ELM classifier2013 International Conference on Recent Trends in Information Technology (ICRTIT), pp 250–256

65.

Sayed GI, Hassanien AE, Schaefer G (2016) An automated computer-aided diagnosis system for abdominal CT liver images 20th conference on medical image understanding and analysis (MIUA 2016), Loughborough Univ, Loughborough, England, pp 68–73

66.

Sakboonyara B, Taeprasartsit P (2019) U-Net and mean-shift histogram for efficient liver segmentation from CT images2019 11th International Conference on Knowledge and Smart Technology (KST), pp 51–56

67.

K S, H LU, H KIM, S K, M T (2018) ROI-based fully automated liver registration in multi-phase CT Images2018 18th International Conference on Control, Automation and Systems (ICCAS), pp 645–649

68.

Raj A, Jayasree M (2016) Automated liver tumor detection using Markov random field segmentation International conference on emerging trends in engineering, science and technology (ICETEST - 2015), Trichur, India, pp 1305–1310

69.

Rafiei S, Nasr-Esfahani E, Najarian K, Karimi N, Samavi S, Soroushmehr SMR (2018) Liver segmentation in CT images using three dimensional to two dimensional fully convolutional network2018 25th IEEE International Conference on Image Processing (ICIP), pp 2067–2071

70.

Qin W, Wu J, Han F et al (2018) Superpixel-based and boundary-sensitive convolutional neural network for automated liver segmentation. Phys Med Biol 63:095017PubMedPubMedCentral

71.

Ponnoprat D, Inkeaw P, Chaijaruwanich J et al (2020) Classification of hepatocellular carcinoma and intrahepatic cholangiocarcinoma based on multi-phase CT scans. Med Biol Eng Comput 58:2497–2515PubMed

72.

Ouhmich F, Agnus V, Noblet V, Heitz F, Pessaux P (2019) Liver tissue segmentation in multiphase CT scans using cascaded convolutional neural networks. Int J Comput Assist Radiol Surg 14:1275–1284PubMed

73.

Ng YS, Xi Y, Qian Y et al (2020) Use of spectral detector computed tomography to improve liver segmentation and volumetry. J Comput Assist Tomogr 44:197–203PubMed

74.

Nayak A, Baidya Kayal E, Arya M et al (2019) Computer-aided diagnosis of cirrhosis and hepatocellular carcinoma using multi-phase abdomen CT. Int J Comput Assist Radiol Surg 14:1341–1352PubMed

75.

Mukherjee DP, Higashiura K, Okada T et al (2013) Utilizing disease-specific organ shape components for disease discrimination: application to discrimination of chronic liver disease from CT data16th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) pp 235-242. Nagoya Univ, Nagoya, Japan

76.

Morshid A, Elsayes KM, Khalaf AM et al (2019) A machine learning model to predict hepatocellular carcinoma response to transcatheter arterial chemoembolization. Radiol Artif Intell 1

77.

Mohagheghi S, Foruzan AH (2020) Incorporating prior shape knowledge via data-driven loss model to improve 3D liver segmentation in deep CNNs. Int J Comput Assist Radiol Surg 15:249–257PubMed

78.

Mofrad FB, Zoroofi RA, Tehrani-Fard AA, Akhlaghpoor S, Sato Y (2014) Classification of normal and diseased liver shapes based on Spherical Harmonics coefficients. J Med Syst 38:20PubMed

79.

Meng L, Tian Y, Bu S (2020) Liver tumor segmentation based on 3D convolutional neural network with dual scale. J Appl Clin Med Phys 21:144–157PubMed

80.

Luo S, Li J (2014) Accurate object segmentation using novel active shape and appearance models based on support vector machine learning2014 International Conference on Audio, Language and Image Processing, pp 347–351

81.

Lu F, Wu F, Hu P, Peng Z, Kong D (2017) Automatic 3D liver location and segmentation via convolutional neural network and graph cut. Int J Comput Assist Radiol Surg 12:171–182PubMed

82.

Selvaraj G, Janakiraman S (2013) Improved feature selection based on particle swarm optimization for liver disease diagnosis. In: 4th International Conference on Swarm, Evolutionary, and Memetic Computing (SEMCCO). Springer-Verlag Berlin, SRM University, Chennai, pp 214–225

83.

Li XH, Huang C, Jia FC, Li ZM, Fang CH, Fan YF (2014) Automatic liver segmentation using statistical prior models and free-form deformation. In: International Workshop on Medical Computer Vision - Algorithms for Big Data (MICCAI-bigMCV), Cambridge, pp 181–188

84.

Li X, Chen H, Qi X, Dou Q, Fu CW, Heng PA (2018) H-DenseUNet: hybrid densely connected unet for liver and tumor segmentation from CT volumes. IEEE Trans Med Imaging 37:2663–2674PubMed

85.

Liu Z, Song YQ, Sheng VS et al (2019) Liver CT sequence segmentation based with improved U-Net and graph cut. Expert Systems with Applications 126:54–63

86.

Linguraru MG, Richbourg WJ, Liu J et al (2012) Tumor burden analysis on computed tomography by automated liver and tumor segmentation. IEEE Trans Med Imaging 31:1965–1976PubMedPubMedCentral

87.

Afifi A, Nakaguchi T (2015) Unsupervised detection of liver lesions in CT images. In: 37th Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society (EMBC). IEEE, Milan, pp 2411–2414

88.

Roth K, Hesser J, Konopczynski T (2020) Mask mining for improved liver lesion segmentation. In: IEEE 17th International Symposium on Biomedical Imaging (ISBI). IEEE, Iowa, pp 943–947

89.

Tran ST, Cheng CH, Liu DG (2021) A multiple layer U-Net, U-n-Net, for liver and liver tumor segmentation in CT. IEEE Access 9:3752–3764

90.

Xu HL, Wang BH, Xue WG et al (2019) Automatic segmentation of liver CT image based on dense pyramid network. In: 1st International Workshop on Multiscale Multimodal Medical Imaging (MMMI). Springer International Publishing, Shenzhen, pp 10–16

91.

Yu AH, Liu Z, Sheng VS et al (2021) CT segmentation of liver and tumors fused multi-scale features. Intell Autom Soft Comput 30:589–599

92.

Zhang Y, Tian J, Zhong C et al (2021) DARN: Deep attentive refinement network for liver tumor segmentation from 3D CT volume. In: 25th International Conference on Pattern Recognition (ICPR). IEEE Computer Society, Electrical Network, pp 7796–7803

93.

Ayalew YA, Fante KA, Mohammed MA (2021) Modified U-Net for liver cancer segmentation from computed tomography images with a new class balancing method. BMC Biomedical Engineering 3:4PubMedPubMedCentral

94.

Chen WF, Ou HY, Liu KH et al (2021) In-series U-Net network to 3D tumor image reconstruction for liver hepatocellular carcinoma recognition. Diagnostics 11:18

95.

Chung M, Lee J, Park S, Lee CE, Lee J, Shin YG (2021) Liver segmentation in abdominal CT images via auto-context neural network and self-supervised contour attention*. Artif Intell Med 113:12

96.

Elmenabawy NA, Elnakib A, Moustafa HED (2020) Deep joint segmentation of liver and cancerous nodules from Ct images2020 37th National Radio Science Conference (NRSC), pp 296–301

97.

Fan TL, Wang GL, Li Y, Wang HR (2020) MA-Net: a multi-scale attention network for liver and tumor segmentation. IEEE Access 8:179656–179665

98.

He K, Liu XM, Shahzad R et al (2021) Advanced deep learning approach to automatically segment malignant tumors and ablation zone in the liver with contrast-enhanced CT. Front Oncol 11:10

99.

Kwon J, Choi K (2020) Trainable multi-contrast windowing for liver CT segmentation. In: IEEE International Conference on Big Data and Smart Computing (BigComp). IEEE, Busan, pp 169–172

100.

Lei T, Zhou WZ, Zhang YX et al (2020) Lightweight v-net for liver segmentation. In: 2020 IEEE International Conference on Acoustics, Speech, and Signal Processing. IEEE, Barcelona, pp 1379–1383

101.

Ben-Cohen A, Klang E, Kerpel A, Konen E, Amitai MM, Greenspan H (2018) Fully convolutional network and sparsity-based dictionary learning for liver lesion detection in CT examinations. Neurocomputing 275:1585–1594

102.

Bevilacqua V, Carnimeo L, Brunetti A et al (2016) Synthesis of a neural network classifier for hepatocellular carcinoma grading based on triphasic CT images. In: 1st International Conference on Recent Trends in Image Processing and Pattern Recognition (RTIP2R). Springer-Verlag Berlin, Karnatak Arts Sci & Commerce Coll, Bidar, pp 356–368

103.

Chen L, Song H, Li Q, Cui YT, Yang J, Hu XHT (2019) Liver segmentation in CT images using a non-local fully convolutional neural network. In: IEEE International Conference on Bioinformatics and Biomedicine (BIBM). IEEE, San Diego, pp 639–642

104.

Frid-Adar M, Diamant I, Klang E, Amitai M, Goldberger J, Greenspan H (2017) Modeling the intra-class variability for liver lesion detection using a multi-class patch-based CNN. In: 3rd International Workshop on Patch-Based Techniques in Medical Images (Patch-MI). Springer International Publishing Ag, Quebec City, pp 129–137

105.

Furuzuki M, Lu HM, Kim H et al (2019) A detection method for liver cancer region based on faster R-CNN. In: 19th International Conference on Control, Automation and Systems (ICCAS). IEEE, Jeju, pp 808–811

106.

Gong H, Yu LF, Leng S et al (2019) A deep learning- and partial least square regression-based model observer for a low-contrast lesion detection task in CT. Med Phys 46:2052–2063

107.

Huang WM, Li N, Lin ZP et al (2013) Liver tumor detection and segmentation using kernel-based extreme learning machine. In: 35th Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society (EMBC). IEEE, Osaka, pp 3662–3665

108.

Jin XY, Du ZH, Zhang T, Li LJ (2017) A disease detection method of liver based on improved convolutional neural network. In: 10th International Symposium on Computational Intelligence and Design (ISCID). IEEE, Hangzhou, pp 96–98

109.

Jin XY, Jin QL, Yang X (2015) A disease detection method of liver based on improved back propagation neural network. In: 8th International Symposium on Computational Intelligence and design (ISCID). IEEE, Hangzhou, pp 111–113

110.

Kim B, Kim J, Lee J-G, Kim DH, Park SH, Ye JC (2019) Unsupervised deformable image registration using cycle-consistent CNN. In: Shen D, Liu T, Peters TM et al (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019. Springer International Publishing, Cham, pp 166–174

111.

Vivanti R, Szeskin A, Lev-Cohain N, Sosna J, Joskowicz L (2017) Automatic detection of new tumors and tumor burden evaluation in longitudinal liver CT scan studies. Int J Comput Assist Radiol Surg 12:1945–1957PubMed

112.

Tao QY, Ge ZY, Cai JF, Yin JX, See S (2019) Improving deep lesion detection using 3D contextual and spatial attention. In: 10th International Workshop on Machine Learning in Medical Imaging (MLMI) / 22nd International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI). Springer International Publishing Ag, Shenzhen, pp 185–193

113.

Liang D, Lin L, Chen X et al (2019) Multi-stream scale-insensitive convolutional and recurrent neural networks for liver tumor detection in dynamic Ct Images2019 IEEE International Conference on Image Processing (ICIP), pp 794–798

114.

Lee S-g, Bae JS, Kim H, Kim JH, Yoon S (2018) Liver lesion detection from weakly-labeled multi-phase CT volumes with a grouped single shot MultiBox detector. In: Frangi AF, Schnabel JA, Davatzikos C, Alberola-López C, Fichtinger G (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2018. Springer International Publishing, Cham, pp 693–701

115.

Yang CJ, Wang CK, Fang YD et al (2021) Clinical application of mask region-based convolutional neural network for the automatic detection and segmentation of abnormal liver density based on hepatocellular carcinoma computed tomography datasets. PLoS ONE 16:e0255605PubMedPubMedCentral

116.

Zhou J, Gandomi AH, Chen F, Holzinger A (2021) Evaluating the quality of machine learning explanations: a survey on methods and metrics. Electronics 10:593

117.

Almotairi S, Kareem G, Aouf M, Almutairi B, Salem MA (2020) Liver tumor segmentation in CT scans using modified SegNet. Sensors (Basel) 20

118.

Anter AM, Hassenian AE (2019) CT liver tumor segmentation hybrid approach using neutrosophic sets, fast fuzzy c-means and adaptive watershed algorithm. Artif Intell Med 97:105–117PubMed

119.

Chen X, Lin LF, Liang D et al (2019) A dual-attention dilated residual network for liver lesion classification and localization on CT images. In: 26th IEEE International Conference on Image Processing (ICIP). IEEE, Taipei, pp 235–239

120.

Deng ZF, Guo QZ, Zhu ZL (2019) Dynamic regulation of level set parameters using 3D convolutional neural network for liver tumor segmentation. J Healthc Eng 2019:17

121.

Huang W, Yang Y, Lin Z et al (2014) Random feature subspace ensemble based extreme learning machine for liver tumor detection and segmentation2014 36th Annual International Conference of the IEEE Engineering in Medicine and Biology Society, pp 4675–4678

122.

Kadoury S, Vorontsov E, Tang A (2015) Metastatic liver tumour segmentation from discriminant Grassmannian manifolds. Phys Med Biol 60:6459PubMed

123.

Zhou JY, Huang WM, Xiong W, Chen WY, Venkatesh SK (2013) Segmentation of hepatic tumor from abdominal CT data using an improved support vector machine framework. In: 35th Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society (EMBC). IEEE, Osaka, pp 3347–3350

124.

Zhang Y, Pan X, Li C, Wu T (2020) 3D liver and tumor segmentation with CNNs based on region and distance metrics. Appl Sci. https://doi.org/10.3390/app10113794CrossRefPubMed

125.

Zhang Y, Jiang B, Wu J et al (2020) Deep learning initialized and gradient enhanced level-set based segmentation for liver tumor from CT images. IEEE Access 8:76056–76068

126.

Zhang X, Tian J, Xiang DH, Li XL, Deng KX (2011) Interactive liver tumor segmentation from CT scans using support vector classification with watershed. In: 33rd Annual International Conference of the IEEE Engineering-in-Medicine-and-Biology-Society (EMBS). IEEE, Boston, pp 6005–6008

127.

Wu Y, Zhou Q, Hu H, Rong G, Li Y, Wang S (2019) Hepatic lesion segmentation by combining plain and contrast-enhanced CT images with modality weighted U-Net2019 IEEE International Conference on Image Processing (ICIP), pp 255–259

128.

Wei Y, Jiang X, Liu K et al (2019) A hybrid multi-atrous and multi-scale network for liver lesion detection. In: Suk H-I, Liu M, Yan P, Lian C (eds) Machine Learning in Medical Imaging. Springer International Publishing, Cham, pp 364–372

129.

Vorontsov E, Tang A, Pal C, Kadoury S (2018) Liver lesion segmentation informed by joint liver segmentation2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018), pp 1332–1335

130.

Vorontsov E, Tang A, Roy D, Pal CJ, Kadoury S (2017) Metastatic liver tumour segmentation with a neural network-guided 3D deformable model. Med Biol Eng Comput 55:127–139PubMed

131.

Todoroki Y, Iwamoto Y, Lin L, Hu H, Chen YW (2019) Automatic detection of focal liver lesions in multi-phase CT images using a multi-channel & multi-scale CNN2019 41st Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), pp 872–875

132.

Sun C, Guo S, Zhang H, Li J, Ma S, Li X (2017) Liver lesion segmentation in CT images with MK-FCN2017 IEEE 2nd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC), pp 1794–1798

133.

Shimizu A, Narihira T, Kobatake H, Furukawa D, Nawano S, Shinozaki K (2013) Ensemble learning based segmentation of metastatic liver tumours in contrast-enhanced computed tomography. IEICE Trans Inf Syst 96-D:864–868

134.

Moawad AW, Fuentes D, Khalaf AM et al (2020) Feasibility of automated volumetric assessment of large hepatocellular carcinomas’ responses to transarterial chemoembolization. Front Oncol 10:572PubMedPubMedCentral

135.

Radu C, Fisher P, Mitrea D et al (2020) Integration of real-time image fusion in the robotic-assisted treatment of hepatocellular carcinoma. Biol (Basel) 9

136.

Haq MNU, Irtaza A, Nida N, Shah MA, Zubair L (2021) Liver tumor segmentation using resnet based mask-R-CNN2021 International Bhurban Conference on Applied Sciences and Technologies (IBCAST), pp 276–281

137.

Anil BC, Dayananda P (2021) Automatic liver tumor segmentation based on multi-level deep convolutional networks and fractal residual network. IETE J Res. https://doi.org/10.1080/03772063.2021.1878066:1-9CrossRef

138.

Aslam MS, Younas M, Sarwar MU et al (2021) Liver-tumor detection using CNN ResUNet. Comput Mater Continua 67

139.

Dey R, Hong Y (2020) Hybrid cascaded neural network for liver lesion segmentation. In: IEEE 17^th International Symposium on Biomedical Imaging (ISBI). IEEE, Iowa, pp 1173–1177

140.

Hamard A, Frandon J, Larbi A et al (2020) Impact of ultra-low dose CT acquisition on semi-automated RECIST tool in the evaluation of malignant focal liver lesions. Diagn Interv Imaging 101:473–479PubMed

141.

AmirHosseini B, Hosseini R (2019) An improved fuzzy-differential evolution approach applied to classification of tumors in liver CT scan images. Med Biol Eng Comput 57:2277–2287

142.

Balagourouchetty L, Pragatheeswaran JK, Pottakkat B, G R, (2020) GoogLeNet-based ensemble FCNet classifier for focal liver lesion diagnosis. IEEE J Biomed Health Inform 24:1686–1694PubMed

143.

Cao SE, Zhang LQ, Kuang SC et al (2020) Multiphase convolutional dense network for the classification of focal liver lesions on dynamic contrast-enhanced computed tomography. World J Gastroenterol 26:3660–3672PubMedPubMedCentral

144.

Das A, Acharya UR, Panda SS, Sabut S (2019) Deep learning based liver cancer detection using watershed transform and Gaussian mixture model techniques. Cogn Syst Res 54:165–175

145.

Devi RM, Seenivasagam V (2020) Automatic segmentation and classification of liver tumor from CT image using feature difference and SVM based classifier-soft computing technique. Soft Comput 24:18591–18598

146.

Jiang HY, Zheng RP, Yi DH, Zhao D (2013) A novel multiinstance learning approach for liver cancer recognition on abdominal CT images based on CPSO-SVM and IO. Comput Math Methods Med 2013:10

147.

Jin XY, Zhang T, Li LJ, Wu HT, Sun B (2016) Lesion recognition method of liver CT images based on random forest. In: 8th International Conference on Information Technology in Medicine and Education (ITME). IEEE, Fuzhou, pp 227–230

148.

Kabe GK, Song YQ, Liu Z (2020) Optimization of FireNet for liver lesion classification. Electronics 9:16

149.

Khalili K, Lawlor RL, Pourafkari M et al (2020) Convolutional neural networks versus radiologists in characterization of small hypoattenuating hepatic nodules on CT: a critical diagnostic challenge in staging of colorectal carcinoma. Sci Rep 10:10

150.

Kumar SS, Moni RS, Rajeesh J (2013) An automatic computer-aided diagnosis system for liver tumours on computed tomography images. Comput Electr Eng 39:1516–1526

151.

Kutlu H, Avci E (2019) A novel method for classifying liver and brain tumors using convolutional neural networks, discrete wavelet transform and long short-term memory networks. Sensors 19:16

152.

Yasaka K, Akai H, Abe O, Kiryu S (2018) Deep learning with convolutional neural network for differentiation of liver masses at dynamic contrast-enhanced CT: a preliminary study. Radiology 286:887–896PubMed

153.

Sreeja P, Hariharan S (2017) Image analysis for the detection and diagnosis of hepatocellular carcinoma from abdominal CT images. In: International Conference on Internet of Things for Technological Development (IoT4TD). Springer-Verlag Singapore Pte Ltd, Gandhinagar, pp 107–117

154.

Shi WQ, Kuang SC, Cao S et al (2020) Deep learning assisted differentiation of hepatocellular carcinoma from focal liver lesions: choice of four-phase and three-phase CT imaging protocol. Abdom Radiol (NY) 45:2688–2697

155.

Romero FP, Diler A, Bisson-Gregoire G et al (2019) End-to-end discriminative deep network for liver lesion classification. In: 16th IEEE International Symposium on Biomedical Imaging (ISBI). IEEE, Venice, pp 1243–1246

156.

Renukadevi T, Karunakaran S (2020) Optimizing deep belief network parameters using grasshopper algorithm for liver disease classification. Int J Imaging Syst Technol 30:168–184

157.

Rajathi GI, Jiji GW (2019) Chronic liver disease classification using hybrid whale optimization with simulated annealing and ensemble classifier. Symmetry-Basel 11:21

158.

Peng J, Kang S, Ning Z et al (2020) Residual convolutional neural network for predicting response of transarterial chemoembolization in hepatocellular carcinoma from CT imaging. Eur Radiol 30:413–424PubMed

159.

Özyurt F, Tuncer T, Avci E, Koç M, Serhatlioğlu İ (2019) A novel liver image classification method using perceptual hash-based convolutional neural network. Arab J Sci Eng 44:3173–3182

160.

Mala K, Sadasivam V, Alagappan S (2015) Neural network based texture analysis of CT images for fatty and cirrhosis liver classification. Appl Soft Comput 32:80–86

161.

Maaref A, Romero FP, Montagnon E et al (2020) Predicting the response to FOLFOX-based chemotherapy regimen from untreated liver metastases on baseline CT: a deep neural network approach. J Digit Imaging 33:937–945PubMedPubMedCentral

162.

Li J, Sun J, Shen NY, Chen EL, Zhang YC (2019) A CAD system for liver cancer diagnosis based on multi-phase CT images features with BP network. In: 11th International Conference on Intelligent Human-Machine Systems and Cybernetics (IHMSC). IEEE, Zhejiang University, Hangzhou, pp 67–70

163.

Liang D, Lin L, Hu H et al (2018) Combining convolutional and recurrent neural networks for classification of focal liver lesions in multi-phase CT images. In: Frangi AF, Schnabel JA, Davatzikos C, Alberola-López C, Fichtinger G (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2018. Springer International Publishing, Cham, pp 666–675

164.

Thuring J, Rippel O, Haarburger C et al (2020) Multiphase CT-based prediction of Child-Pugh classification: a machine learning approach. Eur Radiol Exp 4:9

165.

Wang MY, Fu FF, Zheng BJ et al (2021) Development of an AI system for accurately diagnose hepatocellular carcinoma from computed tomography imaging data. Br J Cancer 125:1111–1121

166.

Wang Q, Wang Z, Sun Y et al (2020) SCCNN: a diagnosis method for hepatocellular carcinoma and intrahepatic cholangiocarcinoma based on Siamese cross contrast neural network. IEEE Access 8:85271–85283

167.

Xu HY, Zou XH, Zhao YN et al (2021) Differentiation of intrahepatic cholangiocarcinoma and hepatic lymphoma based on radiomics and machine learning in contrast-enhanced computer tomography. Technol Cancer Res Treat 20:7

168.

Zhang J, Huang Z, Cao L et al (2020) Differentiation combined hepatocellular and cholangiocarcinoma from intrahepatic cholangiocarcinoma based on radiomics machine learning. Ann Transl Med 8:119PubMedPubMedCentral

169.

Giannini V, Rosati S, Defeudis A et al (2020) Radiomics predicts response of individual HER2-amplified colorectal cancer liver metastases in patients treated with HER2-targeted therapy. Int J Cancer 147:3215–3223PubMed

170.

Homayounieh F, Singh R, Nitiwarangkul C et al (2020) Semiautomatic segmentation and radiomics for dual-energy CT: A pilot study to differentiate benign and malignant hepatic lesions. AJR Am J Roentgenol 215:398–405

171.

Mao B, Zhang LZ, Ning PG et al (2020) Preoperative prediction for pathological grade of hepatocellular carcinoma via machine learning-based radiomics. Eur Radiol 30:6924–6932

172.

Mokrane FZ, Lu L, Vavasseur A et al (2020) Radiomics machine-learning signature for diagnosis of hepatocellular carcinoma in cirrhotic patients with indeterminate liver nodules. Eur Radiol 30:558–570PubMed

173.

Budai BK, Tóth A, Borsos P et al (2020) Three-dimensional CT texture analysis of anatomic liver segments can differentiate between low-grade and high-grade fibrosis. BMC Med Imaging 20:108PubMedPubMedCentral

174.

Huo Y, Terry JG, Wang J et al (2019) Fully automatic liver attenuation estimation combing CNN segmentation and morphological operations. Med Phys 46:3508–3519PubMedPubMedCentral

175.

Kayaaltı Ö, Aksebzeci BH, Karahan İÖ et al (2014) Liver fibrosis staging using CT image texture analysis and soft computing. Appl Soft Comput 25:399–413

176.

Yasaka K, Akai H, Kunimatsu A, Abe O, Kiryu S (2018) Deep learning for staging liver fibrosis on CT: a pilot study. Eur Radiol 28:4578–4585PubMed

177.

Son JH, Lee SS, Lee Y et al (2020) Assessment of liver fibrosis severity using computed tomography-based liver and spleen volumetric indices in patients with chronic liver disease. Eur Radiol 30:3486–3496PubMed

178.

Yin Y, Yakar D, Dierckx R, Mouridsen KB, Kwee TC, de Haas RJ (2021) Liver fibrosis staging by deep learning: a visual-based explanation of diagnostic decisions of the model. Eur Radiol 31:9620–9627PubMedPubMedCentral

179.

Ahmadi K, Karimi A, Fouladi Nia B (2016) New technique for automatic segmentation of blood vessels in CT scan images of liver based on optimized fuzzy c-means method. Comput Math Methods Med 2016:5237191PubMedPubMedCentral

180.

Ben-Cohen A, Klang E, Amitai MM, Goldberger J, Greenspan H (2018) Anatomical data augmentation for CNN based pixel-wise classification2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018), pp 1096–1099

181.

Conze PH, Noblet V, Rousseau F et al (2017) Scale-adaptive supervoxel-based random forests for liver tumor segmentation in dynamic contrast-enhanced CT scans. Int J Comput Assist Radiol Surg 12:223–233

182.

Gensure RH, Foran DJ, Lee VM et al (2012) Evaluation of hepatic tumor response to yttrium-90 radioembolization therapy using texture signatures generated from contrast-enhanced CT images. Acad Radiol 19:1201–1207PubMedPubMedCentral

183.

Huang Q, Sun J, Ding H, Wang X, Wang G (2018) Robust liver vessel extraction using 3D U-Net with variant dice loss function. Comput Biol Med 101:153–162PubMed

184.

Zeng YZ, Zhao YQ, Liao M, Zou BJ, Wang XF, Wang W (2016) Liver vessel segmentation based on extreme learning machine. Phys Med 32:709–716PubMed

185.

Yu W, Fang B, Liu Y, Gao M, Zheng S, Wang Y (2019) Liver vessels segmentation based on 3d residual U-NET2019 IEEE International Conference on Image Processing (ICIP), pp 250–254

186.

Yang W, Lu Z, Yu M, Huang M, Feng Q, Chen W (2012) Content-based retrieval of focal liver lesions using bag-of-visual-words representations of single- and multiphase contrast-enhanced CT images. J Digit Imaging 25:708–719PubMedPubMedCentral

187.

Wang J, Han XH, Xu Y et al (2017) Sparse codebook model of local structures for retrieval of focal liver lesions using multiphase medical images. Int J Biomed Imaging 2017:1413297PubMedPubMedCentral

188.

Li Q, Yu B, Tian X, Cui X, Zhang R, Guo Q (2020) Deep residual nets model for staging liver fibrosis on plain CT images. Int J Comput Assist Radiol Surg 15:1399–1406PubMed

189.

Sun W, Qin N, Huang D, Liu Z, Ni S (2020) QN-S3VM method for evaluation of liver functional reserve2020 Chinese Automation Congress (CAC), pp 5629–5634

190.

Xu M, Wang Y, Chi Y, Hua X (2020) Training liver vessel segmentation deep neural networks on noisy labels from contrast CT imaging2020 IEEE 17th International Symposium on Biomedical Imaging (ISBI), pp 1552–1555

191.

Yang JZ, Fu MH, Hu Y (2021) Liver vessel segmentation based on inter-scale V-Net. Math Biosci Eng 18:4327–4340

192.

Yoshinobu Y, Iwamoto Y, Han XH et al (2020) Deep learning method for content-based retrieval of focal liver lesions using multiphase contrast-enhanced computer tomography images. In: IEEE International Conference on Consumer Electronics (ICCE). IEEE, Las Vegas, pp 598–601

193.

Gu J, Zhao Z, Zeng Z et al (2020) Multi-phase cross-modal learning for noninvasive gene mutation prediction in hepatocellular carcinoma42nd Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society (EMBC), Montreal, Canada, pp 5814–5817

194.

Kobe A, Zgraggen J, Messmer F et al (2021) Prediction of treatment response to transarterial radioembolization of liver metastases: radiomics analysis of pre-treatment cone-beam CT: a proof of concept study. Eur J Radiol Open 8:100375PubMedPubMedCentral

195.

Li X, Qi Z, Du H et al (2022) Deep convolutional neural network for preoperative prediction of microvascular invasion and clinical outcomes in patients with HCCs. Eur Radiol 32:771–782PubMed

196.

Ahmad M, Ai DN, Xie GW et al (2019) Deep belief network modeling for automatic liver segmentation. IEEE Access 7:20585–20595

197.

Zhang Y, Peng C, Peng L et al (2022) DeepRecS: from RECIST diameters to precise liver tumor segmentation. IEEE J Biomed Health Inform 26:614–625PubMed

198.

Zhang Y, He Z, Zhong C, Zhang Y, Shi Z (2017) Fully convolutional neural network with post-processing methods for automatic liver segmentation from CT2017 Chinese Automation Congress (CAC), pp 3864–3869

199.

Bilic P, Christ P, Li HB et al (2023) The liver tumor segmentation benchmark (LiTS). Med Image Anal 84:102680PubMed

200.

Chen XY, Zhang R, Yang PK (2019) Feature fusion encoder decoder network for automatic liver lesion segmentation. In: 16th IEEE International Symposium on Biomedical Imaging (ISBI). IEEE, Venice, pp 430–433

201.

Vivanti R, Joskowicz L, Lev-Cohain N, Ephrat A, Sosna J (2018) Patient-specific and global convolutional neural networks for robust automatic liver tumor delineation in follow-up CT studies. Med Biol Eng Comput 56:1699–1713PubMed

202.

Zhou J, Wang W, Lei B et al (2020) Automatic detection and classification of focal liver lesions based on deep convolutional neural networks: a preliminary study. Front Oncol 10:581210PubMed

203.

Adcock A, Rubin D, Carlsson G (2014) Classification of hepatic lesions using the matching metric. Comput Vis Image Underst 121:36–42

204.

Liang D, Lin LF, Hu HJ et al (2018) Residual convolutional neural networks with global and local pathways for classification of focal liver lesions. In: 15th Pacific Rim International Conference on Artificial Intelligence (PRICAI) / 15th Pacific Rim Knowledge Acquisition Workshop (PKAW). Springer International Publishing Ag, Nanjing, pp 617–628

205.

Wang W, Chen Q, Iwamoto Y et al (2020) Deep fusion models of multi-phase CT and selected clinical data for preoperative prediction of early recurrence in hepatocellular carcinoma. IEEE Access 8:139212–139220

206.

Zhang L, Xia W, Yan ZP et al (2020) Deep learning predicts overall survival of patients with unresectable hepatocellular carcinoma treated by transarterial chemoembolization plus sorafenib. Front Oncol 10:593292PubMedPubMedCentral

207.

Wang J, Han XH, Xu Y et al (2017) Tensor sparse representation of temporal features for content-based retrieval of focal liver lesions using multi-phase medical images2017 IEEE International Symposium on Multimedia (ISM), pp 507–510

208.

Group TFMCS, Bedossa P (1994) Intraobserver and interobserver variations in liver biopsy interpretation in patients with chronic hepatitis C. Hepatology 20:15–20

209.

Sterling RK, Lissen E, Clumeck N et al (2006) Development of a simple noninvasive index to predict significant fibrosis in patients with HIV/HCV coinfection. Hepatology 43:1317–1325PubMed

210.

WHO (2021) Generating evidence for artificial intelligence-based medical devices: a framework for training, validation and evaluation. World Health Organization WHO.int. Available via https://www.who.int/publications/i/item/9789240038462. Accessed 24.01.2023

Titel: Performance and clinical applicability of machine learning in liver computed tomography imaging: a systematic review
verfasst von: Keyur Radiya
Henrik Lykke Joakimsen
Karl Øyvind Mikalsen
Eirik Kjus Aahlin
Rolv-Ole Lindsetmo
Kim Erlend Mortensen
Publikationsdatum: 12.05.2023
Verlag: Springer Berlin Heidelberg
Erschienen in: European Radiology / Ausgabe 10/2023
Print ISSN: 0938-7994
Elektronische ISSN: 1432-1084
DOI: https://doi.org/10.1007/s00330-023-09609-w

Update Radiologie

Bestellen Sie unseren Fach-Newsletter und bleiben Sie gut informiert.

Newsletter bestellen

Klimawandel und Gesundheit: Was Sie in der Hausarztpraxis wissen müssen und tun können

Springer Medizin

Performance and clinical applicability of machine learning in liver computed tomography imaging: a systematic review

Abstract

Objectives

Methods

Results

Conclusion

Key Points

Supplementary Information

Publisher's note

Introduction

Methods

Results

Liver segmentation

Lesion segmentation

Lesion detection

Classification of liver or lesions

Other/miscellaneous

Discussion

Conclusion

Acknowledgements

Declarations

Guarantor

Conflict of interest

Statistics and biometry

Ethical approval

Methodology

Publisher's note

Unsere Produktempfehlungen

e.Med Interdisziplinär

e.Med Radiologie

Supplementary Information

Neu im Fachgebiet Radiologie

Darf man die Behandlung eines Neonazis ablehnen?

Ein Drittel der jungen Ärztinnen und Ärzte erwägt abzuwandern

Endlich: Zi zeigt, mit welchen PVS Praxen zufrieden sind

Akuter Schwindel: Wann lohnt sich eine MRT?

Update Radiologie

Klimawandel und Gesundheit: Was Sie in der Hausarztpraxis wissen müssen und tun können

Springer Medizin

Abstract

Objectives

Methods

Results

Conclusion

Key Points

Supplementary Information

Publisher's note

Introduction

Methods

Results

Liver segmentation

Lesion segmentation

Lesion detection

Classification of liver or lesions

Other/miscellaneous

Discussion

Conclusion

Acknowledgements

Declarations

Guarantor

Conflict of interest

Statistics and biometry

Ethical approval

Methodology

Publisher's note

Unsere Produktempfehlungen

e.Med Interdisziplinär

e.Med Radiologie

Supplementary Information

Weitere Artikel der Ausgabe 10/2023

Evaluation of the impact of a novel denoising algorithm on image quality in dual-energy abdominal CT of obese patients

Can high-resolution magnetic resonance-vessel wall imaging simplify moyamoya disease assessment?

Diagnostic accuracy of state-of-the-art rectal MRI sequences for the diagnosis of extramural vascular invasion in locally advanced rectal cancer after preoperative chemoradiotherapy: dos or maybes?

Development and validation of a deep learning radiomics nomogram for preoperatively differentiating thymic epithelial tumor histologic subtypes

Artificial intelligence–based prediction of cervical lymph node metastasis in papillary thyroid cancer with CT

Correction: Use of blood oxygen level-dependent magnetic resonance imaging to detect acute cellular rejection post-liver transplantation

Neu im Fachgebiet Radiologie

Darf man die Behandlung eines Neonazis ablehnen?

Ein Drittel der jungen Ärztinnen und Ärzte erwägt abzuwandern

Endlich: Zi zeigt, mit welchen PVS Praxen zufrieden sind

Akuter Schwindel: Wann lohnt sich eine MRT?

Update Radiologie