Skip to main content
Erschienen in:

Open Access 03.09.2024 | Review Article

The application value of Rs-fMRI-based machine learning models for differentiating mild cognitive impairment from Alzheimer's disease: a systematic review and meta-analysis

verfasst von: Chentong Wang, Li Zhou, Feng Zhou, Tingting Fu

Erschienen in: Neurological Sciences | Ausgabe 1/2025

Abstract

Background

Various machine learning (ML) models based on resting-state functional MRI (Rs-fMRI) have been developed to facilitate differential diagnosis of mild cognitive impairment (MCI) and Alzheimer's disease (AD). However, the diagnostic accuracy of such models remains understudied. Therefore, we conducted this systematic review and meta-analysis to explore the diagnostic accuracy of Rs-fMRI-based radiomics in differentiating MCI from AD.

Methods

PubMed, Embase, Cochrane, and Web of Science were searched from inception up to February 8, 2024, to identify relevant studies. Meta-analysis was conducted using a bivariate mixed-effects model, and sub-group analyses were carried out by the types of ML tasks (binary classification and multi-class classification tasks).

Findings

In total, 23 studies, comprising 5,554 participants were enrolled in the study. In the binary classification tasks (twenty studies), the diagnostic accuracy of the ML model for AD was 0.99 (95%CI: 0.34 ~ 1.00), with a sensitivity of 0.94 (95%CI: 0.89 ~ 0.97) and a specificity of 0.98 (95%CI: 0.95 ~ 1.00). In the multi-class classification tasks (six studies), the diagnostic accuracy of the ML model was 0.98 (95%CI: 0.98 ~ 0.99) for NC, 0.96 (95%CI: 0.96 ~ 0.96) for early mild cognitive impairment (EMCI), 0.97 (95%CI: 0.96 ~ 0.97) for late mild cognitive impairment (LMCI), and 0.95 (95%CI: 0.95 ~ 0.95) for AD.

Conclusions

The Rs-fMRI-based ML model can be adapted to multi-class classification tasks. Therefore, multi-center studies with large samples are needed to develop intelligent application tools to promote the development of intelligent ML models for disease diagnosis.
Hinweise

Supplementary Information

The online version contains supplementary material available at https://​doi.​org/​10.​1007/​s10072-024-07731-1.

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Abkürzungen
ML
Machine learning
Rs-fMRI
Resting-state functional MRI
MCI
Mild cognitive impairment
AD
Alzheimer's disease
EMCI
Early mild cognitive impairment
LMCI
Late mild cognitive impairment
CT
Computed tomography
PET
Positron emission tomography
sMRI
Structural magnetic resonance imaging
Rs-fMRI
Resting-state functional magnetic resonance imaging
BOLD
Blood-oxygen level-dependent
ReHo
Regional homogeneity
ALFF
Amplitude of Low Frequency Fluctuation
rsFC
Resting state functional connectivity
AI
Artificial intelligence
ML
Machine learning
DL
Deep learning
The PRISMA-DTA Statement
The Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies
RQS
Radiomics Quality Score
PLR
Positive likelihood ratio
NLR
Negative likelihood ratio
DOR
Diagnostic odds ratio
SROC
Summary ROC curve
QUADAS-2
Quality Assessment of Diagnostic Accuracy Studies-2
CDR
Clinical Dementia Rating Scale
MMSE
Modified Mental State Examination
SVM
Support vector machine
NC
Normal population
SPECT
Single photon emission computed tomography
Amyloid-β
cMCI
Converting to AD
CI
Confidence interval

Introduction

Alzheimer's disease (AD) is a progressive neurodegenerative disorder primarily affecting the brain. The disease manifests with neuroinflammatory plaques, neurogenic fiber tangles, neuronal loss and amyloid angiopathy. It is the most common cause of dementia, accounting for an estimated 60 to 80% of patients [1]. Clinically, its early symptoms include difficulty remembering recent conversations, names or events, apathy, and depression. Some patients may exhibit communication problems, confusion, poor judgment, and behavioral changes. In the advanced stages of AD, patients may develop complications such as walking, talking and swallowing [1]. The development of AD has three phases: preclinical AD, MCI due to AD, and dementia due to AD, also known as AD dementia (Fig. 1). The pre-symptomatic dementia stage of AD is known as AD-induced mild cognitive impairment (MCI). The AD stage occurs when the patient experiences a gradual progressive decline in cognitive function due to advancement in AD pathology in the brain. Significant cognitive impairment may interfere with daily functioning leading to AD dementia [2]. This pathological process is also known as the AD continuum and typically starts with subtle changes in the brain, which are not immediately noticeable [3]. MCI can be further categorized into early mild cognitive impairment (EMCI) and late mild cognitive impairment (LMCI) [4], with EMCI presenting with milder episodic memory impairment compared with LMCI. Some studies have shown that the risk of LMCI progression to AD is much higher than that of EMCI [5, 6]. The latest Global Burden of Disease study states that AD is the fifth leading cause of death worldwide [7]. In the United States, approximately 6.5 million people over the age of 65 years are affected by AD [8]. Considering the lack of interventions to prevent, slow down, or cure AD, this number is expected to increase to 13.8 million by 2060 [9]. Consequently, to prevent its detrimental effect on the society, researchers should aim to develop a strategy or intervention to control MCI, thereby reverse or delay AD. In population-based studies, an early systematic review and meta-analysis reported a 26% reversal rate [10]. This underscores the need to explore the transition from MCI to AD.
Several tools, such as computed tomography (CT), positron emission tomography (PET), structural magnetic resonance imaging (sMRI), resting-state functional magnetic resonance imaging (Rs-fMRI), and cerebrospinal fluid analysis of β-amyloid (Aβ) or tau protein (total tau [T-tau] and phosphorylated tau [P-tau]) concentrations have been suggested as potential diagnostic methods for AD. These techniques allow clinicians to comprehensively assess the brain structure and function in patients with AD at the molecular level, cellular level, metabolic level, and from a microcirculatory perspective [11]. However, these tools are currently in the exploratory stage and have various limitations in clinical application. For instance, cerebrospinal fluid (CSF) biomarker testing is clinically unacceptable to patients due to its invasive nature. CT does not accurately examine the subtle changes in brain structure and functions, and is therefore not ideal for early AD [12]. In comparison, PET can detect pathological changes in the nervous system at an early stage and is suitable for early diagnosis of AD. However, the specificity of PET in differentiating AD and MCI is poor [13], and it is unable to accurately identify AD at the subclinical stage. Although sMRI can detect structural changes in the brain, such alterations are not always fully consistent with pathologic observations in AD patients. These findings reveal various limitations in evaluating disease severity and prognosis [14]. Rs-fMRI is used to assess local brain activity and resting-state networks. It circumvents the performance problems inherent in task-state functional magnetic resonance studies [15]. The core principle of resting-state fMRI (rs-fMRI) involves quantifying fluctuations in blood-oxygen-level-dependent (BOLD) signals, which are linked to regional changes in blood flow and oxygen metabolism. These fluctuations provide indirect measures of spontaneous neuronal activity within the brain [16]. Early diagnosis of AD is critical as it allows effective treatment and improves the prognosis of AD patients. Rs-fMRI can identify cognitive changes by detecting and evaluating changes in brain structure and function in a non-invasive manner, making it suitable for early AD and differentiating MCI from AD [17, 18]. Neuroimaging techniques described above require specialized knowledge and experience in interpreting the results. This limits their widespread use in clinical practice, especially in economically underdeveloped regions. Artificial intelligence (AI) tools can analyze the data in brain images and make a diagnosis automatically through techniques such as machine learning (ML) and deep learning (DL) [19]. The AI system outperforms general radiologists and imaging interns in conducting the differential diagnosis of brain diseases, often providing accuracy that matches that of academic neuroradiologists [20]. Therefore, to generate ideas to improve the diagnosis and prevention of AD, we conducted this study to evaluate the significance of rs-fMRI combined with AI techniques [21].
This study aimed to systematically review evidence regarding the clinical value of Rs-fMRI-based ML models in differentiating MCI from AD. Specifically, we explored the accuracy, sensitivity, and specificity of these models, as well as their utility in clinical practice. The findings of this analysis will improve early diagnosis of MCI and AD through the use of Rs-fMRI-based AI learning models. They will also accelerate the establishment of a more personalized treatment plan for AD patients. The results may also provide valuable information about the pathogenesis of MCI and AD. Finally, the study will provide ideas for developing advanced Rs-fMRI-based AI learning models for the diagnosis and treatment of other neurodegenerative diseases.

Methods

Study registration

This systematic review was conducted in line with the Preferred Reporting Items for a Systematic Review and Meta-analysis of Diagnostic Test Accuracy Studies (The PRISMA-DTA Statement) and has been prospectively registered with PROSPERO (CRD42023421722).

Eligibility criteria

Inclusion criteria

The patients included in this systematic review had complete Rs-fMRI images and were diagnosed with MCI and AD. The types of studies included were cohort studies, case–control studies, and cross-sectional studies. We selected studies that were highly relevant to the topic of this meta-analysis in terms of research questions, interventions, and outcome indicators. To ensure that the results were comprehensive and generalizable, the enrolled studies were from diverse research backgrounds, cultures, and regions. Studies were considered eligible if they provided sufficient data for this meta-analysis, included necessary demographic information, interventions, and outcome measures.

Exclusion criteria

The following studies were excluded:
  • (1) Meta-analyses, reviews, guidelines, expert opinions, etc.
  • (2) Studies in which only risk factor analysis was performed and no complete ML model was constructed;
  • (3) Studies without outcomes such as recall, Roc, F1 score, c-statistic, c-index, sensitivity, specificity, accuracy, precision, confusion matrix, diagnostic tetrad, and calibration curve in the assessment of accuracy of ML models.
  • (4) Validation studies of mature scales only.
  • (5) Studies using a single factor to determine prediction accuracy.

Data sources and search strategy

A systematic search was conducted on PubMed, Cochrane, Embase, and Web of Science to identify relevant studies from the inception of these databases up to February 25, 2023. To ensure the inclusion of all relevant studies, we conducted an updated database search on February 8, 2024. The search strategy combined subject terms and free-text keywords with no geographic restrictions. The detailed search strategy is provided in Table S1.

Study selection and data extraction

The identified literature were imported into Zotero followed by exclusion of duplicates using automatic software tagging and manual tagging. Next, the studies were screened by reading the titles and abstracts. Subsequently, we downloaded the full texts of the remaining studies, and conducted further screening by reading the full texts of the studies. Before data extraction, we developed a standardized data extraction spreadsheet, containing study title, country, year of publication, first author, study type, patient source, diagnostic criteria, number of cases, validation method, and model type.
The study selection and data extraction were performed independently by two researchers. The screening and extraction results were cross-checked and in cases of disputes, a third researcher was consulted to reach a consensus.

Risk of bias assessment

The methodological quality of the included studies was evaluated using the Radiomics Quality Score (RQS) and the results were cross-checked by two independent investigators. Any disputes were addressed by consulting a third researcher. In addition, we employed the RQS, a radiomics-specific quality assessment tool, to evaluate the methodological quality of the original study design across 16 of the 36 study components.

Outcomes

The included original studies were categorized into binary classification and multi-class classification tasks based on the type of ML task. For binary classification tasks, the outcome metrics were sensitivity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), specificity, diagnostic odds ratio (DOR), and summary ROC curve (SROC). For the multi-class classification tasks, the outcome metrics were diagnostic accuracy for each category.

Synthesis methods

Statistical analysis was conducted using Stata 15.0. The c-statistic was used as the systematic evaluation index to evaluate the accuracy evaluation of the model. C-statistic and its 95% confidence interval were calculated in the meta-analysis. A random effects model was used to complete the meta-analysis of the c-statistic. Moreover, we utilized a bivariate mixed-effects model to combine the sensitivity and specificity of the model predictions to and prevent the effect of severe data imbalance on the prediction accuracy. A p value < 0.05 was considered statistically significant.

Results

Study selection

The search retrieved 15,870 articles up to February 25, 2023. Subsequently, we updated the search for all relevant studies in the databases up to February 8, 2024 to comprehensively include relevant literature and enhance the reliability of the meta-analysis results. We initially retrieved 19,155 records, from which we identified 4,440 duplicates. After removing duplicates, we screened titles and abstracts to select 45 relevant studies for full-text review. Ultimately, 23 studies met the inclusion criteria for this study [2244]. (Fig. 2).

Study characteristics

A total of 23 studies were included in this study, comprising 5,554 participants. The included studies were case–control investigations. Two of the included studies were conducted in Canada, four in China, one in Egypt, one in Germany, one in India, three in Iran, four in Korea, one in Lithuania, one in Pakistan, one in Singapore, and four in the United States. The included literature comprised 12 articles on machine learning and 13 on deep learning. Regarding classification tasks, 20 articles focused on binary classification, while six addressed multi-class classification. Among the included studies, none performed hyperparameter optimization for the reported DL and ML models. (Table 1).
Table 1
Clinical characteristics of the included studies
NO
First author
Year of publication
Country
Patient source
Diagnostic criteria for mild cognitive impairment
Diagnostic criteria for AD
Total number of Alzheimer's disease cases
Total number of MCI
Total number of cases
Generation of validation set
Number of Alzheimer's disease cases in training set
Number of MCI cases in training set
Number of cases in training set
Number of cases in validation set
Number of cases in test set
Model type
1
Yosra Kazemi
2018
Canada
ADNI
MMSE
MMSE
29
85
197
fivefold cross validation
29
85
197
  
AlexNet
2
Yu Wang
2019
China
ADNI database
  
34
35
104
tenfold cross validation
34
35
104
  
SVM, KNN
3
Bocheng Wang
2022
China
ADNI2 dataset
CDR and MMSE
CDR and MMSE
30
87
160
random sampling
  
114
16
30
 
4
SAMAN SARRAF
2016
Canada
ADNI
MMSE
MMSE
52
131
275
fivefold cross validation
52
131
275
  
MCADNNet (an optimized convolutional neural network (CNN) topology)
5
Modupe Odusami
2021
Lithuania
ADNI
  
25
63
138
random sampling
  
51,443 (number of images, 70% for training and 30% for validation in each group)
27,310
 
ResNet18(Residual Network)
6
ZHE WANG
2018
USA
Singe center
Petersen criteria
NINCDS-ADRDA criteria
10
11
33
Leave-One-Out cross-validation (LOOCV)
     
AdaBoost
7
Harshit Parmar
2020
United States
ADNI
  
30
60
120
five-fold cross validation
     
3D CNN
8
Konrad F. Waschkies
2022
Germany
Multi-center
The performance on the "recall word list" subtest in CERAD was worse than average (> 1.5 SD) CERAD
MMSE
74
132
733
tenfold cross validation
     
multi-class support vector machine
9
PR. Buvaneswari
2021
India
ADNI
MMSE score between 25–32; Wechsler Memory Scale Logical Memory II score 23–29; CDR 0.5; no significant impairment in other cognitive areas; no dementia
MMSE:20–30; CDR: 0.5–1.0
68
69
210
Unclear
     
kernel-SVR method, kSVR
10
Nazanin Beheshti
2022
USA
ADNI and OASIS
  
250
97
648
random sampling
     
CNN and Transformer
11
Sukrit Gupta
2019
Singapore
ADNI
Subjects classified as MCI have memory impairment, objective memory loss as measured by the education-adjusted score on the Wechsler Logistic Memory Scale Memory II, with no dementia and severe impairment in other cognitive domains
The criteria for AD by National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer’s Disease and Related Disorders Association
29
90
168
fivefold cross validation
     
5-layer feed-forward deep neural network (DNN)
12
Bohyun Wang
2022
Korea
ADNI
  
34
89
168
a hold-out verification
  
23
 
99
ZNN
13
Doaa Mousa
2022
Egypt
ADNI
  
114
231
512
tenfold cross validation
     
SVM
14
Tingting Zhang
2021
China
ADNI
The diagnostic criteria for MCI are as follows: (1) MMSE score between 24 and 30. (2) CDR is 0.5. (3) Memory complaints, objective memory loss measured by scores on the Educationally Adjusted Wechsler Memory Scale Logical Memory II. (4) There is no obvious impairment in other cognitive areas and the patient can remember activities of daily living (no dementia)
 
19
85
104
cross validation
     
SVM classifier
15
Mohammadmahdi Rahimiasl
2021
Iran
ADNI
The studies (Xue et al., 2019; Zhang et al., 2019) provide more information on the data collection protocol of ADNI and the diagnostic criteria for AD, MCI and HC
26
63
125
fivefold cross validation
     
L2 regularization logistic regression and linear SVM classifier
16
Farheen Ramzan
2020
Pakistan
ADNI
Cognitive testing (i.e.,MMSE) and CDR
Cognitive testing (i.e.,MMSE) and CDR
25
63
138
random sampling
  
70% ( 595,056 images)
20%(170,016 images)
10%(85,005 images)
ResNet18(Residual Network)
17
Yubraj Gupta
2020
South Korea
ADNI
MCIs group: MMSE score 25–30 points, FAQ: 0–16 points, and GDS: 0–13 points
MCIc group: MMSE: 19–30 points, FAQ: 0–18 points, GDS: 0–10 points
AD group: CDR score: 1 point, the MMSE score: 14–24 points, the FAQ score: 3–28 points; GDS score: 0–7 points
33
61
129
LOOCV
  
70%(compared to the test set)
 
30%
MKL algorithm classifier(Multiple Kernel Learning)
18
Seong-Jin Son
2017
Korea
ADNI
CDR score: 0.5, MMSE score: 24–30
CDR score: 0.5, MMSE score: 24–30
30
40
105
LOOCV
     
random forest classifier
19
Ali Khazaee
2016
Iran
ADNI
MMSE score: 24–30, memory complaints, objective memory loss as measured by the educationally adjusted score of the Wechsler Memory Scale Logical Memory II, CDR: 0.5, no significant impairment in other cognitive areas, basically retained activities of daily living, and no dementia "
MMSE score: 14–26 points, CDR: 0.5 or 1.0, and those mmet the diagnostic criteria for AD in the National Institute of Neurological and Communicative Disorders and Stroke and Alzheimer's Disease and Related Disorders Society (NINCDS/ ADRDA)
34
89
168
tenfold cross validation
     
naïve Bayesian classifier
20
Ali Khazaee
2015
Iran
ADNI
Patients with AD have a MMSE score of 14–26 and a CDR of 0.5 or 1.0, and meet the possible diagnostic standards for AD in the National Institute of Neurological and Communicative Disorders and Stroke and the Alzheimer's and Related Disorders Association (NINCDS/ADRDA)
Patients with MCI have a MMSE scores of 24–30, memory complaints, objective memory loss (measured by educationally adjusted scores on the Wechsler Memory Scale Logical Memory II), a CDR of 0.5, have no severe impairment in other cognitive areas, largely preserve activities of daily living. and have no dementia
34
89
168
Holdout cross validation
23
23
23
 
95
SVM
21
Zhuqing Long
2023
China
Singe center
CDR, MMSE and AVLT
CDR, MMSE and AVLT
44
66
168
tenfold cross validation
     
SVM
22
Saman Sarraf
2023
USA
ADNI
MMSE
MMSE
54
131
284
random sampling
  
226
27
31
CNN
23
Ju-Hyeon Noh
2023
Korea
ADNI
MMSE
MMSE
118
397
699
fivefold cross validation
     
3D-CNN
CDR Clinical dementia rating scale, MMS Mini-mental state exam, AVLT Auditory verbal learning test

Risk of bias assessment

The risk of bias in the included studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) scale [45]. All the included studies used the method of Rs-fMRI alone to classify the diseases, and there were no other measures to interfere with the classification accuracy of the model. Most included studies employed a case–control design, which can introduce population bias. Additionally, the potential for publication bias exists, as studies with positive findings may be more likely to be published than those with negative results. Furthermore, two studies from the same research group were included, which could impact the overall results. Regarding disease diagnosis, all studies used clinical scales as the basis for sample inclusion. Diagnoses such as CSF or imaging biomarkers are not included. Therefore, this clinical diagnosis-based MCI/AD cohort may have biased the results. In addition, there may be bias arising from a literature database. Most of the case data in the literature were retrieved from the ADNI database. Currently, there is no single database that comprehensively host all published medical literature, and the criteria for inclusion vary from country to country, potentially leading to biased results in systematic evaluations. Evaluating publication bias is crucial for accurately understanding research findings, informing clinical practice, enhancing study design, improving research quality, and promoting research transparency.(Fig. 3 and Table S2).

Meta analysis

Binary classification tasks

Synthesized results

Among the included studies, 20 reported binary classification tasks involving 25 models. The binary classification tasks had an average sensitivity of 0.94 (95%CI: 0.89 ~ 0.97), a specificity of 0.98 (95%CI: 0.95 ~ 1.00), a PLR of 58.8 (95%CI: 16.6 ~ 208.6), an NLR of 0.06 (95%CI: 0.03 ~ 0.11), a DOR of 1041 (95%CI: 169 ~ 6425), and an AUROC of 0.99 (95%CI: 0.34 ~ 1.00) (Fig. 4).
The percentage of AD in our included population was about 18%. Its clinical applicability was evaluated using a column-line plot. We utilized the 18% prevalence as a priori probability that in the case of a positive ML result, there is a 93% chance that the presence of AD is real, and if the ML result is negative, there is only a 1% chance that the presence of AD is real.
Of the constructed models, nine exhibited strong diagnostic performance for both AD diagnosis and exclusion, with positive and negative likelihood ratios of 10 and 0.1, respectively. Another seven models effectively diagnosed AD but had limitations in excluding it, while two models excelled at exclusion but struggled with diagnosis. The remaining six models demonstrated suboptimal performance in both diagnostic tasks.
Analysis of the Deeks’ funnel plot revealed no publication bias (p = 0.89).

Subgroup analysis

Subgroup analysis was conducted by the type of validation and type of modeling in the included studies.

Model type

Deep learning

The DL models showed a sensitivity of 0.96 (95%CI: 0.91 ~ 0.99), a specificity of 0.99 (95%CI: 0.94 ~ 1.00), a PLR of 100.9 (95%CI: 14.6 ~ 697.0), an NLR of 0.04 (95%CI: 0.01 ~ 0.10), a DOR of 2853 (95%CI: 159 ~ 51220), and an AUROC of 0.99 (95%CI: 0.56 ~ 1.00). (Figure S1 and S2).

Machine learning

ML models had a sensitivity of 0.91 (95%CI: 0.78 ~ 0.97), a specificity of 0.97 (95%CI: 0.89 ~ 0.99), a PLR of 32.6 (95%CI: 8.0 ~ 132.6), an NLR of 0.09 (95%CI: 0.03 ~ 0.23), a DOR of 361 (95%CI: 74 ~ 1756), and an AUROC of 0.98 (95%CI: 0.33 ~ 1.00). (Figure S3 and S4).
Although ML demonstrated slightly lower overall predictive performance compared to DL, it still yields promising results. The SVMs represent one of the more effective ML methods. The SVM models had a sensitivity of 0.95 (95%CI: 0.76 ~ 0.99), a specificity of 0.95 (95%CI: 0.84 ~ 0.99), a PLR of 20.4 (95%CI: 5.9 ~ 71.1), an NLR of 0.06 (95%CI: 0.01 ~ 0.28), a DOR of 367 (95%CI: 58 ~ 2324), and an AUROC of 0.99 (95%CI: 0.33 ~ 1.00). (Figure S5 and S6).

Type of validation

Cross-validation

Cross-validation methods were used in 14 of the 25 models, with a sensitivity of 0.96 (95%CI: 0.89 ~ 0.98), a specificity of 0.99 (95%CI: 0.93 ~ 1.00), a PLR of 96.8 (95%CI: 13.6 ~ 689.3), an NLR of 0.04 (95%CI: 0.02 ~ 0.11), a DOR of 2184 (95%CI: 143 ~ 33,329), and an AUROC of 0.99 (95%CI: 0.34 ~ 1.00). (Figure S7 and S8).

Test set/validation set

Among the 25 models, 11 models were validated in either the test set or the validation set, with a sensitivity of 0.93 (95%CI: 0.81 ~ 0.97), a specificity of 0.97 (95%CI: 0.88 ~ 0.99), a PLR of 29.4 (95%CI: 7.3 ~ 118.9), an NLR of 0.08 (95%CI: 0.03 ~ 0.22), a DOR of 386 (95%CI: 46 ~ 3269), and an AUROC of 0.98 (95%CI: 0.63 ~ 1.00). (Figure S9 and S10).

Multi-class classification tasks

In the multi-class classification tasks, the accuracy was 98.5% in the normal population (NC), 96.2% in the EMCI, 96.6% in LMCI, and 94.8% in AD. (Fig. 5).

Discussion

Summary of the main findings

Regarding the ML performance in binary classification tasks for AD diagnosis, this systematic review found that ML can effectively diagnose AD with a sensitivity of 0.91 and a specificity of 0.97. This demonstrates that the ML model can complete the disease diagnosis task based on Rs-fMRI.
The DL method had higher diagnostic accuracy compared to traditional ML methods with a sensitivity and specificity of 0.96 and 0.99, respectively, both of which were higher than that of ML model, suggesting that it has better performance in disease diagnosis.
The support vector machines (SVM) were found to be the most effective in predicting disease among the ML models, with a sensitivity of 0.95, which was higher compared with the average value of ML.
Further analysis revealed that the predictive accuracy of cross-validation methods was higher than that of the test set or validation set validation methods. Such methods exhibited good diagnostic accuracy for different populations (NC, EMCL, LMCL, AD) in multi-class classification tasks.

Comparison with previous reviews

Neuroimaging techniques, including sMRI, fMRI, PET, and single photon emission computed tomography (SPECT) have been widely applied in AD diagnosis. These modalities are particularly valuable for early disease detection and differentiating AD subtypes. Among these, PET and sMRI are the most commonly utilized imaging techniques.
PET can detect pathological changes in the nervous system at an early stage to achieve early and accurate diagnosis of AD. PET can also provide information about the function, metabolism, and perfusion status of the nervous system, which can facilitate accurate determination of the severity of the disease, as well as the efficacy of treatments. Positron emission tomography with 2-deoxy-2-[fluorine-18] fluoro-D-glucose coupled with computed tomography (18F-FDG PET) has been used as an adjunct for AD diagnosis for more than 20 years. AD is characterized with low glucose metabolism in the temporoparietal and posterior cingulate regions [46]. Multicenter studies have reported that 18F-FDG PET correctly classifies 95% of AD patients, and can effectively predict the transition of MCI to AD [47]. In contrast, radiotracers that detect amyloid-beta (Aβ) accumulation, tau protein aggregation, and neuroinflammation more directly reflect the pathological state of individuals with AD or MCI progressing to AD (cMCI) [48]. While both are valuable, amyloid-beta (Aβ) imaging with radiotracers currently holds a more established clinical role compared to measures of gray matter atrophy and reduced cerebral glucose metabolism in predicting the progression from MCI to AD [49]. It is, therefore, valuable in the early diagnosis of AD. A previous meta-analysis comprising 5,967 patients explored the diagnostic value of Aβ-PET in AD [13], and reported a sensitivity, specificity, DOR, and AUC of 0.90, 0.80, 35.68, and 0.91, respectively. Subgroup analyses showed that Aβ-PET had a high sensitivity (0.91) and specificity (0.81) in distinguishing AD from normal controls. However, its specificity in differentiating AD from MCI was poor (0.49). Moreover, its sensitivity and specificity in predicting the transition from MCI to AD were 0.84 and 0.62, respectively. Compared with PET, Rs-fMRI had better performance in predicting the progression from MCI to AD.
sMRI can be applied in the detection and evaluation of neurodegenerative lesions in patients with AD owing to its ability to provide detailed visualization of brain structures [14]. For example, sMRI can diagnose AD by measuring the volume and morphology of the hippocampus, a reduction in hippocampus volume is an important feature of AD. In addition, sMRI can detect whole-brain atrophy associated with AD, especially in the medial temporal lobe [50]. However, limitations such as the high dimensionality of raw sMRI images and the lack of obvious structural changes in early AD may decrease its performance in early assessment and diagnosis of the disease. A previous review involving 3,935 participants explored the sensitivity and specificity of sMRI in diagnosing AD [14]. The volume of the total hippocampus was reported in most of the included literature with a pooled mean sensitivity of 0.73 (95% confidence interval (CI) 0.64 to 0.80) and a pooled mean specificity of 0.71 (95% CI 0.65 to 0.77). Atrophy of the medial temporal lobe was reported in some of the included studies with a mean sensitivity of 0.64 (95% CI 0.53 to 0.73) and a mean specificity of 0.65 (95% CI 0.51 to 0.76), involving 1,077 participants. Five studies reported the volume of lateral ventricles, demonstrating a mean sensitivity of 0.57 (95% CI: 0.49–0.65) and a mean specificity of 0.64 (95% CI: 0.59–0.70). Analysis of the most extensively studied brain regions, the hippocampus and medial temporal lobe, revealed low sensitivity and specificity for volumetry as an independent biomarker using sMRI for early AD diagnosis in individuals with MCI.
In this review, we found that Rs-fMRI provided a better diagnosis of early AD compared with CT and sMRI. In addition, Rs-fMRI was more effective in predicting the transition from MCI to AD compared with PET. The advantages of Rs-fMRI in diagnosing AD are detailed in Table 2. As stated in the introduction, measures for evaluating regional brain activity include rsFC, ReHo, and ALFF analysis, all of which can describe Rs-fMRI images. ReHo describes intra-regional coherence and rsFC describes synchronization between regions. Notably, none of them can directly describe the intensity of brain activity in a region, i.e., activity detection is not possible. ALFF reveals the intensity of the BOLD signal for spontaneous regional activity [51]. One study included in our analysis reported the use of fractional anisotropy (FC) as an evaluation indicator [40]. While current imaging techniques have limitations, ongoing technological advancements are expected to expand the application of brain imaging in AD diagnosis and treatment, leading to more effective therapeutic and preventive strategies.
Table 2
Comparison of neuroimaging techniques in the diagnosis of AD
Method
Identifying AD
Early identification
Identifying the ability of MCI to convert to AD
PET
 + 
 + 
-
sMRI
 + 
-
-
Rs-fMRI
 + 
 + 
 + 
The types of tasks in ML are grouped into two: binary classification and multi-class classification. The number of binary classification tasks in the included studies was much higher than that of multi-class classification tasks. However, in clinical practice, clinicians tend to focus more on multi-class classification tasks. This is because during differential diagnosis, AD should be differentiated from several other diseases with similar manifestations, and the stage of the disease and degree of impairment need to be clarified. This cannot be achieved using a simple binary classification task. Therefore, more complex multi-class classification tasks should be designed to enhance the clinical diagnosis of disease.
In this systematic review, data were obtained from included studies, eliminating the need for manual coding. Our analysis confirmed the superior diagnostic performance of DL compared to traditional methods. By intelligently extracting image data, DL can mitigate the variability inherent in manually defined regions of interest. Future research should focus on advancing DL theories to develop sophisticated image analysis tools.

Advantages and limitations of the study

Rs-fMRI provide a non-invasive approach for assessing the functional brain connectivity with potential to detect early damage in AD. MCI occurs during the early stage of AD. In this systematic review, we identified the model types and validation types with high diagnostic accuracy for AD. Moreover, we examined binary and multi-class classification tasks separately and demonstrated the significance of multi-class classification in clinical setting. To minimize limitations associated with the heterogeneity among the studies, we performed subgroup analysis to compare the performance of DL and ML models.
Despite the strong diagnostic evidence mentioned above, this systematic review and meta-analysis has several limitations. One of them is the potential for sample bias due to the heavy reliance on the ADNI database, making it difficult to assess population overlap. The inclusion of only a limited number of case–control studies further restricted the analysis. Notably, two studies by Ali Khazaee et al. [40, 41] were included, both exploring deep learning and machine learning models. This overlap might have influenced the observed lower sensitivity of machine learning models compared to deep learning approaches.
Secondly, there was a potential publication bias in the patient-based analysis. Some unimportant or undesirable results are not always reported. However, such results may affect the diagnostic performance.
Thirdly, several clinical tools have been developed to facilitate the diagnosis of AD, but we only focused on the diagnostic performance of Rs-fMRI. For example, besides imaging methods, cerebrospinal fluid biomarkers are commonly employed in the diagnosis of AD. Unfortunately, none of the studies included in this study used cerebrospinal fluid biomarkers (e.g., CSF/ pet-amyloid biomarkers, etc.) for the diagnosis of AD. This may be due to the fact that CSF is collected using invasive procedures that are not often accepted by patients. There is Only one study reported the use of lumbar puncture to assess their CSF/pet -amyloid biomarker levels [29]. In the study, they compared the gap between CSF/pet -amyloid biomarkers and the predictive performance of RsfMRI. The patient inclusion criteria did not require CSF or PET amyloid biomarker assessment. Additionally, the use of the Clinical Dementia Rating Scale (CDR) and the Modified Mini-Mental State Examination (MMSE) for diagnosing MCI and AD introduced potential methodological biases. To enhance the generalizability and predictive accuracy of future models, larger and more diverse datasets incorporating amyloid biomarker data are warranted.
Fourthly, each type of MCI has a unique etiology. Various factors have been reported including vascular disease, Parkinson's disease (PD), dementia with Lewy bodies (DLB), frontotemporal lobe dementia (FTLD), long-term sleep deprivation, and heavy metal poisoning. Therefore, the progression of MCI to AD may be influenced by several risk factors such as hypertension, diabetes, obesity, early malnutrition, hearing loss, smoking, and alcohol consumption. The studies enrolled in our study did not differentiate the different causes of MCI. The absence of amyloid biomarkers to definitively confirm AD etiology may have introduced bias in diagnosing MCI, potentially affecting the performance of ML and DL models.
Finally, our study adopted a single-modality approach, relying solely on Rs-fMRI data. While we acknowledge the potential benefits of multimodal imaging, particularly in enhancing diagnostic accuracy and classification, for this study, we focused on exploring the capabilities of rs-fMRI. Other research has demonstrated the value of multimodal approaches, such as combining resting-state FDG-PET with fMRI, to investigate the relationship between brain glucose metabolism and activity in AD patients [52]. In addition, a previous study demonstrated that hippocampal subregion and amygdala volumes obtained by sMRI in combination with brain network features and multiple measurement features from Rs-fMRI could effectively achieve early diagnosis and classification of AD [35]. A multimodal approach using a combination of diagnostic tools can provide higher diagnostic accuracy, facilitate early diagnosis, reveal the pathological mechanisms, and reduce misdiagnosis rates. Therefore, we will also explore the possibility of combining Rs-fMRI data with other biomarkers in subsequent studies.

Conclusions

In summary, several studies have explored various techniques for diagnosing MCI and AD in recent years. With the development of brain imaging technology, several tests have been established. Among them, Rs-fMRI has attracted attention because of its non-invasive nature and its ability to recognize early development in AD. The emergence of AI tools has revolutionized various aspects of AD detection and management. For underdeveloped regions where medical resources are relatively scarce, early identification of AD can be achieved using AI methods, which will allow establishment of a better prognosis.
The results of this review demonstrate that Rs-fMRI can diagnosis AD and its accuracy can be improved through AI methods. However, there are some limitations of this study. Therefore, future prospective, multicenter studies are advocated to explore the potential for application of this technique in routine AD diagnosis.

Acknowledgements

The authors have no acknowledgments to report.

Research involving human participants and/or animals 

Not applicable—this study does not involve humans or animals.

Declarations

Not applicable.
For this type of study, informed consent is not required.

Conflicts of interest

The authors have no conflict of interest to report.
Open Access This 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.

e.Med Neurologie & Psychiatrie

Kombi-Abonnement

Mit e.Med Neurologie & Psychiatrie erhalten Sie Zugang zu CME-Fortbildungen der Fachgebiete, den Premium-Inhalten der dazugehörigen Fachzeitschriften, inklusive einer gedruckten Zeitschrift Ihrer Wahl.

e.Med Neurologie

Kombi-Abonnement

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

Literatur
1.
2.
Zurück zum Zitat Jack CR Jr, Albert MS, Knopman DS, McKhann GM, Sperling RA, Carrillo MC, Thies B, Phelps CH (2011) Introduction to the recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7(3):257–262PubMedCrossRef Jack CR Jr, Albert MS, Knopman DS, McKhann GM, Sperling RA, Carrillo MC, Thies B, Phelps CH (2011) Introduction to the recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7(3):257–262PubMedCrossRef
3.
Zurück zum Zitat Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, Gamst A, Holtzman DM, Jagust WJ, Petersen RC, Snyder PJ, Carrillo MC, Thies B, Phelps CH (2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7(3):270–279PubMedCrossRef Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, Gamst A, Holtzman DM, Jagust WJ, Petersen RC, Snyder PJ, Carrillo MC, Thies B, Phelps CH (2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: recommendations from the National Institute on Aging-Alzheimer’s Association workgroups on diagnostic guidelines for Alzheimer’s disease. Alzheimers Dement 7(3):270–279PubMedCrossRef
4.
Zurück zum Zitat Zhou Y (2021) Imaging and multiomic biomarker applications: advances in early alzheimer’s disease. Nova Science Publishers Zhou Y (2021) Imaging and multiomic biomarker applications: advances in early alzheimer’s disease. Nova Science Publishers
5.
Zurück zum Zitat Jessen F, Wolfsgruber S, Wiese B, Bickel H, Mösch E, Kaduszkiewicz H, Pentzek M, Riedel-Heller SG, Luck T, Fuchs A, Weyerer S, Werle J, van den Bussche H, Scherer M, Maier W, Wagner M (2014) AD dementia risk in late MCI, in early MCI, and in subjective memory impairment. Alzheimers Dement 10(1):76–83PubMedCrossRef Jessen F, Wolfsgruber S, Wiese B, Bickel H, Mösch E, Kaduszkiewicz H, Pentzek M, Riedel-Heller SG, Luck T, Fuchs A, Weyerer S, Werle J, van den Bussche H, Scherer M, Maier W, Wagner M (2014) AD dementia risk in late MCI, in early MCI, and in subjective memory impairment. Alzheimers Dement 10(1):76–83PubMedCrossRef
8.
Zurück zum Zitat Omura JD, McGuire LC, Patel R, Baumgart M, Lamb R, Jeffers EM, Olivari BS, Croft JB, Thomas CW, Hacker K (2022) Modifiable Risk Factors for Alzheimer Disease and Related Dementias Among Adults Aged ≥45 Years - United States, 2019. MMWR Morb Mortal Wkly Rep 71(20):680–685PubMedPubMedCentralCrossRef Omura JD, McGuire LC, Patel R, Baumgart M, Lamb R, Jeffers EM, Olivari BS, Croft JB, Thomas CW, Hacker K (2022) Modifiable Risk Factors for Alzheimer Disease and Related Dementias Among Adults Aged ≥45 Years - United States, 2019. MMWR Morb Mortal Wkly Rep 71(20):680–685PubMedPubMedCentralCrossRef
10.
Zurück zum Zitat Canevelli M, Grande G, Lacorte E, Quarchioni E, Cesari M, Mariani C, Bruno G, Vanacore N (2016) Spontaneous Reversion of Mild Cognitive Impairment to Normal Cognition: A Systematic Review of Literature and Meta-Analysis. J Am Med Dir Assoc 17(10):943–948PubMedCrossRef Canevelli M, Grande G, Lacorte E, Quarchioni E, Cesari M, Mariani C, Bruno G, Vanacore N (2016) Spontaneous Reversion of Mild Cognitive Impairment to Normal Cognition: A Systematic Review of Literature and Meta-Analysis. J Am Med Dir Assoc 17(10):943–948PubMedCrossRef
11.
Zurück zum Zitat Dubois B, Epelbaum S, Nyasse F, Bakardjian H, Gagliardi G, Uspenskaya O, Houot M, Lista S, Cacciamani F, Potier MC, Bertrand A, Lamari F, Benali H, Mangin JF, Colliot O, Genthon R, Habert MO, Hampel H (2018) Cognitive and neuroimaging features and brain β-amyloidosis in individuals at risk of Alzheimer’s disease (INSIGHT-preAD): a longitudinal observational study. Lancet Neurol 17(4):335–346PubMedCrossRef Dubois B, Epelbaum S, Nyasse F, Bakardjian H, Gagliardi G, Uspenskaya O, Houot M, Lista S, Cacciamani F, Potier MC, Bertrand A, Lamari F, Benali H, Mangin JF, Colliot O, Genthon R, Habert MO, Hampel H (2018) Cognitive and neuroimaging features and brain β-amyloidosis in individuals at risk of Alzheimer’s disease (INSIGHT-preAD): a longitudinal observational study. Lancet Neurol 17(4):335–346PubMedCrossRef
12.
Zurück zum Zitat Petrella JR, Coleman RE, Doraiswamy PM (2003) Neuroimaging and early diagnosis of Alzheimer disease: a look to the future. Radiology 226(2):315–336PubMedCrossRef Petrella JR, Coleman RE, Doraiswamy PM (2003) Neuroimaging and early diagnosis of Alzheimer disease: a look to the future. Radiology 226(2):315–336PubMedCrossRef
13.
Zurück zum Zitat Ruan D, Sun L (2023) Amyloid-β PET in Alzheimer’s disease: A systematic review and Bayesian meta-analysis. Brain Behav 13(1):e2850PubMedCrossRef Ruan D, Sun L (2023) Amyloid-β PET in Alzheimer’s disease: A systematic review and Bayesian meta-analysis. Brain Behav 13(1):e2850PubMedCrossRef
14.
Zurück zum Zitat Lombardi G, Crescioli G, Cavedo E, Lucenteforte E, Casazza G, Bellatorre AG, Lista C, Costantino G, Frisoni G, Virgili G, Filippini G (2020) Structural magnetic resonance imaging for the early diagnosis of dementia due to Alzheimer’s disease in people with mild cognitive impairment. Cochrane Database Syst Rev 3(3):009628 Lombardi G, Crescioli G, Cavedo E, Lucenteforte E, Casazza G, Bellatorre AG, Lista C, Costantino G, Frisoni G, Virgili G, Filippini G (2020) Structural magnetic resonance imaging for the early diagnosis of dementia due to Alzheimer’s disease in people with mild cognitive impairment. Cochrane Database Syst Rev 3(3):009628
15.
Zurück zum Zitat Gonneaud J, Baria AT, Pichet Binette A, Gordon BA, Chhatwal JP, Cruchaga C, Jucker M, Levin J, Salloway S, Farlow M, Gauthier S, Benzinger TLS, Morris JC, Bateman RJ, Breitner JCS, Poirier J, Vachon-Presseau E, Villeneuve S (2021) Accelerated functional brain aging in pre-clinical familial Alzheimer’s disease. Nat Commun 12(1):5346PubMedPubMedCentralCrossRef Gonneaud J, Baria AT, Pichet Binette A, Gordon BA, Chhatwal JP, Cruchaga C, Jucker M, Levin J, Salloway S, Farlow M, Gauthier S, Benzinger TLS, Morris JC, Bateman RJ, Breitner JCS, Poirier J, Vachon-Presseau E, Villeneuve S (2021) Accelerated functional brain aging in pre-clinical familial Alzheimer’s disease. Nat Commun 12(1):5346PubMedPubMedCentralCrossRef
16.
Zurück zum Zitat Beheshtian E, Jalilianhasanpour R, Modir Shanechi A, Sethi V, Wang G, Lindquist MA, Caffo BS, Agarwal S, Pillai JJ, Gujar SK, Sair HI (2021) Identification of the Somatomotor Network from Language Task-based fMRI Compared with Resting-State fMRI in Patients with Brain Lesions. Radiology 301(1):178–184PubMedCrossRef Beheshtian E, Jalilianhasanpour R, Modir Shanechi A, Sethi V, Wang G, Lindquist MA, Caffo BS, Agarwal S, Pillai JJ, Gujar SK, Sair HI (2021) Identification of the Somatomotor Network from Language Task-based fMRI Compared with Resting-State fMRI in Patients with Brain Lesions. Radiology 301(1):178–184PubMedCrossRef
17.
Zurück zum Zitat Shinn M, Hu A, Turner L, Noble S, Preller KH, Ji JL, Moujaes F, Achard S, Scheinost D, Constable RT, Krystal JH, Vollenweider FX, Lee D, Anticevic A, Bullmore ET, Murray JD (2023) Functional brain networks reflect spatial and temporal autocorrelation. Nat Neurosci 26(5):867–878PubMedCrossRef Shinn M, Hu A, Turner L, Noble S, Preller KH, Ji JL, Moujaes F, Achard S, Scheinost D, Constable RT, Krystal JH, Vollenweider FX, Lee D, Anticevic A, Bullmore ET, Murray JD (2023) Functional brain networks reflect spatial and temporal autocorrelation. Nat Neurosci 26(5):867–878PubMedCrossRef
18.
Zurück zum Zitat Zang YF, Zuo XN, Milham M, Hallett M (2015) Toward a Meta-Analytic Synthesis of the Resting-State fMRI Literature for Clinical Populations. Biomed Res Int 2015:435265PubMedPubMedCentralCrossRef Zang YF, Zuo XN, Milham M, Hallett M (2015) Toward a Meta-Analytic Synthesis of the Resting-State fMRI Literature for Clinical Populations. Biomed Res Int 2015:435265PubMedPubMedCentralCrossRef
19.
Zurück zum Zitat Greener JG, Kandathil SM, Moffat L, Jones DT (2022) A guide to machine learning for biologists. Nat Rev Mol Cell Biol 23(1):40–55PubMedCrossRef Greener JG, Kandathil SM, Moffat L, Jones DT (2022) A guide to machine learning for biologists. Nat Rev Mol Cell Biol 23(1):40–55PubMedCrossRef
20.
Zurück zum Zitat Rauschecker AM, Rudie JD, Xie L, Wang J, Duong MT, Botzolakis EJ, Kovalovich AM, Egan J, Cook TC, Bryan RN, Nasrallah IM, Mohan S, Gee JC (2020) Artificial Intelligence System Approaching Neuroradiologist-level Differential Diagnosis Accuracy at Brain MRI. Radiology 295(3):626–637PubMedCrossRef Rauschecker AM, Rudie JD, Xie L, Wang J, Duong MT, Botzolakis EJ, Kovalovich AM, Egan J, Cook TC, Bryan RN, Nasrallah IM, Mohan S, Gee JC (2020) Artificial Intelligence System Approaching Neuroradiologist-level Differential Diagnosis Accuracy at Brain MRI. Radiology 295(3):626–637PubMedCrossRef
21.
Zurück zum Zitat Tengeler AC, Dam SA, Wiesmann M, Naaijen J, van Bodegom M, Belzer C, Dederen PJ, Verweij V, Franke B, Kozicz T, Arias Vasquez A, Kiliaan AJ (2020) Gut microbiota from persons with attention-deficit/hyperactivity disorder affects the brain in mice. Microbiome 8(1):44PubMedPubMedCentralCrossRef Tengeler AC, Dam SA, Wiesmann M, Naaijen J, van Bodegom M, Belzer C, Dederen PJ, Verweij V, Franke B, Kozicz T, Arias Vasquez A, Kiliaan AJ (2020) Gut microbiota from persons with attention-deficit/hyperactivity disorder affects the brain in mice. Microbiome 8(1):44PubMedPubMedCentralCrossRef
22.
Zurück zum Zitat Kazemi Y, Houghten S (2018) A deep learning pipeline to classify different stages of Alzheimer's disease from fMRI data. In: 2018 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB), St. Louis, MO, USA, pp 1–8. https://doi.org/10.1109/CIBCB.2018.8404980 Kazemi Y, Houghten S (2018) A deep learning pipeline to classify different stages of Alzheimer's disease from fMRI data. In: 2018 IEEE Conference on Computational Intelligence in Bioinformatics and Computational Biology (CIBCB), St. Louis, MO, USA, pp 1–8. https://​doi.​org/​10.​1109/​CIBCB.​2018.​8404980
24.
Zurück zum Zitat Wang B, Li L, Peng L, Jiang Z, Dai K, Xie Q, Cao Y, Yu D (2022) Multigroup recognition of dementia patients with dynamic brain connectivity under multimodal cortex parcellation. Biomed Signal Process Control 76:103725CrossRef Wang B, Li L, Peng L, Jiang Z, Dai K, Xie Q, Cao Y, Yu D (2022) Multigroup recognition of dementia patients with dynamic brain connectivity under multimodal cortex parcellation. Biomed Signal Process Control 76:103725CrossRef
25.
Zurück zum Zitat Sarraf S, Desouza DD, Anderson J, Saverino C (2019) MCADNNet: Recognizing Stages of Cognitive Impairment through Efficient Convolutional fMRI and MRI Neural Network Topology Models. IEEE Access 7:155584–155600PubMedPubMedCentralCrossRef Sarraf S, Desouza DD, Anderson J, Saverino C (2019) MCADNNet: Recognizing Stages of Cognitive Impairment through Efficient Convolutional fMRI and MRI Neural Network Topology Models. IEEE Access 7:155584–155600PubMedPubMedCentralCrossRef
26.
Zurück zum Zitat Odusami M, Maskeliūnas R, Damaševičius R, Krilavičius T (2021) Analysis of Features of Alzheimer’s Disease: Detection of Early Stage from Functional Brain Changes in Magnetic Resonance Images Using a Finetuned ResNet18 Network. Diagnostics (Basel). 11(6):1097 Odusami M, Maskeliūnas R, Damaševičius R, Krilavičius T (2021) Analysis of Features of Alzheimer’s Disease: Detection of Early Stage from Functional Brain Changes in Magnetic Resonance Images Using a Finetuned ResNet18 Network. Diagnostics (Basel). 11(6):1097
27.
Zurück zum Zitat Wang Z, Zheng Y, Zhu DC, Bozoki AC, Li T (2018) Classification of Alzheimer’s Disease, Mild Cognitive Impairment and Normal Control Subjects Using Resting-State fMRI Based Network Connectivity Analysis, IEEE J Transl Eng. Health Med 6:1801009 Wang Z, Zheng Y, Zhu DC, Bozoki AC, Li T (2018) Classification of Alzheimer’s Disease, Mild Cognitive Impairment and Normal Control Subjects Using Resting-State fMRI Based Network Connectivity Analysis, IEEE J Transl Eng. Health Med 6:1801009
28.
Zurück zum Zitat Parmar H, Nutter B, Long R, Antani S, Mitra S (2020) Spatiotemporal feature extraction and classification of Alzheimer’s disease using deep learning 3D-CNN for fMRI data. J Med Imaging (Bellingham) 7(5);056001PubMed Parmar H, Nutter B, Long R, Antani S, Mitra S (2020) Spatiotemporal feature extraction and classification of Alzheimer’s disease using deep learning 3D-CNN for fMRI data. J Med Imaging (Bellingham) 7(5);056001PubMed
29.
Zurück zum Zitat Waschkies KF, Soch J, Darna M, Richter A, Altenstein S, Beyle A, Brosseron F, Buchholz F, Butryn M, Dobisch L, Ewers M, Fliessbach K, Gabelin T, Glanz W, Goerss D, Gref D, Janowitz D, Kilimann I, Lohse A, Munk MH, Rauchmann B-S, Rostamzadeh A, Roy N, Spruth EJ, Dechent P, Heneka MT, Hetzer S, Ramirez A, Scheffler K, Buerger K, Laske C, Perneczky R, Peters O, Priller J, Schneider A, Spottke A, Teipel S, Düzel E, Jessen F, Wiltfang J, Schott BH, Kizilirmak JM (2022) Big five, self-reported depression, and anxiety are predictive for Alzheimer’s disease. medRxiv 2022–11. https://www.medrxiv.org/content/10.1101/2022.11.30.22282930v4.full-text Waschkies KF, Soch J, Darna M, Richter A, Altenstein S, Beyle A, Brosseron F, Buchholz F, Butryn M, Dobisch L, Ewers M, Fliessbach K, Gabelin T, Glanz W, Goerss D, Gref D, Janowitz D, Kilimann I, Lohse A, Munk MH, Rauchmann B-S, Rostamzadeh A, Roy N, Spruth EJ, Dechent P, Heneka MT, Hetzer S, Ramirez A, Scheffler K, Buerger K, Laske C, Perneczky R, Peters O, Priller J, Schneider A, Spottke A, Teipel S, Düzel E, Jessen F, Wiltfang J, Schott BH, Kizilirmak JM (2022) Big five, self-reported depression, and anxiety are predictive for Alzheimer’s disease. medRxiv 2022–11. https://​www.​medrxiv.​org/​content/​10.​1101/​2022.​11.​30.​22282930v4.​full-text
30.
Zurück zum Zitat Buvaneswari PR, Gayathri R (2023) Detection and Classification of Alzheimer’s disease from cognitive impairment with resting-state fMRI. Neural Comput Appl 35(31):22797–22812CrossRef Buvaneswari PR, Gayathri R (2023) Detection and Classification of Alzheimer’s disease from cognitive impairment with resting-state fMRI. Neural Comput Appl 35(31):22797–22812CrossRef
31.
Zurück zum Zitat Beheshti N, Johnsson L (2022) Classification of 4D fMRI images using ML, focusing on computational and memory utilization efficiency. In: Xu X, Li X, Mahapatra D, Cheng L, Petitjean C, Fu H (eds) Resource-Efficient Medical Image Analysis REMIA. Lect Notes Comput Sci, vol 13543. Springer, Cham. https://doi.org/10.1007/978-3-031-16876-5_6 Beheshti N, Johnsson L (2022) Classification of 4D fMRI images using ML, focusing on computational and memory utilization efficiency. In: Xu X, Li X, Mahapatra D, Cheng L, Petitjean C, Fu H (eds) Resource-Efficient Medical Image Analysis REMIA. Lect Notes Comput Sci, vol 13543. Springer, Cham. https://​doi.​org/​10.​1007/​978-3-031-16876-5_​6
32.
Zurück zum Zitat Gupta S, Chan YH, Rajapakse JC (2019) Decoding Brain Functional Connectivity Implicated in AD and MCI. In: Shen D, Liu T, Peters TM, Staib LH, Essert C, Zhou S, Yap P-T, Khan A (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019. Springer International Publishing, Cham, pp 781–789CrossRef Gupta S, Chan YH, Rajapakse JC (2019) Decoding Brain Functional Connectivity Implicated in AD and MCI. In: Shen D, Liu T, Peters TM, Staib LH, Essert C, Zhou S, Yap P-T, Khan A (eds) Medical Image Computing and Computer Assisted Intervention – MICCAI 2019. Springer International Publishing, Cham, pp 781–789CrossRef
34.
Zurück zum Zitat Mousa D, Zayed N, Yassine IA (2022) Alzheimer disease stages identification based on correlation transfer function system using resting-state functional magnetic resonance imaging. PLoS ONE 17(4):e0264710PubMedPubMedCentralCrossRef Mousa D, Zayed N, Yassine IA (2022) Alzheimer disease stages identification based on correlation transfer function system using resting-state functional magnetic resonance imaging. PLoS ONE 17(4):e0264710PubMedPubMedCentralCrossRef
35.
Zurück zum Zitat Zhang T, Liao Q, Zhang D, Zhang C, Yan J, Ngetich R, Zhang J, Jin Z, Li L (2021) Predicting MCI to AD Conversation Using Integrated sMRI and rs-fMRI: Machine Learning and Graph Theory Approach. Front Aging Neurosci 13:688926PubMedPubMedCentralCrossRef Zhang T, Liao Q, Zhang D, Zhang C, Yan J, Ngetich R, Zhang J, Jin Z, Li L (2021) Predicting MCI to AD Conversation Using Integrated sMRI and rs-fMRI: Machine Learning and Graph Theory Approach. Front Aging Neurosci 13:688926PubMedPubMedCentralCrossRef
36.
Zurück zum Zitat Rahimiasl M, Moghadam Charkari N, Ghaderi F (2021) Random walks on B distributed resting-state functional connectivity to identify Alzheimer’s disease and Mild Cognitive Impairment. Clin Neurophysiol. 132(10):2540–2550PubMedCrossRef Rahimiasl M, Moghadam Charkari N, Ghaderi F (2021) Random walks on B distributed resting-state functional connectivity to identify Alzheimer’s disease and Mild Cognitive Impairment. Clin Neurophysiol. 132(10):2540–2550PubMedCrossRef
37.
Zurück zum Zitat Ramzan F, Khan MUG, Rehmat A, Iqbal S, Saba T, Rehman A, Mehmood Z (2019) A Deep Learning Approach for Automated Diagnosis and Multi-Class Classification of Alzheimer’s Disease Stages Using Resting-State fMRI and Residual Neural Networks. J Med Syst 44(2):37PubMedCrossRef Ramzan F, Khan MUG, Rehmat A, Iqbal S, Saba T, Rehman A, Mehmood Z (2019) A Deep Learning Approach for Automated Diagnosis and Multi-Class Classification of Alzheimer’s Disease Stages Using Resting-State fMRI and Residual Neural Networks. J Med Syst 44(2):37PubMedCrossRef
38.
Zurück zum Zitat Gupta Y, Kim JI, Kim BC, Kwon GR (2020) Classification and Graphical Analysis of Alzheimer’s Disease and Its Prodromal Stage Using Multimodal Features From Structural Diffusion, and Functional Neuroimaging Data and the APOE Genotype. Front Aging Neurosci 12:238PubMedPubMedCentralCrossRef Gupta Y, Kim JI, Kim BC, Kwon GR (2020) Classification and Graphical Analysis of Alzheimer’s Disease and Its Prodromal Stage Using Multimodal Features From Structural Diffusion, and Functional Neuroimaging Data and the APOE Genotype. Front Aging Neurosci 12:238PubMedPubMedCentralCrossRef
39.
Zurück zum Zitat Son SJ, Kim J, Park H (2017) Structural and functional connectional fingerprints in mild cognitive impairment and Alzheimer’s disease patients. PLoS ONE 12(3):e0173426PubMedPubMedCentralCrossRef Son SJ, Kim J, Park H (2017) Structural and functional connectional fingerprints in mild cognitive impairment and Alzheimer’s disease patients. PLoS ONE 12(3):e0173426PubMedPubMedCentralCrossRef
40.
Zurück zum Zitat Khazaee A, Ebrahimzadeh A, Babajani-Feremi A (2017) Classification of patients with MCI and AD from healthy controls using directed graph measures of resting-state fMRI. Behav Brain Res 322(Pt B):339–350PubMedCrossRef Khazaee A, Ebrahimzadeh A, Babajani-Feremi A (2017) Classification of patients with MCI and AD from healthy controls using directed graph measures of resting-state fMRI. Behav Brain Res 322(Pt B):339–350PubMedCrossRef
41.
Zurück zum Zitat Khazaee A, Ebrahimzadeh A, Babajani-Feremi A (2016) Application of advanced machine learning methods on resting-state fMRI network for identification of mild cognitive impairment and Alzheimer’s disease. Brain Imaging Behav 10(3):799–817PubMedCrossRef Khazaee A, Ebrahimzadeh A, Babajani-Feremi A (2016) Application of advanced machine learning methods on resting-state fMRI network for identification of mild cognitive impairment and Alzheimer’s disease. Brain Imaging Behav 10(3):799–817PubMedCrossRef
42.
Zurück zum Zitat Long Z, Li J, Fan J, Li B, Du Y, Qiu S, Miao J, Chen J, Yin J, Jing B (2023) Identifying Alzheimer’s disease and mild cognitive impairment with atlas-based multi-modal metrics. Front Aging Neurosci 15:1212275PubMedPubMedCentralCrossRef Long Z, Li J, Fan J, Li B, Du Y, Qiu S, Miao J, Chen J, Yin J, Jing B (2023) Identifying Alzheimer’s disease and mild cognitive impairment with atlas-based multi-modal metrics. Front Aging Neurosci 15:1212275PubMedPubMedCentralCrossRef
43.
Zurück zum Zitat Sarraf S, Sarraf A, DeSouza DD, Anderson JAE, Kabia M, The Alzheimer’s Disease Neuroimaging (2023) OViTAD: Optimized Vision Transformer to Predict Various Stages of Alzheimer’s Disease Using Resting-State fMRI and Structural MRI Data. Brain Sci. 13(2):260PubMedPubMedCentralCrossRef Sarraf S, Sarraf A, DeSouza DD, Anderson JAE, Kabia M, The Alzheimer’s Disease Neuroimaging (2023) OViTAD: Optimized Vision Transformer to Predict Various Stages of Alzheimer’s Disease Using Resting-State fMRI and Structural MRI Data. Brain Sci. 13(2):260PubMedPubMedCentralCrossRef
45.
Zurück zum Zitat Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, Leeflang MM, Sterne JA, Bossuyt PM (2011) QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 155(8):529–536PubMedCrossRef Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, Leeflang MM, Sterne JA, Bossuyt PM (2011) QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 155(8):529–536PubMedCrossRef
46.
Zurück zum Zitat Herholz K, Salmon E, Perani D, Baron JC, Holthoff V, Frölich L, Schönknecht P, Ito K, Mielke R, Kalbe E, Zündorf G, Delbeuck X, Pelati O, Anchisi D, Fazio F, Kerrouche N, Desgranges B, Eustache F, Beuthien-Baumann B, Menzel C, Schröder J, Kato T, Arahata Y, Henze M, Heiss WD (2002) Discrimination between Alzheimer dementia and controls by automated analysis of multicenter FDG PET. Neuroimage 17(1):302–316PubMedCrossRef Herholz K, Salmon E, Perani D, Baron JC, Holthoff V, Frölich L, Schönknecht P, Ito K, Mielke R, Kalbe E, Zündorf G, Delbeuck X, Pelati O, Anchisi D, Fazio F, Kerrouche N, Desgranges B, Eustache F, Beuthien-Baumann B, Menzel C, Schröder J, Kato T, Arahata Y, Henze M, Heiss WD (2002) Discrimination between Alzheimer dementia and controls by automated analysis of multicenter FDG PET. Neuroimage 17(1):302–316PubMedCrossRef
47.
Zurück zum Zitat Shaffer JL, Petrella JR, Sheldon FC, Choudhury KR, Calhoun VD, Coleman RE, Doraiswamy PM (2013) Predicting cognitive decline in subjects at risk for Alzheimer disease by using combined cerebrospinal fluid MR imaging, and PET biomarkers. Radiology 266(2):583–591PubMedPubMedCentralCrossRef Shaffer JL, Petrella JR, Sheldon FC, Choudhury KR, Calhoun VD, Coleman RE, Doraiswamy PM (2013) Predicting cognitive decline in subjects at risk for Alzheimer disease by using combined cerebrospinal fluid MR imaging, and PET biomarkers. Radiology 266(2):583–591PubMedPubMedCentralCrossRef
48.
Zurück zum Zitat Chandra A, Valkimadi PE, Pagano G, Cousins O, Dervenoulas G, Politis M (2019) Applications of amyloid, tau, and neuroinflammation PET imaging to Alzheimer’s disease and mild cognitive impairment. Hum Brain Mapp 40(18):5424–5442PubMedPubMedCentralCrossRef Chandra A, Valkimadi PE, Pagano G, Cousins O, Dervenoulas G, Politis M (2019) Applications of amyloid, tau, and neuroinflammation PET imaging to Alzheimer’s disease and mild cognitive impairment. Hum Brain Mapp 40(18):5424–5442PubMedPubMedCentralCrossRef
49.
Zurück zum Zitat Jack CR Jr, Knopman DS, Jagust WJ, Petersen RC, Weiner MW, Aisen PS, Shaw LM, Vemuri P, Wiste HJ, Weigand SD, Lesnick TG, Pankratz VS, Donohue MC, Trojanowski JQ (2013) Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol 12(2):207–216PubMedPubMedCentralCrossRef Jack CR Jr, Knopman DS, Jagust WJ, Petersen RC, Weiner MW, Aisen PS, Shaw LM, Vemuri P, Wiste HJ, Weigand SD, Lesnick TG, Pankratz VS, Donohue MC, Trojanowski JQ (2013) Tracking pathophysiological processes in Alzheimer’s disease: an updated hypothetical model of dynamic biomarkers. Lancet Neurol 12(2):207–216PubMedPubMedCentralCrossRef
50.
Zurück zum Zitat Frizzell TO, Glashutter M, Liu CC, Zeng A, Pan D, Hajra SG, D’Arcy RCN, Song X (2022) Artificial intelligence in brain MRI analysis of Alzheimer’s disease over the past 12 years: A systematic review. Ageing Res Rev 77:101614PubMedCrossRef Frizzell TO, Glashutter M, Liu CC, Zeng A, Pan D, Hajra SG, D’Arcy RCN, Song X (2022) Artificial intelligence in brain MRI analysis of Alzheimer’s disease over the past 12 years: A systematic review. Ageing Res Rev 77:101614PubMedCrossRef
51.
Zurück zum Zitat He W, Xiong H, Fang J, Gu H (2021) Impact of self-designed Ningxin Anshen Decoction on the resting-state network functional connectivity in patients with mild to moderate generalized anxiety disorders. Ann Palliat Med 10(2):1313–1324PubMedCrossRef He W, Xiong H, Fang J, Gu H (2021) Impact of self-designed Ningxin Anshen Decoction on the resting-state network functional connectivity in patients with mild to moderate generalized anxiety disorders. Ann Palliat Med 10(2):1313–1324PubMedCrossRef
52.
Zurück zum Zitat Marchitelli R, Aiello M, Cachia A, Quarantelli M, Cavaliere C, Postiglione A, Tedeschi G, Montella P, Milan G, Salvatore M, Salvatore E, Baron JC, Pappatà S (2018) Simultaneous resting-state FDG-PET/fMRI in Alzheimer Disease: Relationship between glucose metabolism and intrinsic activity. Neuroimage 176:246–258PubMedCrossRef Marchitelli R, Aiello M, Cachia A, Quarantelli M, Cavaliere C, Postiglione A, Tedeschi G, Montella P, Milan G, Salvatore M, Salvatore E, Baron JC, Pappatà S (2018) Simultaneous resting-state FDG-PET/fMRI in Alzheimer Disease: Relationship between glucose metabolism and intrinsic activity. Neuroimage 176:246–258PubMedCrossRef
Metadaten
Titel
The application value of Rs-fMRI-based machine learning models for differentiating mild cognitive impairment from Alzheimer's disease: a systematic review and meta-analysis
verfasst von
Chentong Wang
Li Zhou
Feng Zhou
Tingting Fu
Publikationsdatum
03.09.2024
Verlag
Springer International Publishing
Erschienen in
Neurological Sciences / Ausgabe 1/2025
Print ISSN: 1590-1874
Elektronische ISSN: 1590-3478
DOI
https://doi.org/10.1007/s10072-024-07731-1

Kompaktes Leitlinien-Wissen Neurologie (Link öffnet in neuem Fenster)

Mit medbee Pocketcards schnell und sicher entscheiden.
Leitlinien-Wissen kostenlos und immer griffbereit auf ihrem Desktop, Handy oder Tablet.

Neu im Fachgebiet Neurologie

Schlaganfall oder Schlaganfall-Imitator?

Ein breites Spektrum von Erkrankungen kann einen Schlaganfall vortäuschen. Bei der notwendigen schnellen Unterscheidung zwischen solchen „stroke mimics“ und echten Schlaganfällen können einige klinische Faktoren und Symptome unterstützend herangezogen werden. 

Ein pulverisierter Plastiklöffel im Gehirn

Das menschliche Gehirn besteht zu etwa 0,5% aus Nanoplastik – Tendenz weiter steigend. Nach Resultaten eine Autopsiestudie reichert sich Plastik im Gehirn 10- bis 30-fach stärker an als in anderen Organen – mit bislang noch völlig unklaren Folgen.

Schlaganfall durch wandernde A. carotis interna

Die Thrombolyse nützte nichts, vielmehr war eine Teilresektion des Zungenbeins nötig: Es hatte eine verlagerte Carotis interna komprimiert und so bei einem älteren Mann einen Schlaganfall ausgelöst. Per Bildgebung lässt sich eine solche Ursache manchmal nur schwer nachweisen.

Epileptischer Anfall nach Darmreinigung

Die Darmreinigung vor einer Koloskopie führte einen älteren Mann auf die Intensivstation: Seine Natriumwerte waren durch die Prozedur so gesunken, dass er Bewusstseinstrübungen und Krampfanfälle entwickelte.

Update Neurologie

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