Computer-Aided tumor diagnosis in 3-D breast elastography
Introduction
Breast cancer is the second leading cause of cancer-related mortality in women [1]. Undoubtedly, the early detection of breast cancer is the most effective way to reduce the death rate [2]. In recent years, since ultrasound has higher sensitivity than mammography in early detection of breast cancer [3] and it is non-invasive, low cost, real time, ultrasound has become one of the most useful modality for detecting and diagnosing the breast tumors. In addition, the developed elastography techniques are mature enough to provide useful information for clinical diagnosis recently. The stiffness differences between benign and malignant tumors may be used to be a basis for classifying tumors in elastography due to the reason that the malignant tumors are usually stiffer than a benign one [4].
The three-dimensional (3-D) US [5] has been developed for clinical assessment of diseases. 3-D US has the real time capability to build the lesion or organ volume by using the computer workstations. The main advantages of 3-D US over 2-D US [6] are to provide a repeatable and more precise method for evaluating disease entities and anatomic structures and to view the cross sectional planes which are usually inaccessible in 2-D US. Recently, research investigators and commercial companies have further developed 3-D elastography to obtain all the elasticity information of a tumor volume. However, elastogram cannot be used for clinical diagnosis without combining other modalities such as 2-D, 3-D or 4-D US [7].
Since the rapid development of hardware and computer applications, many computer-aided diagnosis (CAD) systems are developed for saving physicians’ workloads. Our previous CAD studies are using the 2-D [8], [9] and 3-D US [10], [11], and 2-D elastography [12], [13] to diagnose the tumors. The elastographic features computed from a 3-D elastographic volume rather than using a single elastographic image were utilized to differentiate benign from malignant lesion [12]. It is also proved that the combination of B-mode and elastographic features sufficiently improves the diagnostic performances of 2-D elastography [13]. Similarly, combining 3-D B-mode [10] and 3-D elastographic features may raise the diagnostic rate. Therefore, in this study, a CAD method is proposed based on 3-D morphological and elastographic features extracted from surrounding tumor tissues for classification of benign and malignant breast tumors.
Section snippets
Material
This study was approved by the local ethics committee, and informed consent was obtained from all of the included patients. The data used in this study consisted of 40 biopsy-proved lesions (20 benign and 20 malignant lesions) in 35 women (age range from 23 to 83 years, mean 55.93 ± 14.64 years), which were collected between December 2011 to August 2012 to evaluate the performance of the diagnosis system proposed in this study. In benign lesions, there were 11 cases of fibroadenoma (FA), 5
Method
In the proposed diagnostic scheme, the volume of interest (VOI) containing the tumor is obtained by overlapping the B-mode images and elastographic images, and extracting the intersectional area. Then, the proposed segmentation method is used to segment the tumor within the VOI. After obtaining the tumor contour, the B-mode features including the morphology and texture features are computed based on the tumor contour and the B-mode ultrasound images. The elastographic features which contain
Experimental results
All the cases are classified into benign and malignant tumors by applying the binary logistic regression model with the leave-one-out cross-validation method. The four kinds of proposed feature set and all the combinations of these feature sets are used to classify tumors respectively. The five indicators, which include accuracy, sensitivity, specificity, PPV, NPV, and the Az values of the ROC curve are calculated and listed in Table 2. The classification by using the feature set contains
Discussion
Although the mammography is the most used modality for screening breast cancer, it may yield false-positive results while applying in the dense breasts [26]. In recent, US is utilized as the complementary modality for screening because of its higher sensitivity in dense breast cases, but lower specificity in cancer detection. Breast ultrasound has been proved that it is useful for differentiating malignant from benign masses. Nevertheless, the morphological features of malignant and benign
Conclusion
In this study, the purpose is proposed for diagnosing the breast tumors on the 3-D elastography by using the B-mode and the elastographic feature sets. Firstly, the proposed segmentation methods are used to extract the tumor contour and mask for the B-mode images. Then, to analyze the tumor lesions, two feature sets containing B-mode and elastographic features are extracted by using the tumor contour and mask. Finally, the B-mode, elastography, and combined feature sets are used to classify the
Conflict of interest
The authors declare that they have no financial and personal relationships with other people or organizations that could inappropriately influence their work.
Acknowledgement
The authors thank the Ministry of Science and Technology (MOST 104-2221-E-002-062-MY3) of the Republic of China for the financial support.
References (32)
- et al.
Computer aided classification system for breast ultrasound based on breast imaging reporting and data system (BI-RADS)
Ultrasound Med. Biol.
(2007) - et al.
Breast ultrasound computer-aided diagnosis using BI-RADS features
Acad. Radiol.
(2007) - et al.
Computer-aided diagnosis for the classification of breast masses in automated whole breast ultrasound images
Ultrasound Med. Biol.
(2011) An easy measure of compactness for 2D and 3D shapes
Pattern Recognit.
(2008)- et al.
Comparison of strain and shear wave elastography for the differentiation of benign from malignant breast lesions, combined with B-mode ultrasonography: qualitative and quantitative assessments
Ultrasound Med. Biol.
(2014) Cancer facts and figures 2016
Am. Cancer Soc.
(2016)- C.D. Mathers, D.M. Fat, J. Boerma, The global burden of disease: 2004 update, World Health Organization...
- et al.
Benefit of ultrasonography in the detection of clinically and mammographically occult breast cancer
World J. Surgery
(2008) - et al.
New elastographic classification of breast lesions during and after compression
Diagn. Interv. Radiol.
(2009) - et al.
Downey, Pretorius, Fenster, Three-dimensional ultrasound
(1999)
Clinical utility of three-dimensional US1
Radiographics
Breast sonography–2D, 3D, 4D ultrasound or elastography
Breast Care
Computer-aided diagnosis for 3-D power doppler breast ultrasound
Ultrasound Med. Biol.
Breast elastography diagnosis based on dynamic sequence features
Med. Phys.
Classification of breast tumors using elastographic and b-mode features: comparison of automatic selection of representative slice and physician-selected slice of images
Ultrasound Med. Biol.
Advances In Diagnostic And Therapeutic Ultrasound Imaging
Cited by (8)
A wavelet neural operator based elastography for localization and quantification of tumors
2023, Computer Methods and Programs in BiomedicineViscoelastic parameters of invasive breast cancer in correlation with porous structure and elemental analysis data
2021, Computer Methods and Programs in BiomedicineCitation Excerpt :There are several modalities for non-invasive measurement of mechanical properties of tissues. Elastography is a method in which a stimulus is imposed onto the tissue, and the tissue response is studied using imaging techniques [5–8]. Mechanical imaging or tactile imaging is a branch of elastography used for the medical diagnosis based on the analysis of stress patterns [9].
Methods for the segmentation and classification of breast ultrasound images: a review
2021, Journal of UltrasoundBreast elastography: Diagnostic performance of computer-aided diagnosis software and interobserver agreement
2020, Radiologia Brasileira