Abstract
Magnetic resonance imaging (MRI) has been a prevalence technique for breast cancer diagnosis. Computer-aided detection and segmentation of lesions from MRIs plays a vital role for the MRI-based disease analysis. There are two main issues of the existing breast lesion segmentation techniques: requiring manual delineation of Regions of Interests (ROIs) as a step of initialization; and requiring a large amount of labeled images for model construction or parameter learning, while in real clinical or experimental settings, it is highly challenging to get sufficient labeled MRIs. To resolve these issues, this work proposes a semi-supervised method for breast tumor segmentation based on super voxel strategies. After image segmentation with advanced cluster techniques, we take a supervised learning step to classify the tumor and nontumor patches in order to automatically locate the tumor regions in an MRI. To obtain the optimal performance of tumor extraction, we take extensive experiments to learn parameters for tumor segmentation and classification, and design 225 classifiers corresponding to different parameter settings. We call the proposed method as Semi-supervised Tumor Segmentation (SSTS), and apply it to both mass and nonmass lesions. Experimental results show better performance of SSTS compared with five state-of-the-art methods.
Highlights
Image segmentation is a process of separating a digital image into di®erent regions for specic purposes, such as object recognition and classication.[1]
There are three parameters that need to be set by users in SSTS, which are normalization threshold m for Simple Linear Iterative Clustering (SLIC) over-segmentation, distance threshold d for clustering super-pixels to form patches based on DBSCAN, and r for patch labeling
We can see that the extraction results of SSTS on Magnetic resonance imaging (MRI) of case 1 are similar as the result of MT, which is much better than the result of Fuzzy C means (FCM)
Summary
Image segmentation is a process of separating a digital image into di®erent regions for specic purposes, such as object recognition and classication.[1]. The main weaknesses of the existing breast lesion segmentation include the following aspects: (1) some methods require manual delineation of ROIs as a step of initialization (e.g., Ref. 9), which restricts the automatic segmentation to cases that su±cient expert knowledge should be known in advance[6]; (2) supervised methods require a large amount of labeled images for model construction or parameter learning, while in real clinical or experimental settings, it is highly challenging to get su±cient labeled MRIs due to the limited number of patients and the time constraint; (3) existing work mainly focuses on segmenting mass tumors (either benign or malignant) (e.g., Ref. 5), the investigation on segmenting nonmass lesions is relatively less due to the shape diversity of the lesions, which makes the delineation of the lesions extremely di±cult To tackle these issues, we propose a Semisupervised Tumor Segmentation (SSTS) technique, which constraints the number of threshold parameters that need to be set in advance, and requires a small amount of labeled images. This paper is structured as follows: Sec. 2 illustrates the related work of tumor segmentation; Sec. 3 introduces the proposed SSTS framework in detail; Sec. 4 discusses the experimental settings and results and Sec. 5 concludes this paper and indicates our future work
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
