Axis-guided patch based accurate segmentation for pathological vessels using adaptive weight sparse representation. (March 2020)

• Record Type: Journal Article
• Authors: Hu, Xin; Ding, Deqiong; Li, Zhengzuo; Ge, Quanxu; Jiang, Chunmao; Li, Jing; Zhou, Zhiyuan; Chu, Dianhui
• Is Part Of: Biomedical signal processing and control. Volume 57(2020)
• Publication Date: 2020-03
• Subjects: Patch-based vessel segmentation -- Adaptive weight sparse representation -- Multi-scale training samples -- Pathological vessel; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2019.101817
• Languages: English
• ISSNs: 1746-8094

Abstract

Abstract : Graphical abstract: Addressing the challenges in delineating the pathological vessel boundary accurately, a novel segmentation framework is presented, a vessel axis tracking algorithm + a patch-based sparse representation. The redundant information embedded in the training samples is employed by the patch-based sparse representation method, to deal with the challenges caused by pathological tissues. However, the dense one-to-one comparison of the sparse representation method between target patch and training patch results in heavy computational burden. In order to gain computational efficiency, vessel axis guide segmentation method, multi-scale training samples, multi-scale dictionary, and adaptive weights for scale-specific sub-dictionary atoms are used in this work. The evaluation of our method is promising, on both segmentation accuracy and computational time. Highlights: A pathological vessel segmentation combines axis tracking and sparse representation. Redundant information is embedded in the multi-scale training for accurate segment. The computational burden is reduced by a multi-scale dictionary and adaptive weights. The performance of our method is promising over clinical datasets. Abstract: Background and objective: Over decades of research and development in vessel segmentation, accurate and reliable methods targeting at pathological vessels are few. Addressing this challenge, we try to delineate the vessel boundary accurately with the presence of pathologies. Methods: A novel segmentation framework is presented in this work, a vessel axis tracking algorithm + a patch-based sparse representation. The patch-based algorithm is navigated by the vessel axis tracking algorithm. Within the training process, multi-scale training samples have been used, which has the potential to be both physiologically accurate and computationally effective. The redundant information embedded in the multi-scale training samples is employed to delineate the pathological vessel accurately. To further reduce the computational burden caused by the patch-based sparse representation, a multi-scale dictionary has been generated, and adaptive weights have been assigned to the scale-specific sub-dictionary atoms, for computational efficiency. Results: Our method is evaluated by comparing with two state-of-the-art methods, on synthetic complex-structured datasets and real clinical datasets. The performance of our method is promising over the evaluation, since the overlap ratios of our method are high over all the datasets, around 91%, much better than two state-of-the-art methods.