Nonlocal spatial clustering in automated brain hematoma and edema segmentation. (21st January 2019)

• Record Type: Journal Article
• Authors: Tu, Wei; Kong, Linglong; Karunamuni, Rohana; Butcher, Ken; Zheng, Lili; McCourt, Rebecca
• Is Part Of: Applied stochastic models in business and industry. Volume 35:Number 2(2019)
• Page Start: 321
• Page End: 329
• Publication Date: 2019-01-21
• Subjects: active contour without edges -- computed tomography -- medical image segmentation -- nonlocal spatial clustering; Stochastic analysis -- Periodicals; Stochastic processes -- Periodicals; Business mathematics -- Periodicals; Finance -- Mathematical models -- Periodicals; Industrial management -- Mathematical models -- Periodicals; 338.00151923
• DOI: 10.1002/asmb.2431
• Languages: English
• ISSNs: 1524-1904

Abstract

Abstract: Hematoma and edema volume are potential predictors of 30‐day mortality rate and functional outcome (degree of disability or dependence in daily activities after a stroke) for patients with intracerebral hemorrhage. The manual segmentation of hematoma and edema from computed tomography scans is common practice but a time‐consuming and labor‐intensive task. Automated segmentation is an appealing alternative, but it is challenging because of the poorly defined boundary between edema and the surrounding healthy brain tissue. There is limited literature on this problem, and we aim to help fill the gap between the theoretical development of segmentation methods and the practical need. Our framework is fully automated and requires no supervision. The method uses nonlocal regularized spatial fuzzy C‐means clustering in the initialization stage and the active contour without edges method in the refinement stage. To evaluate it, we used 30 subjects with different sizes, shapes, and locations of hematoma and edema. Compared with the manual segmentation results from two independent raters, our method performs hematoma segmentation well, with an average dice score coefficient of 0.92. Although there is a lack of ground truth in edema segmentation due to the high inter and intrarater variation, our results are comparable with manual segmentation results.