Computer-aided diagnosis of cavernous malformations in brain MR images. (June 2018)

• Record Type: Journal Article
• Authors: Wang, Huiquan; Ahmed, S. Nizam; Mandal, Mrinal
• Is Part Of: Computerized medical imaging and graphics. Volume 66(2018)
• Page Start: 115
• Page End: 123
• Publication Date: 2018-06
• Subjects: Cavernous malformation -- Computer-aided diagnosis -- Skull stripping -- Template matching -- Principal component analysis -- Support vector machine; Diagnostic imaging -- Periodicals; Imaging systems in medicine -- Periodicals; Diagnosis, Radioscopic -- Data processing -- Periodicals; Diagnostic Imaging -- Periodicals; Imagerie pour le diagnostic -- Périodiques; Diagnostic imaging; Periodicals; Electronic journals; Electronic journals; 616.0754
• DOI: 10.1016/j.compmedimag.2018.03.004
• Languages: English
• ISSNs: 0895-6111

Abstract

Highlights: Cavernous malformation is one of the most common epileptogenic lesions. An efficient automated detection of Cavernoma in MRI image is proposed. Three steps: brain extraction, candidate caveronoma detection, final classification. 95% sensitivity, 90% specificity and 91% accuracy are achieved. The technique has low computational complexity. Abstract: Cavernous malformation or cavernoma is one of the most common epileptogenic lesions. It is a type of brain vessel abnormality that can cause serious symptoms such as seizures, intracerebral hemorrhage, and various neurological disorders. Manual detection of cavernomas by physicians in a large set of brain MRI slices is a time-consuming and labor-intensive task and often delays diagnosis. In this paper, we propose a computer-aided diagnosis (CAD) system for cavernomas based on T2-weighted axial plane MRI image analysis. The proposed technique first extracts the brain area based on atlas registration and active contour model, and then performs template matching to obtain candidate cavernoma regions. Texture, the histogram of oriented gradients and local binary pattern features of each candidate region are calculated, and principal component analysis is applied to reduce the feature dimensionality. Support vector machines (SVMs) are finally used to classify each region into cavernoma or non-cavernoma so that most of the false positives (obtained by template matching) are eliminated. The performance of the proposed CAD system is evaluated and experimental results show that it provides superior performance in cavernoma detection compared to existing techniques.