Automated Hemorrhage Slices Detection for CT Brain Images

• Published in: Visual Informatics: Sustaining Research and Innovations pp. 268 - 279
• Main Authors: Tong, Hau-Lee, Ahmad Fauzi, Mohammad Faizal, Haw, Su-Cheng
• Format: Book Chapter
• Language: English
• Published: Berlin, Heidelberg Springer Berlin Heidelberg 2011
• Series: Lecture Notes in Computer Science
• Subjects: Computed tomography; Hemorrhage slices detection; Visual Informatics
• ISBN: 3642251900; 9783642251900
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-642-25191-7_26

This paper presents an automated method to detect the hemorrhage slices for Computed Tomography(CT) brain images. The proposed system can be divided into two stages which are preprocessing stage and detection stage. Preprocessing basically is to prepare and enhance the images for the detection stage. During detection stage, new midline detection approach is proposed to divide the intracranial area into left and right hemispheres. Then histogram features are extracted from left and right hemispheres for the dissimilarity comparison. All the feature components will be channeled into support vector machine (SVM) classifier to determine the existence of the hemorrhage. Ten-fold cross validation was applied during the SVM classification. The experiments were performed on 450 CT images and results were evaluated in terms of recall and precision. The recall and precision obtained from experimental results for hemorrhage slices detection are 84.86% and 96.82% respectively.