Rough Classifier Based on Region Growth Algorithm for Identifying Liver CT Image
Ching‐Hsue Cheng, Liang‐Ying Wei · Tamkang University Institutional Repository (TKUIR) · 2016
Over decades, liver cancer is a rising cause of death in Taiwan, and more and more researchers are concerned about detecting hepatic tumors in computed tomography (CT) images. For clinical applications in terms of diagnosis and treatment planning, image segmentation on abdominal CT is indispensable. Patients with a large number of CT images need specialist physicians to identify, and detecting tumor location correctly from many CT images has been a major challenge subsequently. Therefore, this paper proposed a novel computer-aided detection (CAD) method that had high classification accuracy for identifying tumors. The proposed method used a region growing algorithm to segment liver CT images, employed REDUCT sets to reduce attributes, and then utilized a rough set algorithm to enhance classification performance. To evaluate the classification performances, the proposed method was compared with five different classification methods: decision tree (C4.5 and REP (reduced error pruning)), multilayer perceptron, Naïve Bayes, and support vector machine (SVM). The results indicate that the proposed method is superior to the listing methods in terms of classification accuracy.