Liver Segmentation using Fast-Generalized Fuzzy C-Means (FG-FCM) from CT Scans
Yasmeen Al-Saeed, Hassan Soliman, Mohammed Mahfouz Elmogy · 2020 International Conference on Data Analytics for Business and Industry: Way Towards a Sustainable Economy (ICDABI) · 2020
Liver tumors are one of the leading deaths causes across the world. The early diagnosis of those tumors will be beneficial in reducing mortality rates. Liver computer-aided diagnosis (CAD) is used to assist radiologist decision. Accurate liver segmentation is an essential stage in CAD systems. This paper proposed an automatic, unsupervised liver segmentation based on Fast-Generalized Fuzzy C-Means (FG-FCM). This paper's primary concern is to accurately segment the liver from the rest of the abdomen organs on computed tomography (CT) scans. The proposed method was tested and evaluated using three benchmark datasets, MICCAI-Sliver07, LiTS17, and 3Dircadb, with 250 subjects. Experimental results show the proposed method's reliability with an accuracy of 94.70%, sensitivity of 93.81%, a specificity of 91.90%, a Dice similarity coefficient of 92.51%, and an area under the curve of 93.09%. The proposed method is advantageous in terms of noise robustness and in-homogeneity correction when compared with FCM.