Segmentation of Anomalies in Abdomen CT Images by Convolution Neural Network and Classification by Fuzzy Support Vector Machine

The application of computer-aided algorithms for disease diagnosis and treatment is gaining prominent in the past years, and the role of machine learning algorithms is inevitable. This chapter focuses on the segmentation of liver and anomalies like tumor and cyst from abdomen CT images using deep learning convolution neural network (DLCNN) and classification of tumor stages by fuzzy support vector machine (FSVM). The segmentation result of DLCNN outperforms the backpropagation neural network, group method data handling neural network, and decision tree algorithm. The FSVM-based tumor classification results were superior when compared with classical SVM. This chapter focuses on the following: (i) machine learning algorithms for classification and segmentation of medical images, (ii) role of DLCNN in medical image segmentation, (iii) role of FSVM in anomalies classification, and (iv) validation of segmentation and classification results by performance metrics. The simulation results are generated in Matlab 2015a and Java and validated on real-time abdomen CT images.