ABSTRACT
In computer vision, the role of segmentation is inevitable for the analysis of desired region of interest. Teleradiology refers to the transmission of medical images from one place to another through Internet for the detailed study by radiologist and physicians. Preprocessing, segmentation, and compression are the vital stages in teleradiology-based medical image analysis. The fuzzy clustering is an Artificial Intelligence technique and is prominent in medical image processing for the analysis of tumors and delineation of anatomical organs. This chapter proposes an improved Fuzzy C-Means based on adaptive regularized Gaussian kernel with the nonlinear tensor diffusion filter for the analysis of abdomen CT images. The proposed segmentation approach was compared with other clustering approaches and efficient results were produced. The less parameter tuning and improved accuracy makes it an attractive solution for region of interest extraction in medical images. The performance of proposed clustering algorithm was validated by cluster validity metrics on real-time medical datasets. The hardware implementation was carried out using Raspberry Pi embedded processor, and it is well suited for Internet of Things-based teleradiology application.
