We present a noninvasive method for the detection and an advanced segmentation of Hepatocellular carcinoma (HCC) based on Computed Tomography (CT) images.This proposed method basically starts with the processing of the data set. 60 CT images are prepared for the segmentation process. Image data is divided into two groups; 50 CT images of the HCC, and 10 CT images of the normal liver. The ground truth images are created with the specialist abdominal radiologist. Images are in 256x256 µm size in JPEG format. For the segmentation part, the Statistical Region Merging method is used. The proposed method consists of three main parts, these are thresholding, segmentation, and estimation of ROC parameters. By using the database and the ground truth, according to the simulation results, the average of the sensitivity, specificity, and accuracy are obtained as 0.7476 %, 0.9723 %, and 0.9502 %, respectively. In conclusion, HCC is the most common primary malignant tumor in the liver. It is considered an important and life-threatening disease. Early detection of liver cancer has become very important for the patients. The Region Merging Segmentation Method is a very useful liver segmentation technique for detection of the HCC.
Key words: Liver cancer, diagnosis, computed tomography, statistical region merging method
|
