Feature Extraction in Liver Cancer Based on Abdominal CT Scan Images using Contour Analysis and GLCM Method

Abstract

Liver cancer is a type of cancer that affects the largest organs of the stomach, where some are grown from the liver and some grow in other organs, then spread to the liver. One of the technologies used to analyze and diagnose liver cancer is CT Scan (Computer Tomography Scanner). The CT Scan is often preferred for diagnosing liver cancer, especially as being considered of high accurate imaging, high imaging speed and relatively lower cost. However, the results of the CT Scan are often different depending on the accuracy and experience of the doctor so that it can lead to different diagnoses. In this study, a system was created that was able to extract features from CT Scan images of liver cancer to recognize the object of cancer and distinguish it from other objects. This system will be tested on 50 data abdominal CT Scan images with a diagnosis of liver cancer, where 21 data for benign liver cancer and 29 data for malignant liver cancer. This research has three main stages, that is preprocessing to improve image quality using scaling image, histogram equalization, and median filtering. Segmentation to identify the object being observed and separate it from the background using watershed method and binary thresholding with accuracy is 90%. The last is feature extraction based on cancer area, edge irregularity, and texture to identify liver cancer.