Characterization of Hepatocelluler Carcinoma in CT images Using Texture Classification Method

Abstract

The objectives of the study were to characterize Hepatocellular carcinoma (HCC) in CT images using texture analysis so far for classification and delineation of the HCC and liver, in order to have an objective method to improve the accuracy of the diagnosis and to reduce the number of required invasive procedures . The study was carried out on 200 patient with Hepatocellular (HCC) carcinoma underwent abdominal CT Images, from both gender. The study was conducted from August 2012 to August 2015 in Kingdom of Saudi Arabia- Southern area –Najran City ( King Khalid Hospital) and Khartoum state ( Modern medical center ) with Helical Multi detector CT scanner Siemens machines . The classification processes were carried out using Interactive Data Language (IDL) program. After all images were classified the data entered into SPSS with its classes to generate a classification score using stepwise linear discriminate analysis to select the most discriminate features that can be used in the classification of abdominal tissues in CT images. Then the delineation of HCC done by furthers processing of the classification using region label function. Similar method was obtained for the same set using grey scale as input instead of the texture. The result shown that the texture analysis reveal a different underlying pattern of the HCC compared to the normal liver and other abdominal tissues with classification sensitivity 96.5%, and the combination of the texture features throughout the different triphasic image phases provides the highest predictive overall accuracy of 89.1 % using stepwise linear discriminant analysis .