LAUSR

A progressive fatal disease causing more threats to human lives is chronic hepatic disease (CHD). In most of the developing countries, the mortality and morbidity rate have increased due to CHD. Invasive and noninvasive methods are used to measure the pathogenicity of the liver. In this study, ultrasonographic images, clinical findings, and laboratory findings are used to determine the stages of CHD. The stages of CHD are (a) chronic hepatitis, (b) compensated cirrhosis, and (c) decompensate cirrhosis. The histopathological analysis is adopted by the invasive method to conduct a liver biopsy. Results of liver biopsy have shown some kind of complications such as pain after liver biopsy, pneumothorax, bleeding, or puncture of the biliary tree and rarely death due to heavy bleeding. In such situations, noninvasive procedures are used as an alternative for liver biopsy. In this study, Hough‐based histogram‐oriented gradient features are extracted and classified using combined multi‐support vector machine and hidden Markov model classifiers. A large number of feature models containing 42 chronic hepatitis, 49 compensated cirrhosis, and 47 decompensate cirrhosis were selected specifically for the experimental study. The results outperformed using the abovementioned features and have achieved an overall accuracy of about 99% for the normal detector, 91.43% for the chronic hepatitis detector, and 96.72% for the cirrhosis detector.