Automatic Classification Algorithm for Diffused Liver Diseases Based on Ultrasound Images

Diffuse liver diseases such as fatty and cirrhosis, are leading causes of disability fatality across the world. Early diagnosis these is extremely important to save lives improve effectiveness treatment. This study proposes a non-invasive method for diagnosing using ultrasound images, by classifying tissue normal, steatosis, or feature extraction, selection, classification. First, correlation, homogeneity, variance, entropy, contrast, energy, long run emphasis, percentage, standard deviation determined. Second, most efficient features selected based on Fisher discriminant manual selection methods. Third, three voting-based sub-classifiers used, namely, normal/steatosis, normal/cirrhosis, steatosis/cirrhosis classifiers. The final classification majority function. Our provides two key contributions: combination different methods, avoiding limitations each while benefiting from their strengths; classifier categorization into sub-classifiers, where overall decision individual sub-classifier. We obtained recognition accuracies classifiers 95%, 95.74%, 94.23%, respectively, an accuracy which outperforms other