An Efficient and Automated Classification Scheme for Diagnosing Fatty Liver Disorder using Ultrasonic Images

Abstract

In the treatment of abdominal diseases like fatty liver disorder, ultrasonic images-based investigation is considered as the primary step of diagnosis. But, the noisy feature of ultrasonic images combined with the least contrasting features introduces maximum complexity during the process of automated classification. This paper contributes an Improved Active Contour Segmentation scheme for effective segmentation. Then Gray Level Co-occurrence Matrix (GLCM) and fractal features are extracted from the segmented ultrasonic images and the classification is achieved using an enhanced forest classification incorporated in the artificial neural networks (ANN) for accurate detection. The results of the proposed automated approach are investigated using classification accuracy, mean classification accuracy, true positive rate, false positive rate and true negative rate. The results of the proposed scheme also infers a classification accuracy rate of 98.73% and a mean classification accuracy rate of 97.65% compared to the baseline automated liver disorder classification techniques.