Experimental Analysis for Non-Alcoholic Fatty Liver Disease in Ultrasound Images Based on Efficient Net Classifier

Liver dysfunction is most commonly brought on by fatty liver disease, which is widespread in both developed and developing countries. If not addressed, fatty liver might lead to serious complications. As juvenile obesity rates rise, so does the number of children diagnosed with nonalcoholic fatty liver disease (NAFLD). In order to distinguish nonalcoholic fatty liver disease (NAFLD) from healthy liver tissue, we wanted to perform digital image analysis (DIA) on ultrasound (US) pictures of the liver and create a machine learning (ML) based categorization model. This study compares EfficientNet Classifier and K-Nearest Neighbors (KNN) for its ability to detect the presence of nonalcoholic fatty liver disease in ultrasound images. The histological outcome of a liver biopsy may have been consulted as a reference for liver fat, as well as ultrasound scans of persons who went through laparoscopic surgery were additionally utilized as a reference for liver fat. EfficientNet Classifier and K-Nearest Neighbors were then used to classify the photos into one of two categories. For EfficientNet categorization, an artificial model for predicting liver fat from ultrasound images achieved an accuracy of 87%. However, when using K-Nearest Neighbors for categorization, 85% accuracies were achieved. Automatic detection of liver fat utilizing the suggested technique based on textural characteristics utilizing the GLCM and the EfficientNet categorization from ultrasound pictures can aid physicians and radiologists in making a definitive diagnosis.