Novel Hybrid Feature Extraction for Enhanced Liver Abnormality Detection using Bio-Inspired Classifiers

The early and precise identification of liver abnormality is essential for better patient care and enhanced prognosis. Conventional diagnostic approaches, including biopsies and imaging techniques, may be invasive, protracted, and necessitate specialist analysis. This research introduces an innovative hybrid feature extraction method integrated with bio-inspired classifiers to improve the classification of liver disorders. The method employs ultrasonic images, to detect cirrhotic patterns and categorize liver abnormalities. In this work the hybrid Fast Hadamard EHO techniques is used for feature extraction. The five classifiers such as Gaussian Mixture Model (GMM), DFA, Firefly, Harmonic Search and hybrid algorithm such as Firefly GMM and Harmonic search GMM were used for classification of liver abnormality. The Hybrid classifier Harmonic GMM gives prominent accuracy of 96.43% and a minimum error rate of 3.57%.