A Hybrid Model using CNN and Genetic Algorithm for Pattern Classification of Interstitial Lung Disease

Our objective is to develop an advanced computational approach to classify patterns of Interstitial Lung Disease(ILD) using a Hybrid model of Convolutional Neural Network(CNN) and Genetic Algorithm (GA). ILD encompasses a group of lung disorders that are often difficult to diagnose due to the complex and overlapping patterns observed in medical imaging. By leveraging the optimization capabilities of GAs, this work seeks to enhance the accuracy of ILD pattern classification. The approach finds the best hyperparameter to train the CNN classifier which identify the most essential nine classes of ILD patterns such as consolidation, fibrosis, emphysema, micronodules, and ground-glass opacities(GGO) and healthy. Here we have considered learning rate, dropout rate and filters as hyper parameters. Our experimental results show outstanding performance, attaining an accuracy of 98.54% and an F1 score of 98.55%. This model is expected to aid healthcare professionals in making more accurate and timely diagnoses of ILD, ultimately improving patient outcomes.