Diagnosis Of Interstitial Lung Disease By Pattern Classification

Diagnosis of interstitial lung disease (ILD) using pattern classification consists of image preprocessing, feature extraction, selection and classification. Feature extraction is initially done using textons and then LTCOP method is used. Classification is initially done using ANN, KNN and Deep CNN classifiers. Deep CNN produces greater accuracy than ANN and KNN classifiers. Feature selection is initially done using ReLu activation and then histogram method is used. Hybrid kernel based SVM classification is a new method that produces more accuracy compared to ANN, KNN and Deep CNN classifiers. Performance of classification are determined using confusion matrix, recall rate, precision, Faverage and accuracy.