Lung disease classification using reticular pattern scoring and five class features with greedy stepwise based on GLCM

Authors:
Publisher website
Google Scholar
Abstract
Medical imaging is an important area in today's healthcare system. Diagnosis using medical images relies heavily on a clinician's ability and experience. However this reliance can offer a room for diagnosis inaccuracies when clinicians are faced with long hours and heavy workload. Therefore the role of Computer Aided Diagnosis (CADx) systems to aid clinicians to make faster and more accurate decisions is an important and growing one. Classification is an important stage in CADx. The objective of this study is to classify disease in the lung using a new five-class system. This study proposed a five-class system describe local features of Gray-level Co-occurrence Matrix (GLCM). This increase the features used to 20 features for each image. This study used 10 normal and 10 diseased patients. Six classifiers were used to show a diverse and non-bias performance of the features. Greedy stepwise algorithm was used as a feature selection method. With feature selection, the accuracies were above 0.80, which indicates the strength of the five-class feature. The highest accuracy was 0.89 using Random Forest classifier.

This publication has 11 references indexed in Scilit: