A System to Classify Chronic Obstructive Pulmonary Disease using Pre-trained-Densenet201 with TSA

Abstract

COPD (Chronic obstructive pulmonary disease), ranking as the 3rd most common cause of death worldwide, frequently goes undiagnosed. Yet, the detection of COPD in its early stages is challenging due to the limited presence or mild nature of initial symptoms. In this work, the DL (deep learning) model DenseNet201 is utilized for classifying COPD using the PFT (Pulmonary Function Test) images. Initially, the pre-processing is carried out using the MF (median filter). After the noise elimination process, automated feature extraction and classification is carried out using the Pre-trained-DenseNet201 with TSA (tunicate search algorithm). The presented model provided satisfactory outcomes, attaining the accuracy of 0.985 and an AUC value of 98.73. These results surpass those reported in prior studies utilizing the similar database. Furthermore, the presented approach exhibits superior performance compared to various contemporary methods trained concurrently. This study represents the inaugural application of the Pre-trained-DenseNet201-with TSA model to this specific dataset for the purpose of COPD identification.