A robust chronic obstructive pulmonary disease classification model using dragonfly optimized kernel extreme learning machine

Chronic obstructive pulmonary disease (COPD) is considered to be one of the most commonly occurring respiratory disorders and is proliferating at an extremely high rate in the recent years. The proposed system aims to classify the various stages of COPD using a COPD patient dataset comprising 101 patients and 24 varied factors related to the disease. In addition, a self-acquired dataset containing 560 lung CT images was also used. The obtained hybrid database is normalized, augmented, and preprocessed using bilateral filter and contrast enhanced using dynamic histogram equalization. Segmentation is then performed using SuperCut algorithm. Feature extraction is done by binary feature fusion technique involving UNet and AlexNet. Kernel extreme learning machine-based classification is conducted further, and the results produced are optimized using dragon fly optimization algorithm. The proposed system produced an enhanced accuracy of 98.82%, precision of 99.01%, recall of 94.98%, F1 score of 96.11%, specificity of 98.09%, MCC value of 94.33%, and AUC value of 0.996 which are far better when compared with other existing systems.