Classification of Pulmonary Emphysema using Deep Learning

Emphysema is one of the lung diseases that comprise COPD. Emphysema is a long-term lung illness that results in alveolar destruction, which destroys tiny air sacs in the lung that allow fast gas exchange or exchange of oxygen and carbon dioxide molecules between the bloodstream and the alveoli. It is one of the two most common kinds of COPD. Another significant COPD consequence is chronic bronchitis. Heart failure collapsed lungs, and big holes in the lungs are all possible consequences of emphysema. Emphysema is an incurable lung airway condition. This research chooses chest X-rays rather than CT scans. This article aims to create a deep learning model that can identify emphysema patients from chest radiographic images. This paper chooses 1696 chest X-ray images from an NIH Chest X-ray dataset. ResN et50 is used as a deep learning model since it is proved effective in medical image classification. This model is trained using multiple epochs and mini-batch sizes. It achieved 98.82 percent accuracy, 1.0 value precision for emphysema, 0.9885 for AUC, 97.70 percent recall/sensitivity for emphysema, 100 percent specificity, 98.85 percent balanced accuracy, and an 18.3303 geometric mean. Additionally, we obtained a false-positive rate of 0%.