Publications

Abstract : In recent years, chest X-rays have emerged as a cost-effective means of diagnosing lung abnormalities. Physicians and radiologists often utilize these images to diagnose a wide range of pulmonary diseases. However, in cases where a diagnosis is uncertain, X-rays must be sent to medical experts for examination, which can be time-consuming. To address this issue, computer-aided diagnosis has emerged as a promising solution that reduces the potential for human errors and enables faster and more accurate diagnosis. This study presents a comparative analysis of various deep learning architectures, demonstrating that selective kernel networks (SkNets) consistently outperform other state-of-the-art models in accuracy, precision, recall, and F1-score. Specifically, SkNets consistently achieve higher accuracy than comparable ResNet models, with our proposed model achieving an impressive accuracy of 95%, a recall of 95.3%, precision of 94.8%, and an F1-score of 95%. The findings in this study highlight the potential of SkNets to serve as a reliable and effective tool for accurately diagnosing pulmonary diseases, with significant implications for the medical community.