Segmentation of Lung Regions from Chest X-Rays Using AI-Based Techniques

History has vividly demonstrated the rapid and devastating impact of the COVID-19 pandemic on a global scale, leading to significant health losses and profoundly affecting the lives of billions worldwide. To combat its spread, early detection has emerged as a primary strategy. A crucial opportunity to enhance patient care and healthcare strategies against COVID-19 lies in the ability to detect lung infections and delineate their borders through medical imaging. The prompt identification of this illness can be achieved most efficiently through the examination of chest radiographs and CT images. Chest X-rays are very easy, cost-effective, and primary imaging modalities used by medical practitioners Therefore, it becomes imperative to segment chest X-rays of the lungs accurately to aid in the accurate and timely diagnosis of the disease so that it can be effectively managed. In this work, the performance of the various artificial intelligence (AI)-based techniques based on the U-Net model and transfer learning is analyzed for the segmentation of chest X-rays and results are evaluated with regard to different parameters, such as accuracy, F1-score, JSC etc.