An Artificial Intelligence-Driven Deep Learning Model for Chest X-ray Image Segmentation

Abstract

Artificial Intelligence and CAD systems are becoming highly popular in medical diagnosis. The application of AI and deep learning in radiology is revolutionizing the medical industry with fast and accurate diagnosis. The Chest X-ray is one of the most significant radiological diagnostic methods being used for its easy availability, cost effectivity, and low radiation doses. The application of deep learning methods in chest X-rays has shown tremendous success in lesion detection. However, the chest-X ray contains a large non-region of interest in the form of the background that interrupts the AI system for accurate lesion detection. Towards the motive of solving the problem, this work proposes a robust and accurate deep learning-based UNet segmentation model to segment the region of interest, i.e., the lung region, and remove the background present in the X-ray images. Our model can successfully and accurately segment chest X-ray images. The model performed with an accuracy of 96.35% with dice coefficient and Jaccard index of 94.88% and 90.38%, respectively. Performing with high accuracy, dice, and Jaccard, our system proves its efficacy and robustness for efficiently segmenting the lung region in purpose for further diagnosis of numerous lung diseases, including COVID-19 and other pneumonia. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.