Abstract
Lung segmentation from chest X-ray images plays an important role in the early diagnosis of pulmonary diseases such as pneumonia, lung cancer, and COVID-19. Manual segmentation is time-consuming and susceptible to observer variability, creating the need for automated and reliable approaches. This study proposes an Attention-Augmented U-Net model for accurate lung region segmentation. The proposed framework enhances the conventional U-Net architecture by integrating Attention Gates (AGs) into skip connections between encoder and decoder layers, enabling the network to focus on relevant lung regions while suppressing background structures. Publicly available chest X-ray datasets were used for training and evaluation with preprocessing techniques including normalization and data augmentation. Experimental results achieved a Dice coefficient of 0.93 and improved segmentation performance compared with conventional U-Net models. The proposed system may support radiologists in computer-aided diagnosis and collaborative clinical workflows.

This publication has 22 references indexed in Scilit: