Implementing an accurate image segmentation to extract the lung shape from X-ray images is a vital step in designing a CAD system that diagnoses various types of chest diseases. Lung segmentation is a complex process due to the blurred regions that separate the lung area and the rest of the image. The conventional image segmentation techniques do not meet the ambitions to achieve precise lung segmentation. In this paper, we utilized the Seg-Net semantic segmentation model as a practical approach to distinguish the lung region pixels in X-ray images. The model involves an encoder network that extracts the data from the input images, and a corresponding decoder network that maps the low-resolution encoder feature maps to full input-resolution feature maps for pixel-wise classification. The model is trained and tested using 539 X-ray images from Shenzhen’s publicly available dataset. The robust performance of the Seg-Net model is investigated in terms of global accuracy, dice coefficient, and intersection over union. The model achieved global accuracy (97.71%), dice coefficient (96.95%), and Jaccard index (94.08%). The experimental results indicate that the Seg-Net model is an effective tool for performing complicated segmentation tasks and extracting regions of interest such as lung area, eye vessels, lesions, and tumors from medical images.
Semantic Segmentation – CNN – Seg-Net – Chest X-ray – Deep Learning
How to Cite This Article
Hasan, Dathar Abas and Jader, Umed Hayder
"Semantic Lung Segmentation from Chest X-Ray Images Using Seg-Net Deep CNN Model,"
Polytechnic Journal: Vol. 13:
2, Article 1.