| Abstract | . Title of the Paper : AI-DRIVEN SEGMENTATION OF CHEST RADIOGRAPHS
FOR ENHANCED EARLY DETECTION OF LUNG CANCER
Journal Name : Journal of the Maharaja Sayajirao University of Baroda
Volume : Vol.59, No.4 , 2025
ISSN Number : 0025-0422
Early detection of lung cancer significantly improves patient survival rates, yet traditional
interpretation of chest radiographs remains challenging due to subtle abnormalities and overlapping
anatomical structures. This study presents an AI-driven approach for segmenting lung nodules in
chest X-rays using a deep learning model based on the U-Net architecture. The proposed system was
trained and evaluated on publicly available datasets, including JSRT and Montgomery County chest
radiographs. Preprocessing steps such as contrast enhancement, resizing, and data augmentation were
applied to improve model performance and generalization. The model achieved a Dice coefficient of
0.89, Intersection over Union (IoU) of 0.82, and accuracy of 92.3%, demonstrating high-quality
segmentation of suspicious lung regions. Visual and statistical comparisons confirm that the system
preserves diagnostic quality while minimizing false positives. The findings indicate that AI-based
segmentation can serve as a reliable tool for assisting radiologists in identifying early-stage lung
cancer. Future work will explore integration with real-time clinical workflows and further refinement
through hybrid architectures and attention mechanisms.
|
|---|
