Attention-Guided Deep Learning Framework for Automated Lung Disease Classification from CT scans

Abstract

This work introduces a deep learning novel framework for identifying and classifying automated lung disease from CT scanimages. The proposed technique employs a ResNet-50 backbone with dual attention to improve the normal, benign, and malignantl ung condition feature discrimination. The methods attention-guided feature extraction, thorough preprocessing, and inter pretable visualization via Grad-CAM contribute to the proposed work’s superior performance. Experimental results demonstrate the value of the proposed approach for supporting clinical diagnosis