A system and a method for domain-adaptive explainable lung disease classification

A system (100) for automated lung disease classification from chest X-ray images is disclosed. The system (100) comprises an
image acquisition module (102) configured to receive chest X-ray images and a pre-processing engine (104) configured to
generate pre-processed image data. A self-supervised pre-learning module (106) learns generalized feature representations from
heterogeneous datasets, while a domain adaptation module (108) aligns feature distributions between source and target domains
to produce domain-invariant features. A hybrid feature extraction backbone (110), comprising a DenseNet sub-module (110a), a
ResNet sub-module (110b), and a transformer encoder module (110c), extracts multi-scale local and global features. A regional
attention module (114) enhances localization of disease-relevant regions, and a graph attention network module (118) models
spatial and semantic relationships. A feature fusion module (120) generates a unified feature representation, and a classification
engine (122) performs multi-class lung disease classification with improved accuracy and robustness across varying imaging
conditions.