An explainable weakly supervised model for multi-disease detection and localization from thoracic X-rays

Anwesh Kabiraj, Tanushree Meena, Kalyan Tadepalli, Sudipta Roy

Abstract: Highlights•Localization of anomaly without having annotated localization in training.•Generalized model to achieve state of the art accuracy in terms of IoBB as well as dice scores.•Novel class activation map pooling to achieve better results than SOTA methods.•The CAN and ADN helps the model to utilize all components of features.•The explainability of the overall process in each layer by showing deconvolutions and saliency maps to unboxing the black box.