Spatial non‐local attention for thoracic disease diagnosis and visualisation in weakly supervised learning. Issue 11 (30th July 2019)

• Record Type: Journal Article
• Authors: Yang, Menglin; Li, Ding; Zhang, Wensheng
• Is Part Of: IET image processing. Volume 13:Issue 11(2019)
• Page Start: 1922
• Page End: 1930
• Publication Date: 2019-07-30
• Subjects: feature extraction -- diseases -- medical image processing -- patient diagnosis -- image segmentation -- image classification -- supervised learning -- image annotation; thoracic disease diagnosis -- weakly supervised learning -- computer‐aided diagnosis -- coarse image‐level annotations -- pathological area -- nonlocal aware features -- fine‐grained expertise -- clinical diagnosis -- visualisation method -- chest X‐ray dataset -- spatial nonlocal attention mechanism -- ResNet‐SNA; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/iet-ipr.2019.0032
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract : Weakly supervised learning is capable of achieving fine‐grained tasks with coarse annotations, which has shown great potential in computer‐aided diagnosis. This study aims to achieve thoracic disease diagnosis in a weakly supervised manner only with coarse image‐level annotations. Except for considering the performance of disease diagnosis, the study concentrates more on discovering the location of the pathological area which is used as visualised evidence for interpretability of diagnosis and the following retrospective analysis. To harvest more associated pathological areas, spatial non‐local attention mechanism to learn non‐local aware features is investigated. Further, a simple, effective, and widely applicable model ResNet‐spatial non‐local attention (SNA) is developed for these two objectives. Besides, an effective visualisation method compatible with the proposal is introduced. The effectiveness of the proposed ResNet‐SNA was validated on the large publicly available chest X‐ray dataset, ChestX‐ray14. Compared with the baseline model, the proposed model improved by 7.96% averaged over 14 diseases, achieving 0.8247 area under the scores up to the highest classification results compared with related works. For localisation, the proposed model improved the performance significantly without using any extra information. More importantly, the proposal only requires image‐level annotations without fine‐grained expertise, which is cost‐effective and expected to apply in clinical diagnosis.