Weakly-Supervised Automatic Biomarkers Detection And Classification Of Retinal Optical Coherence Tomography Images

Optical coherence tomography (OCT) is a primary imaging technique for ophthalmic diagnosis, which has been widely used in retinal disease diagnosis. The retinal biological markers, or biomarkers, play an important role in the management of chronic eye conditions. Hence, it is critical to detect and classify biomarkers from OCT B-scan for patients. In this paper, we proposed a novel weakly-supervised approach utilized healthy data and image-level labels through integrating adversarial generative network and guided attention into one framework. The framework includes an anomaly detection network and a classification network. The anomaly detection network reconstructs the input images with biomarkers to the normal and compared to locate biomarkers area. Inspired by guided attention in network, we utilized the discriminator trained in the anomaly detection network as a classifier twice to reduce model parameters and obtain a complete attention map with class information to get biomarkers classes. Experiment results on a large dataset demonstrate the effectiveness of the proposed detection and classification framework.