Multimodal Generalized Zero Shot Learning for Gleason Grading using Self-Supervised Learning
This work addresses the challenge of quick, non-invasive prostate cancer diagnosis for medical applications, but it is incremental as it builds on existing GZSL and self-supervised learning techniques.
The paper tackles the problem of predicting Gleason grades for prostate cancer diagnosis from non-invasive MRI images in a generalized zero-shot learning setting, where training images for some grades are unavailable. It proposes a method using a conditional variational autoencoder and cycle GANs to generate synthetic features, achieving performance close to fully supervised methods and outperforming other GZSL approaches.
Gleason grading from histopathology images is essential for accurate prostate cancer (PCa) diagnosis. Since such images are obtained after invasive tissue resection quick diagnosis is challenging under the existing paradigm. We propose a method to predict Gleason grades from magnetic resonance (MR) images which are non-interventional and easily acquired. We solve the problem in a generalized zero-shot learning (GZSL) setting since we may not access training images of every disease grade. Synthetic MRI feature vectors of unseen grades (classes) are generated by exploiting Gleason grades' ordered nature through a conditional variational autoencoder (CVAE) incorporating self-supervised learning. Corresponding histopathology features are generated using cycle GANs, and combined with MR features to predict Gleason grades of test images. Experimental results show our method outperforms competing feature generating approaches for GZSL, and comes close to performance of fully supervised methods.