Computer Vision News - May 2020

2 Summary Rese rch 8 It may be important to look at how to improve the robustness of the models against patients with abnormalities, variations in the sampling trajectory, residual motion artefacts in training data, and using 3D spatial correlations for improved quality of volumetric images. One problem currently shown in the research is the "hallucination" effect. Realistic-looking artefacts is a major concern in the reconstruction of medical images, with the potential to mislead radiologists and result in worse patient outcomes. To further explore this fact, in this study a simple test was performed to check for the generalization accuracy for novel cases that are not seen during the training phase. The idea was for the generative model to learn a manifold of structured MR images. In this direction, quantifying possible hallucination risks for GANCS, and devising regularization approaches to make the reconstruction more robust, is an important area that needs to be further researched. Till next time! "Quantifying possible hallucination risks for GANCS and devising regularization approaches to make the reconstruction more robust is an important area that needs to be further researched!"