Computer Vision News - May 2020

3 Summary Deep GANCS MRI 7 A notable fact is that the SNR gain is observed when the number of patients is higher than 130 with an ideal number referenced of 150 patients. It seems that, with this number, the full advantage of both learning from historical data and the complexity of the networks is achieved. Comparing the performance loss based on the SSIMand SNRmeasure, it can be shown that GANCS provides a significant speedup on 102-103 but also a slight performance loss (0.76 vs 0.8 and 0.64 for the Input). For quality assessment, the expert opinion of two paediatric radiologists of the Stanford Medical Centre is provided and the rating is provided in the gold-standard as well as the GANCS and CS-WV schemes. They individually assess the reconstruction and then asked for a side-by-side comparison as well. Finally, the diagnostic quality seems to do better than the CS-WV.More studies need to be performed in order to evaluate and validate this approach for image segmentation and interpretation.