Abnormal-to-Normal-Brain-MRI-translation-using-Cycle-Consistent-Adversarial-Loss

Using CycleGAN architecture with a problem specific custom loss to translate an MRI scan of brain containing tumor to its normal counterpart and vice versa via the cycle consistency, in order to address the issue of data scarcity in medical imaging

Implementations (Pytorch): CycleGAN | CycleGAN + Global Shortcut | CycleGAN + GS + Anomaly Loss | Least Squares ANTGAN + Identity Loss

Abnormal to Normal Translation

Normal to Abnormal Translation