Challenging Current Semi-supervised Anomaly Segmentation Methods for Brain MRI

In this work, we tackle the problem of Semi-Supervised Anomaly Segmentation (SAS) in Magnetic Resonance Images (MRI) brain, which is task automatically identifying pathologies brain images. Our work challenges effectiveness current Machine Learning (ML) approaches application domain by showing that thresholding Fluid-attenuated inversion recovery (FLAIR) MR scans provides better anomaly segmentation maps than several different ML-based detection models. Specifically, our method achieves Dice similarity coefficients and Precision-Recall curves competitors on various popular evaluation data sets for tumors multiple sclerosis lesions. (Code available under: https://github.com/FeliMe/brain_sas_baseline )