MIAS-SAM: Medical Image Anomaly Segmentation without thresholding
It addresses a specific challenge in medical image analysis by eliminating the thresholding step, which is incremental but practical for improving segmentation workflows.
This paper tackles the problem of segmenting anomalous regions in medical images without needing to set a threshold, achieving accurate results across three datasets (Brain MRI, Liver CT, and Retina OCT) as measured by DICE score.
This paper presents MIAS-SAM, a novel approach for the segmentation of anomalous regions in medical images. MIAS-SAM uses a patch-based memory bank to store relevant image features, which are extracted from normal data using the SAM encoder. At inference time, the embedding patches extracted from the SAM encoder are compared with those in the memory bank to obtain the anomaly map. Finally, MIAS-SAM computes the center of gravity of the anomaly map to prompt the SAM decoder, obtaining an accurate segmentation from the previously extracted features. Differently from prior works, MIAS-SAM does not require to define a threshold value to obtain the segmentation from the anomaly map. Experimental results conducted on three publicly available datasets, each with a different imaging modality (Brain MRI, Liver CT, and Retina OCT) show accurate anomaly segmentation capabilities measured using DICE score. The code is available at: https://github.com/warpcut/MIAS-SAM