SkinDeepRESEARCHSteve Seguin

Use case · Idea to test

Mark regions in an MRI

Return a mask that a reviewer can inspect.

  1. MRI volume
  2. Predict a region mask
  3. Review the boundaries
Illustration of the proposed workflow.

A medical-image research tool could mark structures or specified findings in a scan for expert review. A region mask is more useful for this task than a written coordinate or a single click point.

Where the small model fits

A scan encoder feeds a segmentation head that labels locations across the image volume. Intermediate masks could be compared with the full model to study whether computation can stop earlier; that requires a medical model and its own training.

What would need to work

Compare with a conventional full-depth segmentation model on patient-separated cases, including timing. Check missed regions, boundary errors and failures across scanners. Overall pixel accuracy can hide a small important finding. No medical segmentation model has been tested here.

Research proposal only. SkinDeep has no clinical model, patient-use demo or validated medical performance for this task.

A dataset to start with

BraTS provides a research task for brain-tumor segmentation in multimodal MRI. It does not cover every anatomy or MRI use case. BraTS MRI segmentation task.