About:

This module segments your uploaded CT brain scan into 7 important regions including "lateral ventricles [red]", "cerebral matter [green]", and subarachnoid space [blue]" among others. It utilizes adopted Segment Anything (SAM) architecture using masked prompts for each relevant class: we see significant improvements in segmentation scores (15% ventricles, 19.1% subarachnoid) compared to its Unet-based predecessor. A Support Vector Machine is then used to determine if the subject has possible Normal Pressure Hydrocephalus (NPH).

Paper: Automated Segmentation of CT Scans for Normal Pressure Hydrocephalus.

Input

The input should be a single image or a folder of images in nifti format (ending in .nii.gz).

Outputs

• Info: CSV file of volumetric measurements and NPH prediction.
• Segmentation Mask:A segmentation mask in nifti format. If folder of images is used, a folder of segmentation masks are returned.

User Manual

  Click Here for more detailed instruction on how to run this module.