Babels.AI

iUS BrainShift Prediction Studio

About this app

iUS BrainShift is a deep-learning tool that predicts and visualizes intraoperative brain shift from paired pre- and post-resectionultrasound sweeps. It was trained on volumetric intraoperative ultrasound data using a 3D U-Net model developed at Babels.AI.

You can upload your own pre/post sweeps (.mnc, .mnc.gz, or .nii.gz), or test it using built-in sample data. Grad-CAM visualization highlights spatial regions influencing predicted deformation.

For research use only, not for clinical decision-making.

Back to Babels.AI

Babel et al., Operative Neurosurgery (2026). View article

© 2026 Babels.AI. Research deployment environment.

About iUS BrainShift

Brain shift during tumor resection makes pre-operative neuronavigation less accurate as surgery goes on. iUS BrainShift is a deep-learning tool that predicts and visualizes that shift from paired pre- and post-resection intraoperative ultrasound (iUS) sweeps.

It uses a 3D U-Net trained on volumetric intraoperative ultrasound, and Grad-CAM maps highlight the regions that most influence the predicted deformation. Sweeps can be uploaded as MINC or NIfTI files, or sample data can be used.

Predicts
Brain deformation between pre- and post-resection ultrasound volumes.
Inputs
Paired ultrasound sweeps (.mnc, .mnc.gz or .nii.gz), or built-in sample data.
Model
3D U-Net with Grad-CAM visualization.

Key terms

Reference: Babel et al., Operative Neurosurgery (2026)

For research use only, not for clinical decision-making.