Stanford researchers slash brain ultrasound processing time 2,500x
New AI model corrects skull distortions in tFUS 2,535x faster than traditional methods
A new few-shot deep learning framework corrects phase-amplitude aberrations in transcranial focused ultrasound (tFUS), using patient CT images to predict per-element corrections for a 96-element 3D phased-array transducer. Pretrained on diverse skull geometries and fine-tuned with only ten target points, the framework adapts to unseen patients without full patient-specific simulation. In leave-one-out cross-validation across 12 skulls, it achieved a mean phase CMAE of 0.155 rad, amplitude rMAE of 9.089%, focal centroid error of 0.467 mm, Dice score of 94.422%, and peak pressure ratio of 92.332%—about 2,535 times faster than conventional time-reversal simulation.
- Uses a 96-element 3D phased-array transducer with patient CT-based corrections
- Achieves 2,535x speedup over conventional time-reversal simulations
- Validated on 12 skulls with 94.4% target accuracy and 0.467 mm focal precision
Why It Matters
Enables real-time, patient-specific tFUS therapy planning, accelerating non-invasive brain treatment development and deployment.