AI-Empowered Electrical Current Distribution Map Prediction from Magnetic Resonance Images in Non-Invasive Brain Stimulation
Create deep learning model that predicts the electrical current distribution maps of a patient given an MRI image of their brain. We hope the model will be able to bypass many of the pipelines currently needed for brain segmentation, allowing for greater accessibility, accuracy, and speed when compared to similar, existing methods.
​
Principal Investigator: Dr. Ruogu Fang
Department: Biomedical Engineering
​
Currently In Progress
Responsibilities
-
Automated processes of finding both tissue thicknesses and a wide variety of fiducial coordinates (ex. center of electrode pad, closest point in bone tissue to area of largest conductivity, etc.).
-
Gathered and organized training data.
-
Assisted in creating the model.
Broader Impact
Electrical current maps are used frequently to tailor certain types of brain treatments to the patient. We hope our model will allow greater access to these maps, by not using common tools that may be locked behind a paywall, and generate them faster.