Related Experiment Video
Updated: May 10, 2025
14:14
Targeting Neuronal Fiber Tracts for Deep Brain Stimulation Therapy Using Interactive, Patient-Specific Models
Published on: August 12, 2018
8.8K
A novel transfer learning framework for non-uniform conductivity estimation with limited data in personalized brain
Yoshiki Kubota1, Sachiko Kodera1, Akimasa Hirata1
1Department of Electrical and Mechanical Engineering, Nagoya Institute of Technology, Gokiso-cho, Showa-ku, Nagoya 466-8555, Japan.
Physics in Medicine and Biology
|April 25, 2025
Summary
This study introduces a new transfer learning method for accurate non-uniform conductivity estimation in human head models, improving transcranial magnetic stimulation (TMS) personalization. The approach enhances conductivity estimation accuracy significantly, even with limited data.
Area of Science:
- Medical Imaging
- Computational Neuroscience
- Biophysics
Background:
- Personalized transcranial magnetic stimulation (TMS) requires accurate individualized head models.
- Estimating non-uniform conductivity in these models is challenging due to limited ground truth data.
Purpose of the Study:
- To develop a novel transfer learning approach for automatic non-uniform conductivity estimation in human head models.
- To improve conductivity estimation accuracy with limited data for personalized TMS.
Main Methods:
- A transfer learning-based method was developed, utilizing segmentation models generated from T1- and T2-weighted MRI.
- A Transformer was integrated into the segmentation model, and Attention Gates with Residual Connections were used for conductivity estimation.
- The approach was designed for efficient learning from small datasets.
Main Results:
- The proposed method achieved a 2.4% improvement in segmentation accuracy and a 29.1% increase in conductivity estimation accuracy compared to CondNet.
- Superior conductivity estimation accuracy was demonstrated even with only three training cases, outperforming CondNet.
- Generated conductivity maps led to improved brain electrical field simulations.
Conclusions:
- The developed method significantly enhances conductivity estimation for individualized head models.
- This approach shows high utility in brain electrical field simulations and potential for other medical image analysis tasks.
