Real-time visualization and precision targeting in transcranial magnetic stimulation

项目摘要 标题：经颅磁刺激中的实时可视化和精确靶向 经颅磁刺激（TMS）是一种非侵入性的基于设备的神经调节技术，用于探测 神经网络和治疗精神疾病，如重度抑郁症（MDD）和强迫症， 强迫症（OCD）。TMS的治疗效果依赖于放置TMS线圈来准确刺激 潜在的疾病相关的大脑目标。由于TMS诱发的电场（E场）受复杂的 组织结构和大脑几何形状，它依赖于使用计算算法，如边界元 模型（BEM）和有限元模型（FEM），以估计刺激部位。但相对长 这些方法的计算时间对于在脑电刺激期间刺激目标的实时可视化是一个限制 映射和计算治疗计划的最佳线圈位置。在这项拨款中，我们建议发展 一种基于深度神经网络的加速电场预测的方法。我们将开发一种新的深层神经- 通过使用使用具有各向异性的FEM算法计算的训练数据来预测E场的网络架构 组织电导率然后，我们将训练的神经网络集成到3DSlicer软件模块中，用于真实的- 时间电场可视化。此外，我们将开发一种计算算法来搜索最佳线圈 定位以在临床可行的时间内最大化对所选脑目标的刺激。的结果 这笔赠款将把深度学习技术转化为一种有用的工具，以加强TMS在临床中的应用。 research.

Project Summary Title: Real-time visualization and precision targeting in transcranial magnetic stimulation Transcranial magnetic stimulation (TMS) is a non-invasive device-based neuromodulation technique for probing neuronal networks and treating mental disorders such as major depressive disorder (MDD) and Obsessive- Compulsive Disorder (OCD). The treatment efficacy of TMS relies on placing TMS coils to accurately stimulate the underlying disease-related brain target. Since TMS-evoked electric field (E-field) is affected by complex tissue structures and brain geometry, it relies on using computational algorithms, such as boundary element modeling (BEM) and finite element modeling (FEM), to estimate the stimulation site. But the relative long computation time of these methods is a limitation for real-time visualization of the stimulation target during brain mapping and for computing the optimal coil position for treatment planning. In this grant, we propose to develop a deep-neural-network based method to accelerate E-field prediction. We will develop a novel deep-neural- network architecture to predict E-field by using training data computed using the FEM algorithm with anisotropic tissue conductivity. Then, we will integrate the trained neural network into a 3DSlicer software module for real- time E-field visualization. Moreover, we will develop a computational algorithm to search for the optimal coil position to maximize the stimulation of a selected brain target within a clinically feasible time. The outcome of this grant will transform deep-learning techniques into a useful tool to enhance the application of TMS in clinical research.