数值模拟驱动的用于评估磁共振射频电磁辐射的儿童个性化建模研究

It is very hard to precisely estimate the localised specific absorption rate (SAR) for the children under high-intensity magnetic resonance imaging due to the lacking of the individual digital model. This situation thus hinders the optimization for the scan intensity,the sequence design as well as the image quality with the concern that the electromagnetic radiation limits might be exceeded. This study aims to solve this issue by proposing a children's subject-specified modeling method based on constructing the models for evaluating the electromagnetic radiation for various high-intensity scans from one-time low-intensity scan. The multiple graphic processing units (GPU) accelerated parallel Finite-Difference Time-Domain (FDTD) codes have been applied to evaluate and to optimize the modeling schema. This study would focus on the assessment and the optimization for the modeling schema based on SAR errors and the cluster analysis, the hybrid head anatomical modeling method based on the semi-automatical segmentation with anatomical prior knowledge probability template and the manual guidance, the extraction of the subject-specified anatomical information from the low-intensity images and to formulate the subject-specified modeling procedures by means of DARTEL registration as well as morphological methods, and the simplified torso modeling based on Gaussian mixture model. This study can help to optimize the sequence and the coil design and improve the imaging quality while ensuring the children's radiofrequency electromagnetic safety. It also has clinical significance for the cases who need to receive frequent high-intensity and new sequence scans.

儿童在接受高场强磁共振扫描时，因缺乏具有个体特征的数字模型，难于准确地评估其局部电磁辐射比吸收率（SAR）。操作者出于对电磁辐射超标的顾虑，很难有针对性地优化扫描场强和序列，提高图像质量。为此，本研究提出了利用一次较低场强扫描图像建立适用于评估不同高场强设备对儿童电磁辐射的个性化建模方法。研究将利用多GPU并行FDTD模拟计算评估和优化建模方案。将重点研究基于SAR误差和聚类分析进行的建模方案评估和优化流程，基于解剖先验概率模板的自动分割方法和人工标定引导的半自动分割方法相结合的头部解剖学模型建立方案，对低场强磁共振图像进行个性化建模信息提取，基于DARTEL配准与形态学方法的头部建模，及基于高斯混合模型图像分割的躯干简化建模等科学问题。本研究对在确保儿童射频电磁场安全性条件下，优化扫描序列和线圈设计，提高成像质量有重要意义。对需多次接受高场强和新设计序列扫描的案例更具有显著的临床价值。

儿童或婴儿等年幼群体在接受高场强磁共振扫描时，因缺乏具有个体特征的数字模型，难于准确地评估其局部电磁辐射比吸收率（SAR）。在此情况下，操作者出于对电磁辐射超标的顾虑，偏向于使用保守的图像序列，很难有针对性地优化扫描场强和序列，提高图像质量。在本项目研究中，提出了利用较低场强扫描图像建立适用于评估不同高场强设备对儿童电磁辐射的个性化建模方法。研究利用多GPU并行FDTD模拟计算评估和优化建模方案。重点研究了基于SAR误差和聚类分析进行的建模方案评估和优化流程，开发了基于解剖先验概率模板的自动分割方法和人工标定引导的半自动分割方法相结合的头部解剖学模型建立方案，以便对低场强磁共振图像进行个性化建模信息提取，研究了基于DARTEL配准与形态学方法的头部建模，及基于高斯混合模型图像分割的躯干简化建模等科学问题。本研究对在确保儿童射频电磁场安全性条件下，优化扫描序列和线圈设计，提高成像质量有重要意义。对需多次接受高场强和新设计序列扫描的案例更具有显著的临床价值。本项目研究达到了预设成果，共发表SCI文章14篇，EI检索会议论文7篇，形成电磁辐射行业标准2项。生物电磁学年会发表论文5篇，申请专利5项，并获得专利转化24万元。项目组成员在多个国内外学术组织担任重要学术职务。