Novel DWI Methods to Minimize Postoperative Deficits in Pediatric Epilepsy Surgery.

项目摘要/摘要 对于数以百万计的癫痫耐药癫痫患者，手术切除致痫脑组织 往往是唯一剩下的治疗选择。特别是在年轻患者中，他们没有获得癫痫控制或 遭受药物的不可接受的副作用，人们对癫痫发作的影响有进一步的担忧 关于发育：即使是短暂但反复的癫痫发作也会导致认知退化和有害的心理社会 效果。这促使人们迫切需要研究一种更结构化、更定量、更非侵入性的工具， 它能够通过准确地提供 应用术前影像数据评估术后预后的可能性 在全脑水平上进行分析。该项目的总体目标是开发一种新的收益-风险分析工具。 用于对儿童耐药灶性癫痫的术前评估。为了实现这个目标，我们将确认一个国家-- 最先进的基于深度学习的扩散磁共振技术，以提供切除边缘(即距离 在致痫区域和口才区域之间)，从而产生最大效益(即，发作自由和长时间 术语神经认知改善)和最小化风险(即口才功能缺陷，包括 运动/语言/听觉/视觉)。在NIH的支持下，我们已经建立了扩散加权成像的最大值a 用卡尔曼滤波进行后验概率(DWI-MAP)分析，可以为患者个体提供 最佳切除范围，成功避免93%/91%/90%的运动/语言/视觉障碍 ≥患者75%的患者受益于癫痫发作自由。最近我们也发现了一个很深的 卷积神经网络(DCNN)可以提供良好的准确率(94%-100%)来分类真阳性纤维束 这表明，基于DCNN的脑区分类在 检测不同功能的脑白质通路。该项目的目标1将调查组合 使用卡尔曼滤波的基于DCNN的区域分类甚至可以更好地预测切除边缘，结果是 在一个大的队列中，癫痫发作自由和避免功能缺陷。AIM 2将调查高级DWI 整合DCNN和DWI Connectome的方法通过提供1)术前帮助决定是否适时手术 术后神经认知改善的高可能性背后的影像标志物和2)机制 洞察脑结构重组与术后言语智商改善的关系。这样做的结果 该项目预计将通过将深度学习转化为- 基于扩散磁共振成像技术优化手术切缘，预测术后神经认知结果， 并确定了术后脑重组的具体机制，为进一步优化奠定了基础 手术干预前的临床效益-风险分析。

Project Summary/Abstract For the millions of epilepsy patients with drug-resistant seizures, surgical resection of epileptogenic brain tissue is often the only remaining therapeutic option. Especially in young patients who do not obtain seizure control or suffer from unacceptable side effects from medications, there are further concerns about the effect of seizures on development: even brief but repetitive seizures cause cognitive regression and detrimental psychosocial effects. This motivates a particular urgency to investigate a more structured, quantitative, and non-invasive tool, which is capable of informing families and providers to decide timely surgery by accurately providing the probabilities of both favorable and unfavorable postoperative outcomes using data from preoperative imaging analysis at the whole-brain level. The overall goal of this project is to develop a novel tool of benefit-risk analysis for the presurgical evaluation of pediatric drug-resistant focal epilepsy. Toward this goal, we will validate a state- of-the-art deep learning-based diffusion MRI technique to provide the resection margin (i.e., the distance between epileptogenic area and eloquent area) resulting in maximized benefits (i.e., seizure freedom and long- term neurocognitive improvement) and minimized risk (i.e., deficits in eloquent functions including motor/language/hearing/vision). With NIH support, we have established diffusion-weighted imaging maximum a posteriori probability (DWI-MAP) analysis with Kalman filter, which can provide individual patients with the optimal resection margin, yielding successful avoidance of motor/language/visual deficits in 93%/91%/90% of patients with ≥75% of patients benefiting from seizure freedom. Recently, we have also found that deep convolutional neural network (DCNN) can provide an excellent accuracy (94-100%) to classify true positive tracts of eloquent brain areas, suggesting that DCNN-based tract classification may outperform the DWI-MAP in detecting diverse function-specific white matter pathways. Aim 1 of this project will investigate if a combination of DCNN-based tract classification with Kalman filter even better predicts the resection margin, resulting in seizure freedom and avoidance of functional deficits at a large cohort. Aim 2 will investigate if an advanced DWI approach integrating DCNN and DWI connectome helps decide timely surgery by providing 1) preoperative imaging markers underlying high likelihood of postoperative neurocognitive improvements and 2) mechanistic insight in structural brain reorganization associated with postoperative verbal IQ improvement. The results of this project are expected to ultimately improve clinical management of pediatric epilepsy by translating deep learning- based diffusion MRI technique to optimize the surgical margin, predict the postoperative neurocognitive outcome, and determine specific mechanism of postoperative brain reorganization, which will be validated for optimizing clinical benefit-risk analysis before surgical intervention.