MR Fingerprinting for Epilepsy

摘要 全球6500万人患有癫痫；其中约30%的人对 药物治疗，但可以通过手术治愈。局灶性皮质发育不良(FCD)是 医学上难治性癫痫，通过常规MRI的视觉分析经常被遗漏， 使手术治疗变得非常困难。我们建议开发一种新的定量磁共振成像 专门针对癫痫患者的采集和分析框架，可以提供更多 灵敏和特定的脑结构测量，从而改进FCD的检测和亚型 预测。为此，将分三个步骤制定和验证量化框架： (1)开发高分辨率磁共振指纹(MRF)扫描， 高效、准确和精确地同时量化多个组织属性图。 这些定量地图已显示出在检测和识别 以细微的信号异常为特征。(2)开发图像后处理方法，以 分析量化地图，这将提供量化测量，以突出其他 形态特征，如灰白色边界模糊，皮质异常厚度和 折叠。(3)开发基于机器学习的特征筛选和预测工具 表征区分FCD亚型的组级特征，并预测个体级FCD 位置和子类型。因为FCD的检测和亚型预测都是相关的 对于癫痫发作结果，癫痫专家可以使用此工具提供更个性化和 定制咨询服务。我们提议的工作的结果是，通过将 目前的标准--视觉/定性MRI审查到量化框架，包括 数据采集、后处理和决策支持工具，这最终将导致更好的 治疗计划，减少不必要的术前评估测试(特别是侵入性 评估)，并改善具有破坏性和破坏性的患者的术后癫痫预后 使难治性癫痫失效。我们收购/分析的量化性质 这些方法也有可能被其他中心统一采用，具有很高的一致性。

Abstract Epilepsy affects 65 million people worldwide; approximately 30% of them do not respond to medications but can be cured by surgery. Focal cortical dysplasia (FCD), a major pathology for medically intractable epilepsies, is frequently missed by visual analysis of the conventional MRI, making surgical treatment very difficult. We propose to develop a novel quantitative MRI acquisition and analysis framework specific for epilepsy patients, which could provide more sensitive and specific measures of brain structure, thereby improving FCD detection and subtype prediction. To this end, the quantitative framework will be developed and validated in three steps: (1) Develop high-resolution Magnetic Resonance Fingerprinting (MRF) scan that allows simultaneous quantification of multiple tissue property maps efficiently, accurately and precisely. These quantitative maps have shown to be more sensitive and specific on detecting and characterizing subtle signal abnormalities. (2) Develop image post-processing methods to analyze quantitative maps, which will provide quantitative measurements that highlight additional morphological features, such as gray-white boundary blurring, abnormal cortical thickness and folding. (3) Develop machine-learning-based feature screening and prediction tools to characterize group-level features differentiating FCD subtypes, and predict individual-level FCD location and subtyping. Because detection and subtype prediction of FCD are both associated with seizure outcomes, epileptologists can use this tool to provide more personalized and customized counseling. The result of our proposed work promises a paradigm shift by converting the current standard-care of visual/qualitative MRI review to a quantitative framework, including data acquisition, post-processing and decision support tool, that would eventually lead to better treatment planning, reduction in unnecessary pre-surgical evaluation tests (especially invasive evaluation), and improved post-operative seizure outcomes in patients with devastating and disabling medically intractable epilepsy. The quantitative nature of our acquisition/analysis methods also makes it possible to be uniformly adopted by other centers with high consistency.