Quantitative (Perfusion and Diffusion) MRI Biomarkers to Measure Glioma Response

摘要 我们研究计划的持续目标是优化和传播有效的基于成像的 脑瘤个人化治疗策略。神经肿瘤学中的当前反应评估(RANO) 仅结合解剖成像的标准不足以区分肿瘤和治疗效果。 (TE)。在没有明确确认肿瘤进展的情况下，不建议对以下几种情况改变治疗方法 在标准治疗后的几个月。因此，患者不能转而使用潜在更有效的药物。 治疗--这一限制可以通过更可靠的成像技术来克服。 为此，在之前的资金周期中，我们演示了几种定量成像的可行性 (气)可靠地区分肿瘤和治疗效果并预测治疗反应的工具。这些QI工具 包括机器学习方法来校准T1w图像，从而能够创建量化增量T1 (QDT1)映射。QDT1能够检测到真正的对比度增强病变体积(CELV)。QDT1 与我们成熟的动态磁化率对比(DSC)MRI方法一起，用于确定rCBV(相对 脑血容量)，用来产生一个新的生物标记物，肿瘤负荷分数(FTB)，以描绘 基于体素的CELV内肿瘤范围。这些基于灌注的QI工具与我们的 扩散磁共振成像技术，包括功能扩散图(FDM)和最近的RSI(限制 波谱成像)，提供脑肿瘤的全面评估及其与治疗效果的区别。 现在，为了将这项技术转化为临床试验和日常实践，一些最终更新和 按照这里的建议，需要进行临床验证研究。第一，在临床环境中简化采用和测试 随着简化的QI技术的开发，建议对单个QI技术进行改进 工作流程(目标1)。为了促进DSC-MRI和FTB生物标记物的广泛采用，将进行以下研究 以确认单剂DSC-MRI方法可以取代标准的双剂量方法，而不需要 影响rCBV的准确性或FTB图的创建(目标1.1)。此外，配准和分割 算法将进行更新，以包括可变形配准和深度学习方面的最新进展 CELV、非强化病变体积(NELV)和每个QI指标的纵向报告(目标1.2)。 最后，将创建一个包含这些改进的精简工作流程(目标1.3)。目标2研究 将使用临床试验数据(目标2.1-2.2)和日常临床实践(目标2.3-2.4)测试QI工具和工作流程。 这一新的QI-RANO工作流程的临床验证，与证据相比，显示出更好的预测 根据目前的衡量标准，有可能导致脑肿瘤负担评估方式的范式转变。

Abstract The continuing goal of our research program is to optimize and disseminate effective imaging-based strategies to personalize brain tumor treatment. Current Response Assessment in NeuroOncology (RANO) criteria, which incorporate anatomic imaging only, are insufficient for distinguishing tumor from treatment effect (TE). Without definitive confirmation of tumor progression, no treatment changes are recommended for several months after standard therapies. Thus, patients are precluded from switching to potentially more effective therapies—a limitation that could be overcome with more reliable imaging techniques. To this end, during the previous funding cycle, we demonstrated the feasibility of several quantitative imaging (QI) tools to reliably distinguish tumor from treatment effect and predict treatment response. These QI tools include a machine-learning approach to calibrate T1w images enabling the creation of quantitative delta T1 (qDT1) maps. The qDT1 enable the detection of true contrast enhancing lesion volume (CELV). The qDT1 together with our proven dynamic susceptibility contrast (DSC) MRI methods, for determination of rCBV (relative cerebral blood volume), are used to generate a new biomarker, fractional tumor burden (FTB), to delineate the extent of tumor within CELV on a voxel-wise basis. These perfusion-based QI tools in combination with our diffusion MRI technology, which includes functional diffusion maps (FDMs) and more recently RSI (restriction spectrum imaging), provide a comprehensive assessment of brain tumor and its distinction from treatment effect. Now, in order to translate this technology for use in clinical trials and daily practice, some final updates and clinical validation studies are needed as proposed here. First, to ease adoption and testing in the clinical setting improvements are proposed for the individual QI technologies along with the development of a streamlined workflow (Aim 1). To improve the widespread adoption of DSC-MRI and FTB biomarker, studies will be performed to confirm that a single-dose DSC-MRI method can replace the standard double-dose method without affecting the accuracy of rCBV or the creation of FTB maps (Aim 1.1). Also, registration and segmentation algorithms will be updated to include deformable registration and recent advances in deep learning for longitudinal reporting of CELV, non-enhancing lesion volumes (NELV) and each of the QI metrics (Aim 1.2). Finally, a streamlined workflow that incorporates these improvements will be created (Aim 1.3). The Aim 2 studies will test the QI tools and workflow using clinical trial data (Aim 2.1-2.2) and daily clinical practice (Aim 2.3-2.4). Clinical validation of this new QI-RANO workflow, with evidence showing improved prediction in comparison to current measures, has the potential to cause a paradigm shift in how brain tumor burden is assessed.