Characterization and Quantification of Treatment Response for Patients with Glioblastoma multiforme by Referencing to a multi-scale Atlas of Anatomic and Metabolic Imaging Data

成人高级别胶质瘤（GBM）患者在标准护理治疗后的中位生存期为14个月。GBM的一个特征是微血管异常，血管源性水肿增加，并高度浸润到正常大脑。虽然解剖图像被广泛用于这些特征的诊断和临床评估，但治疗后观察到的变化是非特异性的，可能代表肿瘤与胶质瘤、水肿、坏死和炎症的组合。磁共振波谱成像（MRSI）是解决此类歧义的一种很有前途的技术，它可以提供有关组织的生物和代谢特性的信息。一般来说，脑部疾病，如肿瘤或多发性硬化症，会引起大脑强烈的代谢变化，这使得它们非常适合mri表征。然而，磁共振光谱成像仍未广泛应用于临床的标准模式。造成这种差异的一个原因是该方法在技术上具有挑战性，需要额外的培训和专业知识。此外，由于缺乏一个标准的参考框架来定义不同空间位置的代谢信息的正常值，因此容易由于观看者的主观成分而产生评价的不确定性。该项目的目的是通过提供可靠和稳健的指标来描述病变和周围组织的空间和时间变化，为GBM患者的脑mri数据的自动分析提供一个参考框架。第一步将是获取健康志愿者的核磁共振成像数据，以构建正常人大脑的代谢图谱，该图谱将在验证后提供给研究界。此外，该图谱将作为参考标准，用于识别来自加州大学旧金山分校400多名患者的mri数据中的异常模式，并将其与图像引导组织样本的组织学特性联系起来。建立的关系将允许开发分类算法，以评估对治疗的反应和评估连续患者扫描中的肿瘤进展。这些方法将为预测结果和描述新疗法提供客观标准。这些发现将改善诊断，帮助选择适当的治疗方法，并加速发现和评估改进的治疗方法。

The median survival for adult patients with high-grade glioma (GBM) following standard of care treatment is 14 months. A hallmark of GBM is the presence of abnormal microvasculature, increased vasogenic edema and a high degree of infiltration into normal brain. While anatomic images are widely used for diagnosis and clinical assessment of these features, the changes observed following treatment are non-specific and may represent a combination of tumor with gliosis, edema, necrosis and inflammation. A promising technology for resolving such ambiguities is magnetic resonance spectroscopic imaging (MRSI), which provides information about the biological and metabolic properties of the tissue. Generally, brain diseases like tumors or multiple sclerosis, cause strong metabolic changes in the brain, which makes them very suited for MRSI characterization. However, MR spectroscopic imaging is still not widely used as a standard modality in clinical context. One reason for this discrepancy is that the method is technically challenging, which requires additional training and expertise. In addition, the lack of a standard reference framework, which defines the normal values of metabolic information at different spatial locations, makes it prone to evaluation uncertainties due to the subjective component of the viewer.The objective of this project is to provide a reference framework for the automated analysis of brain MRSI data from patients with GBM by providing reliable and robust metrics that describe the spatial and temporal changes in the lesion and surrounding tissue. The first step will be to obtain MRSI data from healthy volunteers in order to construct a metabolic atlas of the normal human brain, which will be provided to the research community upon validation. Furthermore, the atlas will be used as a reference standard to identify abnormal patterns in MRSI data obtained from over 400 patients at UCSF and link them to histological properties from image-guided tissue samples. The relationships that are established will allow classification algorithms to be developed for assessing response to treatment and evaluating tumor progression in serial patient scans. The methods will provide objective criteria for predicting outcome and characterizing new therapies. The findings will improve diagnosis and assist in selecting appropriate treatments, as well as accelerating the discovery and evaluation of improved treatments.