影响胶质母细胞瘤化疗疗效的主要原因是肿瘤耐药。肿瘤组织中的DNA修复酶-O6-甲基鸟嘌呤DNA甲基转移酶（MGMT）基因是公认的化疗耐药基因，其甲基化程度决定了肿瘤细胞内MGMT浓度高低，进而影响化疗疗效；因此，定量预测基因甲基化程度可指导胶质瘤精准化疗。针对目前MGMT基因启动子甲基化程度定量缺乏无创预测方法的关键问题，本项目组在前期胶质瘤图像特征分析的研究基础上，提出一种基于影像组学对MGMT基因启动子甲基化程度定量实现无创预测的方法。研究内容包括：（1）提取并量化治疗前肿瘤的影像组学特征；（2）以定量PCR方法定量肿瘤组织MGMT基因启动子甲基化水平为参照标准，建立基于影像组学特征的甲基化程度定量预测模型；（3）使用临床独立样本进行模型验证及优化。本研究将为MGMT基因启动子甲基化程度定量预测提供一种无创的方法，定量预测结果将为胶质母细胞瘤精准化疗提供个性化指导。

Drug resistance is the main cause for chemotherapeutic treatment failure in glioblastoma. The O6-methylguanine-DNA methyltransferase (MGMT) gene is a recognized drug-resistance gene, and the MGMT promoter methylation status is closely associated with the intracellular MGMT concentration which determine the treatment efficacy. Therefore, quantification of MGMT gene promoter methylation status before treatment is helpful for guiding precise chemotherapy in glioblastoma. To cope with the scarcity of non-invasive method to quantifying the MGMT gene promoter methylation status in clinical practices, based on our previous studies on gliomas with image texture feature analysis, we proposed a novel radiomics-based quantification method to non-invasively quantify the MGMT gene promoter methylation status. This project would like to: 1) to extract and quantify the radiomics features of pretreatment tumor images; 2) to build a quantification predicting model based on tumor radiomics features with the quantifying PCR-measured MGMT gene promoter methylation status as the reference standard; and 3) to test the predicting performance of the model with an independent cohort and accordingly optimize the model. The result of this project is intended to offer clinicians a tool to non-invasively quantifying the MGMT gene promoter methylation status, and aid in treatment decision-making which lead to personalized chemotherapy in glioblastoma.