基于多尺度影像组学的胶质母细胞瘤多指标联合诊疗模型研究

胶质母细胞瘤（GBM）是星形细胞肿瘤中恶性度最高的脑胶质瘤，目前最有效的治疗方案是在GBM形成的早期及时发现并实施针对性的治疗，因此，早期发现肿瘤并精准治疗成为关键。医学成像可为GBM的临床无创诊疗提供有效的依据支持，但现有基于影像组学的GBM诊疗存在所利用的影像序列单一、肿瘤表型分析尺度单一以及所构建的诊疗模型预测指标单一等问题，导致GBM表型刻画出现偏倚，诊疗精准度偏低。本项目结合GBM临床影像检查实际和诊疗中多指标考量的需求，拟开展GBM的多尺度影像组学研究。根据不同影像序列特性进行GBM表型分析，研究影像多尺度解析方法，将以往基于GBM瘤体剖析方式拓展至瘤内亚区域和空间生境区域，并深入挖掘不同影像序列和不同尺度的GBM表征的相关性和互补性，揭示GBM表型的深层次机制，从而建立GBM多指标联合诊疗预测模型。本项目研究成果对于更好地诊疗和处理肿瘤疾病具有重要学术研究价值和广泛应用前景。

Glioblastoma multiforme (GBM) is the most malignant glioma in astrocytoma. Scientific evidence has shown that early detection and treatment is the most effective approach to GBM care, and can significantly reduce the mortality rates. Therefore, it is critical to detect and diagnose tumours at an early stage. Medical imaging can provide an effective basis for the clinical non-invasive diagnosis and treatment of GBM. However, the present radiomics analysis based GBM diagnosis and treatment mainly focus on the use of single-sequence image data to quantitatively analyse the tumour phenotype with a single scale and associate the phenotype with a single index, leading to bias in GBM phenotype characterization and low accuracy of diagnosis and treatment. To tackle these limitations, this grant application proposes to establish a multi-scale radiomics analysis framework according to the practical utility of using the multiple sequences images and multiple indexes for diagnosis and treatment of GBM. The multi-scale radiomics analysis will be performed for evaluation of the GBM phenotype using the multiple sequences of magnetic resonance imaging. The multi-scale measurement methods will be established, which extends the traditional image analysis by incorporating the intra-tumoral sub-regions and spatial habitat regions. The proposal will also explore the correlation and complementarity between different image sequences, as well as between different scales of GBM phenotype, to reveal the in-depth information of GBM phenotype. Based on these analyses, the predictive model for the joint prediction of multiple indexes in GBM will be established. The outcome of this research will be of significant academic advances and has broad application prospects for clinical practice of tumours.