基于影像-代谢组学特征预测鼻咽癌放射性脑损伤的智能化模型研究

Building a new model of intelligent diagnosis and treatment for disease, characterized by the fusion of radiomics and metabolomics information, is an important direction to achieve the early diagnosis and prediction of treatment outcome for disease. In this project, we focus on the early prediction and diagnosis of the radiation brain injury in patients with nasopharyngeal carcinoma. Using radiomics-metabolomics techniques and machine-learning analytical strategy, we aim to analyze the dynamic changing pattern of radiomics and metabolomics features from the acquired longitudinal MRI and serological data, to explore the potential inherent relationship between the alterations in brain structure and serum metabolites, and finally, to establish the individualized predictive model for screening out patients at high risk of radiation brain injury following radiotherapy. The successful construction of this individualized predictive model will provide new theoretical and practical basis for the intelligent diagnosis of radiation-induced brain injury.

融合影像与代谢等多组学信息建立疾病的智能诊疗新模式，是实现疾病早期诊断及疗效预测的重要方向。本项目以早期预测和诊断鼻咽癌放射性脑损伤为研究重点，拟利用鼻咽癌患者纵向随访的 MRI 影像学和血清代谢学样本大数据，采用影像组学-代谢组学技术，结合机器学习分析方法，分析两个组学特征的动态演变模式，探究脑结构和血清代谢变化潜在内在联系，构建个体化预测模型，筛选出肿瘤放疗后脑损伤的高危患者，为肿瘤放疗所致的放射性脑损伤的智能化诊断提供新的理论和实践基础。

融合影像人工智能信息建立疾病的智能诊疗新模式，是实现疾病早期诊断及疗效预测的重要方向。本研究以早期预测和诊断鼻咽癌放射性脑损伤为研究重点，拟利用鼻咽癌患者纵向随访的 MRI 影像学样本大数据，采用影像组学人工智能技术，分析两个组学特征的动态演变模式，探究脑结构和血清代谢变化潜在内在联系，构建个体化预测模型，筛选出肿瘤放疗后脑损伤的高危患者，为肿瘤放疗所致的放射性脑损伤的智能化诊断提供新的理论和实践基础。