多模态MRI影像组学预测脑胶质瘤分子分型的研究进展OA

Glioma,as the most common primary malignant tumor in the central nervous system(CNS),is characterized by high heterogeneity.Accurate molecular subtyping is conducive to formulating treatment strategies and improving prognosis for glioma patients.Although glioma can be diagnosed through surgical procedures or biopsies,such methods are invasive,carrying risks of sampling bias and postoperative complications.Multimodal MRI radiomics,a prominent area of research in disease diagnosis,is capable of integrating the strengths of various MRI imaging techniques.By extracting high-throughput imaging features spanning morphology,texture,functional metabolism and other dimensions,and leveraging machine learning,deep learning as well as statistical analysis tools to build predictive models,this technique has demonstrated significant potential for non-invasive assessment of glioma molecular markers.This paper reviews the recent advances in multimodal MRI radiomics for non-invasively predicting glioma molecular subtypes,points out current research limitations,and suggests future research directions,with the aim of ultimately providing imaging evidence and clinical guidance for preoperative precise diagnosis and the formulation of personalized treatment regimens for glioma patents.