MRI-based radiomics for prediction of isocitrate dehydrogenase subtype in glioblastoma multiforme through artificial intelligence models: A systematic review and meta analysis.

Abstract

2074 Background: Gliomas are aggressive tumours with poor prognosis. Isocitrate Dehydrogenase (IDH) mutations are present in approximately 12% of all Glioma tumours and are considered biomarkers for prognosis and response to chemotherapeutic agents. IDH mutant gliomas have better prognosis in comparison to IDH wild type Gliomas. IDH mutant Gliomas exhibit features like T2-Flair mismatch sign, reduced blood flow seen on perfusion-weighted images and reduced enhancement on MRI, which aid in identification of IDH mutation. Radiomic imaging techniques extract quantitative features from medical images like MRI and CT scans with the help of advanced algorithms and the extracted data can be utilized in the development of specific artificial intelligence (AI) models like Neural Networks for the prediction of IDH mutation. Thus, MRI based Radiomics is an emerging non invasive technique in comparison to conventional biopsy, for the determination of IDH mutation. The meta-analysis conducted aims to analyse the diagnostic potential of Radiomic imaging in predicting IDH mutations in Gliomas. Methods: A systematic search was conducted in PubMed, Google Scholar and Scopus. PRISMA guidelines were followed. A boolean expression was constructed to retrieve and select articles from major medical databases. The R Studio package was used to evaluate the potential of the diagnostic test. The Meta, Metadata and Mada packages were utilised to evaluate Pooled accuracy, sensitivity and specificity. Results: A total of 35 studies and 7522 radiomic features were assessed through this meta analysis. The Pooled Sensitivity and Specificity were estimated to be 86.70% ([74.85; 87.51], 95% CI, p< 0.0001, I^2 = 92.7% [90.9%; 94.1%]) and 82.75% ([0.7912; 0.8587], 95% CI, p<0.0001, I^2 = 82.8% [77.4%; 86.9%]) utilising the random effects model. The pooled Accuracy was found to be 81.28% ([0.6037; 0.9253], 95% CI, p>0.01, I^2 = 0.0% [0.0%; 45.4%]). Conclusions: Through the compilation of previously conducted studies, MRI based Radiomics show High Pooled Sensitivity of 86.70% and High Pooled Specificity of 82.75% in the detection of Isocitrate Dehydrogenase mutations in Gliomas. Pooled Accuracy rate of 81.8% indicates steady reliability of Radiomics in the prediction of IDH mutations. MRI based Radiomics is a dependable and consistent non invasive technique in the detection of IDH mutations in GBM and can be utilized for the generation of predictive models, enhancing clinical diagnosis and tailored management based on IDH mutation.