Abstract
INTRODUCTION: Artificial intelligence (AI) and machine learning (ML) are increasingly being used in medical practice, particularly in neuroradiology. Glioblastoma (GBM), a highly aggressive brain tumor, poses significant diagnostic challenges, especially in differentiating true tumor progression (TP) from pseudoprogression (PsP) using conventional MRI. ML and AI techniques offer a promising approach, but their diagnostic accuracy remains debated. OBJECTIVE: To evaluate the overall diagnostic accuracy of ML algorithms applied to various imaging modalities in differentiating TP from PsP in patients with GBM. METHODS: A systematic review and diagnostic meta-analysis was performed to assess the diagnostic efficacy of ML algorithms applied in MRI of GBM patients to differentiate between TP and PsP. PubMed, Scopus, and Cochrane databases were searched up. Pooled sensitivity, specificity, diagnostic odds ratio (DOR), accuracy, and area under the curve (AUC) were calculated using a random-effects model. Heterogeneity and publication bias were also assessed. RESULTS: Eleven studies involving a total of 676 patients were included. ML models demonstrated a pooled sensitivity of 0.899 (95% CI: 0.843–0.936; I² = 16.3%) and a specificity of 0.766 (95% CI: 0.647–0.854; I² = 38.8%). The diagnostic odds ratio was 27.059 (95% CI: 11.094–65.999; I² =61.9%) and the overall accuracy was 0.85 (95% CI: 0.788–0.897; I² = 66%). The AUC was estimated at 0.88 (95% CI: 0.727–0.896), indicating strong diagnostic performance. Deek’s funnel plot showed no significant publication bias (p = 0.283). CONCLUSION: ML-based models demonstrate high diagnostic accuracy in distinguishing TP from PsP in GBM patients, offering a promising tool to enhance radiological assessment and support clinical decision-making. However, further standardization across algorithms and datasets are needed to facilitate their integration into routine clinical practice.