IMG-20. Artificial intelligence and radiomics in preoperative differentiation of solitary fibrous tumor and meningioma: a systematic review and meta-analysis

Abstract Radiomics and artificial intelligence (AI) have shown promise in the preoperative differentiation of brain tumors. Solitary fibrous tumor or hemangiopericytoma (SFT/HPC) is an aggressive tumor with a high risk of recurrence and metastasis, which bears radiologic similarities with meningioma, making accurate pre-operative diagnosis challenging. This study reviewed current literature on the role of radiomics and AI in diagnosing SFT/HPC and conducted a meta-analysis of studies using radiomics with machine learning (ML) for its utility in pre-operative differentiation of SFT/HPC from meningioma. A total of 287 records were screened, with 14 studies included in the systematic review and 6 in the meta-analysis. The pooled analysis showed strong diagnostic performance of ML models in differentiating SFT/HPC from meningioma, with a pooled AUC of 0.93 (95% CI: 0.91–0.95); accuracy of 0.87 (95% CI: 0.83–0.91); sensitivity of 0.90 (95% CI: 0.85–0.94); specificity of 0.82 (95% CI: 0.78–0.87); positive predictive value of 0.91 (95% CI: 0.88–0.94); and negative predictive value of 0.81 (95% CI: 0.67–0.96). These findings highlight the potential utility of radiomics with ML for pre-operative differentiation of SFT/HPC from meningioma, which may guide surgical planning and treatment decisions. However, there remains a gap between research and clinical application due to concerns about bias and real-world applicability.