Radiomics applications for diagnostic and prognostic performance in trigeminal neuralgia: Systematic review and meta-analysis.

PURPOSE: This study aimed to determine the diagnostic performance, specifically the sensitivity, specificity, and area under the curve (AUC), of radiomics-based models for the identification and prediction of TN.

MATERIALS AND

METHODS

This systematic search, registered on PROSPERO (CRD420251074464), was conducted in PubMed, Embase, and Cochrane to identify studies evaluating radiomics-based models in TN. Eligible studies assessed diagnostic or prognostic performance with reported outcomes, regardless of publication date or design. Study selection and data extraction were performed independently by two reviewers. Risk of bias was assessed using QUADAS-2, and methodological rigor was evaluated via the Radiomics Quality Score (RQS).

RESULTS

A total of nine studies met the eligibility criteria, comprising 2033 participants. Studies were stratified by objective: six evaluated diagnostic performance (identification of symptomatic trigeminal nerves) and three evaluated prognostic performance (prediction of outcomes after MVD or PBC). For diagnostic models, the pooled AUC was 0.86 (95% CI, 0.81-0.92), with substantial heterogeneity (I² = 91.8%; τ² = 0.0043; p < 0.0001). For prognostic models, the pooled AUC was 0.83 (95% CI, 0.78-0.89), with no heterogeneity (I² = 0.0%; τ² = 0; p = 0.567).

CONCLUSIONS

This systematic review and meta-analysis indicates that machine-learning radiomics models show promising diagnostic and prognostic performance for symptomatic trigeminal neuralgia. Nonetheless, current evidence is limited by predominantly single-center, retrospective studies lacking external validation. Radiomics holds substantial potential as an objective adjunct to clinical assessment; further high-quality research is needed to strengthen its applicability in TN management.

PLAIN LANGUAGE SUMMARY: Trigeminal neuralgia is a condition that causes severe facial pain and can be difficult to assess. This study reviewed published research on radiomics, which uses artificial intelligence to analyse patterns in medical images, to assess how well it can identify trigeminal neuralgia and predict treatment outcomes. This study found that radiomics-based models showed good accuracy for both diagnosis and prediction, although most studies were from single centres and lacked external testing. This matters because artificial intelligence may help support future care decisions, although more high-quality studies are needed before routine use.