Interobserver variability in assessing preoperative imaging biomarkers for cerebellar mutism syndrome: a multiobserver pilot study.

Background: Cerebellar mutism syndrome is a well-known complication of paediatric posterior fossa tumour surgery. In recent years, several imaging biomarkers have been suggested to predict cerebellar mutism syndrome based on its probable pathoanatomical causes.

Objective: This study investigates the reliability of preoperative imaging biomarkers for cerebellar mutism syndrome in paediatric posterior fossa tumours. Specifically, it examines the interobserver agreement on the size, invasion, and compression of selected regions of interest with structured magnetic resonance imaging (MRI) reporting.

Materials and methods: Preoperative brain MRI scans from ten paediatric patients with posterior fossa tumours, conducted at a single institution, were analysed. The scans were evaluated by three neuroradiologists from three different institutions across two countries using a structured reporting format. The interobserver agreement was assessed using intraclass correlation coefficient and Fleiss' kappa. All estimates were reported with a 95% confidence interval.

Results: The study found good to excellent agreement in measuring tumour size, tumour volume, and the Evans index. Substantial agreement was found in tumour pathology and location. However, the interobserver agreement was unreliable for invasion and compression of the included anatomical structures.

Conclusion: Findings from this study challenge the reliability of preoperative imaging biomarkers for cerebellar mutism syndrome, emphasising the need for further investigation into consistent and reproducible biomarkers relevant to this syndrome.