Neurosurg Focus 61(3):E8, 2026
Could preoperative imaging help anticipate the biological behavior of spinal meningiomas?
This study applies radiomics—quantitative analysis of imaging features—to distinguish tumor pathology groups.
The model showed promising discrimination using open-source tools and internal validation.
Its potential role is to inform counselling, but clinical adoption requires independent testing.
Objective
To develop a preoperative radiomics classifier for spinal meningioma pathology and explore factors associated with disease control.
Methods
Investigators retrospectively reviewed institutional surgical records from 2012–2025. Manually segmented contrast-enhanced images underwent PyRadiomics feature extraction. An ensemble classifier used 10 features and nested cross-validation.
The study’s “high-grade” category combined WHO grade 2 tumors with grade 1 tumors showing atypia; this is not equivalent to conventional WHO high-grade classification.
Main results
Among 74 patients, the model achieved an area under the receiver-operating-characteristic curve of 0.879 and a mean F1 score of 0.748.
The clinical analysis included 55 patients, with median radiographic follow-up of 22.2 months and four recurrences. Subtotal resection was associated with poorer progression-free survival than gross-total resection: HR 10.62 (95% CI 1.46–77.05; P=.019). Institutional abstract
Interpretation
The imaging approach merits further investigation, but discrimination within one institution does not establish performance across scanners, imaging protocols or patient populations. An AUC of 0.879 should not be described as “87.9% diagnostic accuracy.”
The recurrence analysis is exploratory. Its wide confidence interval signals considerable uncertainty, and it does not justify pursuing complete resection at the expense of neurological function.
Limitations
The small retrospective cohort and internal validation limit generalizability. The unconventional pathology grouping complicates comparison with other grading studies. Four recurrence events provide limited support for prognostic modelling.
Publisher and PubMed retrieval failed; verification relied on the authors’ institutional abstract. Full methodological details and supplementary material could not be assessed.
Clinical takeaway
Radiomics is a promising research adjunct for preoperative assessment of spinal meningiomas. This model should not yet determine resection extent, radiation treatment or surveillance schedules without external validation and confirmation of clinical utility.