Combined radiomics nomogram of different machine learning models for preoperative distinguishing intraspinal schwannomas and meningiomas: a multicenter and comparative study

Z Xu ¹, Y-H Wang ², Z-Y Cheng ¹

⁰Medical Imaging Center, First Affiliated Hospital of Jinan University, No.613 West Huangpu Avenue, Tianhe District, Guangzhou 510630, Guangdong, China.

Clinical Radiology +

|

June 7, 2024

View abstract on PubMed

Summary

A new combined model using multiparameter magnetic resonance imaging (MRI) radiomics and clinical data effectively distinguishes intraspinal schwannomas from meningiomas. This tool aids in clinical decision-making and prognosis prediction.

Area Of Science

Neurosurgery
Radiology
Oncology

Background

Intraspinal schwannomas and meningiomas are common primary tumors.
Accurate preoperative differentiation is crucial for treatment planning and patient outcomes.
Distinguishing these tumors can be challenging based on conventional imaging alone.

Purpose Of The Study

To develop and validate a novel combined model for differentiating intraspinal schwannomas from meningiomas.
To integrate multiparameter magnetic resonance imaging (MRI) radiomics with clinical features.
To assess the model's diagnostic performance and clinical utility.

Main Methods

Analysis of preoperative MRI and clinical data from 209 patients across three institutions.
Construction of a nomogram using a training cohort (n=111).
Internal and external validation of the nomogram in test (n=48) and independent validation (n=50) cohorts.
Performance evaluation using ROC curves (AUC), DCA, and calibration curves.

Main Results

The combined nomogram significantly outperformed radiomics-only and clinical-only models.
Achieved high AUC values: 0.994 (training), 0.962 (test), and 0.949 (external validation).
Decision curve analysis indicated the nomogram provided the best net benefit; calibration curves showed good agreement.

Conclusions

A nomogram integrating clinical and radiomic features is a powerful tool for distinguishing intraspinal schwannomas and meningiomas.
This model holds significant clinical value for decision-making and prognosis prediction.
Further research can explore its application in diverse clinical settings.