Related Experiment Video
Updated: Jul 3, 2025
09:17
Digital Spatial Profiling for Characterization of the Microenvironment in Adult-Type Diffusely Infiltrating Glioma
Published on: September 13, 2022
2.3K
T2-FLAIR mismatch sign and machine learning-based multiparametric MRI radiomics in predicting IDH mutant 1p/19q
1Department of Radiology, The First Affiliated Hospital of Nanjing Medical University, Nanjing, Jiangsu Province 210029, China.
Clinical Radiology
|February 15, 2024
Summary
Machine learning and MRI radiomics accurately predict 1p/19q non-co-deletion in lower-grade gliomas. This approach aids in non-invasive pathological diagnosis, guiding clinical decisions for glioma patients.
Area of Science:
- Neuro-oncology
- Radiology
- Machine Learning
Background:
- Lower-grade gliomas (LGGs) require accurate molecular subtyping for prognosis and treatment.
- 1p/19q non-co-deletion is a key genetic marker in IDH-mutant LGGs.
- Current methods for determining 1p/19q status can be invasive or time-consuming.
Purpose of the Study:
- To evaluate the T2-weighted (T2)-fluid-attenuated inversion recovery (FLAIR) mismatch sign for predicting 1p/19q non-co-deletion in LGGs.
- To assess the efficacy of machine learning-based multiparametric MRI radiomics in predicting this genetic alteration.
- To develop a combined model integrating imaging features for improved diagnostic accuracy.
Main Methods:
- 146 patients with pathologically confirmed IDH-mutant LGGs were included.
- T2-FLAIR mismatch sign and conventional MRI features were analyzed.
- Radiomics features were extracted from various MRI sequences (T1WI, T2WI, FLAIR, ADC, CE-T1WI).
- A stacking model was constructed using the best-performing models and the T2-FLAIR mismatch sign.
Main Results:
- The T2-FLAIR mismatch sign was significantly more common in the 1p/19q non-co-deleted group (AUC=0.692).
- The developed stacking model achieved high performance, with an AUC of 0.925 in the training cohort and 0.886 in the testing cohort.
- The stacking model demonstrated excellent accuracy (0.882 training, 0.864 testing) in predicting 1p/19q non-co-deletion.
Conclusions:
- Multiparametric MRI radiomics, particularly when combined with the T2-FLAIR mismatch sign, is a powerful tool for predicting 1p/19q non-co-deletion in LGGs.
- This non-invasive approach can serve as a valuable supplementary diagnostic tool, potentially reducing the need for invasive biopsies.
- The findings offer guidance for clinical practice, aiding in the stratification and management of glioma patients.
