Related Experiment Video
Updated: Nov 10, 2025
Machine Learning Algorithms for Early Detection of Bone Metastases in an Experimental Rat Model
Published on: August 16, 2020
Machine Learning-Based Multiparametric Magnetic Resonance Imaging Radiomics for Prediction of H3K27M Mutation in
Sedat Giray Kandemirli1, Burak Kocak2, Shotaro Naganawa3
1Department of Radiology, University of Iowa Hospital and Clinics, Iowa City, Iowa, USA.
Objective:
H3K27M mutation in gliomas has prognostic implications. Previous magnetic resonance imaging (MRI) studies have reported variable rates of tumoral enhancement, necrotic changes, and peritumoral edema in H3K27M-mutant gliomas, with no distinguishing imaging features compared with wild-type gliomas. We aimed to construct an MRI machine learning (ML)-based radiomic model to predict H3K27M mutation in midline gliomas.
Methods:
A total of 109 patients from 3 academic centers were included in this study. Fifty patients had H3K27M mutation and 59 were wild-type. Conventional MRI sequences (T1-weighted, T2-weighted, T2-fluid-attenuated inversion recovery, postcontrast T1-weighted, and apparent diffusion coefficient maps) were used for feature extraction. A total of 651 radiomic features per each sequence were extracted. Patients were randomly selected with a 7:3 ratio to create training (n = 76) and test (n = 33) data sets. An extreme gradient boosting algorithm (XGBoost) was used in ML-based model development. Performance of the model was assessed by area under the receiver operating characteristic curve.
Results:
Pediatric patients accounted for a larger proportion of the study cohort (60 pediatric [55%] vs. 49 adult [45%] patients). XGBoost with additional feature selection had an area under the receiver operating characteristic curve of 0.791 and 0.737 in the training and test data sets, respectively. The model achieved accuracy, precision (positive predictive value), recall (sensitivity), and F1 (harmonic mean of precision and recall) measures of 72.7%, 76.5%, 72.2%, and 74.3%, respectively, in the test set.
Conclusions:
Our multi-institutional study suggests that ML-based radiomic analysis of multiparametric MRI can be a promising noninvasive technique to predict H3K27M mutation status in midline gliomas.
Insights
Machine learning analysis of MRI scans can predict H3K27M mutations in midline gliomas. This noninvasive radiomic model shows promise for identifying these mutations, aiding in diagnosis and prognosis.
Area of Science:
- Neuro-oncology
- Radiology
- Machine Learning
Background:
- H3K27M mutations in gliomas have significant prognostic implications.
- Previous MRI studies have shown variable imaging features in H3K27M-mutant gliomas, lacking distinct characteristics compared to wild-type gliomas.
- Accurate prediction of H3K27M mutation status is crucial for guiding treatment and understanding disease progression.
Purpose of the Study:
- To develop and validate a machine learning (ML)-based radiomic model using multiparametric MRI to predict H3K27M mutation status in midline gliomas.
- To assess the noninvasive capability of radiomic analysis in differentiating H3K27M-mutant from wild-type midline gliomas.
- To improve diagnostic accuracy for H3K27M mutations, which are critical for patient prognosis.
Main Methods:
- A multi-institutional cohort of 109 patients (50 H3K27M-mutant, 59 wild-type) was analyzed.
- Radiomic features were extracted from conventional MRI sequences (T1-weighted, T2-weighted, T2-FLAIR, postcontrast T1-weighted, ADC maps).
- An extreme gradient boosting (XGBoost) algorithm was employed to build an ML model, with performance evaluated using the area under the receiver operating characteristic curve (AUC).
Main Results:
- The XGBoost model, incorporating feature selection, achieved an AUC of 0.791 in the training set and 0.737 in the test set.
- In the test set, the model demonstrated an accuracy of 72.7%, precision of 76.5%, recall of 72.2%, and F1-score of 74.3%.
- The study included a significant proportion of pediatric patients (55%).
Conclusions:
- ML-based radiomic analysis of multiparametric MRI is a promising noninvasive method for predicting H3K27M mutation status in midline gliomas.
- This approach could aid in the early identification of H3K27M mutations, facilitating timely clinical management.
- The findings support the integration of advanced imaging analysis techniques in the diagnostic pathway for brain tumors.
More Related Videos
04:09Predicting Treatment Response to Image-Guided Therapies Using Machine Learning: An Example for Trans-Arterial Treatment of Hepatocellular Carcinoma
Published on: October 10, 2018
06:44Modeling Brain Metastases Through Intracranial Injection and Magnetic Resonance Imaging
Published on: June 7, 2020
