Related Experiment Video
Updated: Aug 27, 2026
Quantitative Immunohistochemistry of the Cellular Microenvironment in Patient Glioblastoma Resections
Published on: July 31, 2017
Nomogram combining visual imaging and radiomics analysis for distinguishing diffuse hemispheric glioma, H3 G34-mutant
Bingxin Pang1,2, Mengyuan Yuan1, Xiaochen Wang1
1Beijing Tiantan Hospital, Capital Medical University, Beijing, China.
Objective:
This study aims to develop an MRI-based predictive model to distinguish diffuse hemispheric glioma, H3 G34-mutant (H3 G34-mutant DHG) from adult-type diffuse high-grade gliomas.
Method:
Preoperative MRI images were retrospectively collected from 129 patients with pathologically confirmed diffuse high-grade gliomas, including 50 patients with H3 G34-mutant DHG and 79 patients with adult-type diffuse high-grade gliomas. Visual imaging features were extracted in accordance with the Visually Accessible Rembrandt Images (VASARI) scoring system. Subsequently, the Least Absolute Shrinkage and Selection Operator (LASSO) regression was performed to remove redundant and highly collinear features. Univariable and multivariable logistic regression analyses were conducted to screen significant variables for model construction. Radiomic features were extracted from the whole tumor on multi-sequence MRI scans and selected via Mann-Whitney U test, Spearman correlation, Minimum redundancy maximum relevance (MRMR) algorithm, and LASSO regression. A radiomic predictive model was then constructed using the support vector machine (SVM) algorithm. Significant variables derived from clinical-visual imaging features and radiomic features were integrated into a combined predictive model, and a visualized nomogram was developed. Model discrimination performance was evaluated using the area under the curve (AUC), and clinical utility was assessed through decision curve analysis (DCA).
Results:
Age and diffusion restriction were independent predictors of H3 G34-mutant DHG. The nomogram exhibited superior overall performance compared with the clinical-visual imaging model and radiomics model. It yielded optimal AUC values of 0.976 in the training set and 0.929 in test set, along with favorable sensitivity and specificity. Calibration curves and DCA revealed satisfactory calibration and promising clinical utility for this nomogram.
Conclusion:
H3 G34-mutant DHG may display distinct imaging characteristics. A nomogram integrating clinical-visual imaging and radiomic features may facilitate preoperative discrimination between H3 G34-mutant DHG and adult-type diffuse high-grade gliomas.