Related Experiment Video
Updated: Aug 28, 2025
09:53
Role of Diffusion MRI Tractography in Endoscopic Endonasal Skull Base Surgery
Published on: July 5, 2021
3.7K
A radiomics-based study for differentiating parasellar cavernous hemangiomas from meningiomas
Chunjie Wang1, Lidong You1,2, Xiyou Zhang1
1Department of Radiology, The Second Hospital of Dalian Medical University, No. 467 Zhongshan Road, Shahekou District, Dalian City, 116023, China.
Scientific Reports
|September 15, 2022
Summary
Radiomic models using T2-weighted imaging and ADC maps with SVM and KNN classifiers effectively differentiate parasellar hemangiomas from meningiomas, outperforming traditional MRI and neuroradiologists.
Area of Science:
- Radiology
- Medical Imaging
- Machine Learning in Medicine
Background:
- Parasellar tumors, specifically cavernous hemangiomas and meningiomas, present diagnostic challenges.
- Accurate differentiation is crucial for appropriate treatment planning.
Purpose of the Study:
- To evaluate radiomic models for distinguishing parasellar cavernous hemangiomas from meningiomas.
- To compare the classification performance of different MRI sequences and machine learning classifiers.
Main Methods:
- Retrospective study of 96 patients with parasellar tumors (40 cavernous hemangiomas, 56 meningiomas).
- Radiomics features extracted from five MRI sequences (T1WI, T2WI, CE-T1WI, ADC maps).
- Machine learning classifiers (SVM, KNN) evaluated using feature selection methods.
Main Results:
- Radiomic models using T2-weighted imaging (T2WI) and apparent diffusion coefficient (ADC) maps with Support Vector Machine (SVM) and K-Nearest Neighbors (KNN) classifiers showed high performance (AUC > 0.90, F-score > 0.80).
- These models outperformed standard MRI (AUC 0.805) and neuroradiologists (AUC 0.756, 0.545).
- T2WI demonstrated broader applicability than ADC values due to a higher detection rate.
Conclusions:
- Radiomic models incorporating T2WI and ADC data with SVM and KNN classifiers offer a promising approach for differentiating parasellar hemangiomas and meningiomas.
- This AI-driven method enhances diagnostic accuracy compared to conventional techniques.
