Related Experiment Video
Updated: Oct 22, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.0K
Automatic Meningioma Segmentation and Grading Prediction: A Hybrid Deep-Learning Method.
Chaoyue Chen1, Yisong Cheng1,2, Jianfeng Xu3
1Department of Neurosurgery, West China Hospital, Sichuan University, No. 37 GuoXue Alley, Chengdu 610041, China.
Journal of Personalized Medicine
|August 27, 2021
Summary
A deep-learning system accurately grades meningioma tumors before surgery. This AI tool aids in distinguishing low-grade from high-grade tumors, improving preoperative assessment.
Area of Science:
- Neurosurgery
- Artificial Intelligence
- Medical Imaging
Background:
- Meningiomas are common primary brain tumors.
- Accurate preoperative grading is crucial for treatment planning.
- Current grading methods can be subjective and time-consuming.
Purpose of the Study:
- To evaluate a deep-learning system for preoperative meningioma grading.
- To assess the system's ability in automatic tumor detection, segmentation, and grade prediction.
Main Methods:
- A retrospective study of 643 patients from two institutions.
- Development of a deep-learning system using a modified U-Net for segmentation and DenseNet for grading.
- Integration of segmentation and grading models into a cascade network structure.
Main Results:
- The segmentation model achieved a Dice coefficient of 0.920 ± 0.009.
- The grading model showed an AUC of 0.918 ± 0.006 and accuracy of 0.901 ± 0.039.
- The system demonstrated robust performance on an external validation dataset.
Conclusions:
- Deep learning-based assessment systems can facilitate preoperative grading of meningioma.
- The developed system shows high accuracy and robustness in meningioma classification.
- This AI approach has the potential to improve surgical planning and patient outcomes.
