Related Experiment Video
Updated: Aug 8, 2025
10:25
Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping
Published on: September 25, 2019
48.2K
A framework for brain tumor detection based on segmentation and features fusion using MRI images
Almetwally Mohamad Mostafa1, Mohammed A El-Meligy2, Maram Abdullah Alkhayyal1
1Department of Information Systems, College of Computer and Information Sciences, King Saud University, P.O. BOX 51178, Riyadh 11543, Saudi Arabia.
Brain Research
|February 26, 2023
Summary
A new automated method enhances brain tumor detection using magnetic resonance imaging (MRI) segmentation and feature fusion. This approach significantly improves accuracy and outperforms existing techniques for diagnosing brain tumors.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Brain tumors, characterized by irregular cell growth, are diagnosed using methods like MRI.
- Current 3D MRI segmentation for brain tumors is operator-dependent and time-consuming.
- There is a need for robust, automated brain tumor detection systems.
Purpose of the Study:
- To develop and validate a novel, automated brain tumor detection system.
- To improve the accuracy and efficiency of brain tumor diagnosis from MRI scans.
- To overcome the limitations of manual segmentation in 3D MRI.
Main Methods:
- Image pre-processing using Gaussian Filter (GF) and SynthStrip for skull stripping.
- Automated segmentation and feature fusion for brain tumor identification.
- Training and testing on established benchmarks (Figshare and Harvard datasets).
Main Results:
- Achieved high performance metrics: 99.8% accuracy, 99.3% recall, 99.4% precision, 99.5% F1 score, and 0.989 AUC.
- Demonstrated superior performance compared to existing Deep Learning (DL), classical, and segmentation-based methods.
- Cross-validation on the Harvard dataset yielded 99.3% identification accuracy.
Conclusions:
- The proposed automated method offers a significant advancement in brain tumor detection.
- This approach provides a robust and accurate alternative to conventional diagnostic techniques.
- The system demonstrates high potential for clinical application in neuro-oncology.
