Related Experiment Video
Updated: Jul 7, 2026
06:48
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
8.5K
A deep learning based NeuroFusionNet approach for automated brain tumor diagnosis from MRI
Omara Mustafa1, Salem Alhatamleh2, Hamad Yahia Abu Mhanna3
1Department of Radiology, Korean Medical Center, Lusail, Qatar.
Frontiers in Neuroinformatics
|May 4, 2026
Summary
NeuroFusionNet, a novel deep learning framework, enhances brain tumor classification from MRI scans by integrating generative adversarial networks (GANs) with transfer learning. This approach achieves high accuracy, outperforming existing methods.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computer Vision
Background:
- Brain tumor diagnosis from MRI is challenging due to image variability and manual interpretation limitations.
- Automated methods are needed to improve diagnostic accuracy and efficiency.
Purpose of the Study:
- To develop and validate NeuroFusionNet, a deep learning framework for automated brain tumor classification using MRI.
- To enhance classification performance by combining GAN-based synthetic image generation with transfer learning.
Main Methods:
- NeuroFusionNet utilizes a VGG16 backbone, fine-tuning its last ten layers while keeping earlier layers frozen.
- Generative Adversarial Networks (GANs) create synthetic MRI images to augment real data.
- Both real and synthetic images are processed through the VGG16 network for feature extraction and classification.
Main Results:
- NeuroFusionNet achieved high classification accuracies of 99.05% and 98.75% on two public MRI datasets.
- The framework outperformed several state-of-the-art neural network architectures.
- Consistent superior performance was observed across different datasets.
Conclusions:
- NeuroFusionNet demonstrates significant effectiveness for brain tumor classification on public MRI datasets.
- The proposed framework shows promise for improving automated diagnostic tools in neuro-oncology.
- Further external validation is recommended to confirm generalizability.
