Related Experiment Video
Updated: Jun 26, 2026
10:25
Brain Infarct Segmentation and Registration on MRI or CT for Lesion-symptom Mapping
Published on: September 25, 2019
Bridging modalities: a deep learning framework for brain tumor classification via CT-MRI integration and model
Ahmad Almadhor1, Shtwai Alsubai2, Najib Ben Aoun3,4
1Department of Computer Engineering and Networks, College of Computer and Information Sciences, Jouf University, Sakaka, Saudi Arabia.
Frontiers in Computational Neuroscience
|May 11, 2026
Summary
Deep learning models, specifically Convolutional Neural Networks (CNNs), show high accuracy in classifying brain tumors using CT and MRI scans. These AI approaches are effective for early neurological disorder diagnosis, especially with combined imaging data.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Neuroscience
Background:
- Artificial intelligence (AI) and machine learning (ML) show promise in medical image analysis.
- Their application in neurology and psychiatry, particularly for brain tumor classification, is underexplored.
Purpose of the Study:
- To explore deep learning for automated brain tumor classification using multimodal neuroimaging data (CT and MRI).
- To evaluate custom CNNs versus transfer learning (ResNet-18) for this task.
Main Methods:
- Trained and validated custom CNN and ResNet-18 models separately on CT and MRI datasets.
- Extended models to a combined dataset using multimodal fusion.
- Compared performance metrics between the two model families.
Main Results:
- The custom CNN achieved 97% accuracy on CT and 99% on MRI, outperforming ResNet-18 (95% CT, 97% MRI).
- On the combined dataset, CNN achieved 98% accuracy versus ResNet-18's 94%.
- Lightweight CNNs demonstrated high effectiveness and adaptability for neuroimaging tumor detection.
Conclusions:
- Lightweight CNNs are highly effective for neuroimaging-based brain tumor detection, especially with multimodal data.
- AI-driven systems require exploration of modality-specific features and model adaptability for neurological disorders.
