Brain Tumor MRI Classification Model (ResNet50)
This is a ResNet50-based image classification model fine-tuned to classify brain tumor MRIs into four categories. This model was trained as part of a project and achieved high accuracy on the test set.
Model Description
This model was trained on the Brain Tumor MRI Dataset. It uses a pre-trained ResNet50 architecture from torchvision, where the final layers were fine-tuned for the specific task of identifying brain tumors from MRI scans.
The model classifies images into the following categories:
gliomameningiomanotumorpituitary
Training Procedure
- Architecture: ResNet50 (Fine-Tuning)
- Optimizer: Adam with differential learning rates
- Loss Function: CrossEntropyLoss
- Epochs: 15
- Scheduler: CosineAnnealingLR
Evaluation Results
The model achieved excellent performance, demonstrating its effectiveness on this dataset.
- Best Validation Accuracy: 97.29%
- Final Test Set Accuracy: 96.95%
Classification Report (Test Set)
| precision | recall | f1-score | support | |
|---|---|---|---|---|
| glioma | 0.94 | 0.97 | 0.95 | 300 |
| meningioma | 0.95 | 0.93 | 0.94 | 306 |
| notumor | 0.99 | 0.99 | 0.99 | 405 |
| pituitary | 0.99 | 0.99 | 0.99 | 300 |
| accuracy | 0.97 | 1311 |
Disclaimer: This model is intended for educational and research purposes only and should not be used for medical diagnosis.