This model uses a CNN to classify tumors in MRI scans into 4 classes: no tumor, meningioma tumor, pituitary tunmor, or glioma tumor.
It's able to achieve 75% accuracy on unseen test set, and while this is not industry-grade accuracy since the model is trained on a CPU with limited capacity, it can be a useful tool to help medical students study and build up intuition with MRI images, or help medical professionals get a quick sense of what a tumor might be.
Example photos you can use to test the model:
