ResNet-CBAM-ViT: A Hybrid Deep Learning Model for Robust Brain Tumor Classification in MRI using Frequency and Attention-Based.

This project focuses on automated classification of brain tumors from MRI scans using a hybrid deep learning architecture that combines Convolutional Neural Networks (CNNs) and Vision Transformers (ViT).

Our custom model architecture leverages the local feature extraction power of CNNs and the global attention capabilities of ViT to accurately classify brain tumor types into four categories: Glioma, Meningioma, Pituitary Tumor, and No Tumor.

The model is trained on a preprocessed dataset of axial brain MRI images, and has shown robust performance in validation and test phases.

This is the official ResNet-CBAM-ViT implementation repository with PyTorch.

ResNet-CBAM-ViT: Brain Tumor Classification Architecture

Setup

Clone the repository.

!git clone https://github.com/nafiuny/BrainTumorClassifier_ResNet-CBAM-ViT.git

%cd BrainTumorClassifier_ResNet-CBAM-ViT

!pip install -r requirements.txt

Download dataset

!python download_dataset.py

!python preprocess.py --train_dir Brain_Tumor_MRI_Dataset/Training --output_dir data_preprocessed/data
!python preprocess.py --test_dir Brain_Tumor_MRI_Dataset/Testing --output_dir data_preprocessed/data

!python split_train_val.py \
    --x_path data_preprocessed/X_train_final.npy \
    --y_path data_preprocessed/y_train_final.npy \
    --output_dir data_preprocessed

Train

Train BrainTumorClassifier_ResNet-CBAM-ViT.
You can modify the values of the input parameters as per your requirements. Alternatively, you can run the script with the default settings without changing any parameters.

!python train.py \
    --x_train data_preprocessed/X_train.npy \
    --y_train data_preprocessed/y_train.npy \
    --x_val data_preprocessed/X_val.npy \
    --y_val data_preprocessed/y_val.npy 

Test

You can easily test the trained model either with default settings or by specifying your own paths for datasets and checkpoints.

!python test.py

or

!python test.py \
    --x_test data_preprocessed/X_test_final.npy \
    --y_test data_preprocessed/y_test_final.npy \
    --checkpoint checkpoints/model_epoch_70.pth 

Single Image Prediction

To quickly predict the tumor class of a single MRI image, simply run the following command.

!python predict.py --image_path /path/to/your/image.jpg