Implementing Vision Transformer for Classifying 2D Biomedical Images

Arindam Halder, Sanghita Gharami, Priyangshu Sadhu, Pawan Kumar Singh^{[0000-0002-9598-7981]}, Marcin Wozniak*^{[0000-0002-9073-5347]}, Muhammad Fazal Ijaz*^{[0000-0001-5206-272X]}

(* Corresponding Authors)

This is the official implementation of the paper, "Implementing vision transformer for classifying 2D biomedical images" published in "Scientific Reports (Nature)"

Directory Structure

+-- gradcam
|   +-- .
|   +-- GradCam_BloodMNIST.ipynb
|   +-- GradCam_BreastMNIST.ipynb
|   +-- GradCam_PathMNIST.ipynb
|   +-- GradCam_RetinaMNIST.ipynb
+-- images
+-- BloodMNIST.ipynb
+-- BreastMNIST.ipynb
+-- PathMNIST.ipynb
+-- RetinaMNIST.ipynb

Note: images folder contains images of various classes of all the four datasets (BloodMNIST, BreastMNIST, PathMNIST, RetinaMNIST) used

Running Codes

Clone the project repository: git clone https://github.com/arindam369/MedMNIST-ViT.git   cd MedMNIST-ViT

Run BloodMNIST.ipynb, BreastMNIST.ipynb, PathMNIST.ipynb, RetinaMNIST.ipynb files to view the result of our model on the respective datasets. gradcam directory contains the codes of visualising GradCAM Analysis on the respective datasets using our proposed model. Use the images of the images directory to perform GradCAM Analysis.

Datasets

The datasets utilized in this study can be found at: albertvillanova/medmnist-v2 | MedMNIST

Citation

If you find this work helpful, consider citing our study:

@article{Halder2024,
   title={Implementing vision transformer for classifying 2D biomedical images},
   author={Halder, Arindam and Gharami, Sanghita and Sadhu, Priyangshu and Singh, Pawan Kumar and Woźniak, Marcin and Ijaz, Muhammad Fazal},
   journal={Scientific Reports},
   volume={14},
   number={1},
   pages={12567},
   year={2024},
   month={may},
   day={31},
   issn={2045-2322},
   doi={10.1038/s41598-024-63094-9},
   url={https://doi.org/10.1038/s41598-024-63094-9}
}