Evaluating UNET and Mask R-CNN models for their generalizability with multimodal image datasets and detecting nuclei in tumours

by Ajay Singh, Yakup Kohen.

For MBP1413 - 2024

Git hub repo for UNET and Mask RCNN models.

2024_MBP1413_Final_Report_Ajay_Yakup.pdf - Report

Final_2024_MB1314_Ajay_Yakup_MONAI_to_share.ipynb - Can be opened on Google colab for MONAI teaching

Monai_model.pth- finalized model for MONAI. This can be loaded to MONAI on google colab to run predictions on the UNET model trained.

We demonstrate UNET and Mask R-CNN architectures' ability to segment cell nuclei from a 2018 Data science bowl on Kaggle.
670 images were contained in the training dataset, of which was utilized in the project.

Data generated can be found on google drive: UNET: https://drive.google.com/drive/folders/132ZV7R1SpAEMkkm3FN-kZ22Xlp015HHU?usp=sharing