Update: This work has been accepted for presentation and publishing on 2025 International Conference on Machine Vision (ICMV 2025)! 🎉

We propose UNet-HoVerGNN, a hybrid framework for nuclei instance segmentation and classification that combines CNN-based segmentation with graph-based reasoning. Nuclei centroids are used to construct $k$-nearest neighbor graphs, where node and edge features are extracted from a pretrained U-Net encoder. A GNN refines classification by capturing spatial and contextual dependencies. Evaluated on MoNuSAC, PanNuke, and CoNSeP, our model achieves strong results: 0.603 AJI, 0.641 F1, 0.622 PQ on MoNuSAC; 0.644 AJI, 0.682 F1, 0.661 PQ on PanNuke; and 0.585 AJI, 0.608 F1, 0.595 PQ on CoNSeP, outperforming or matching state-of-the-art models like HoVerNet and similar graph-based methods. Code is available at: GitHub Repository Dataset on Kaggle Models on Kaggle Reproduced experiments

To run the training and evaluation notebooks:

• Download the dataset from Kaggle and place it in a local folder.
• Redirect the DATA_PATH variable in the notebook to point to the dataset location.
• Ensure required dependencies (PyTorch, PyTorch Geometric, etc.) are installed.

Supported datasets:

• MoNuSAC: Multi-organ nuclei segmentation and classification.
• PanNuke: Pan-cancer nuclei classification.
• CoNSeP: Colorectal cancer nuclei segmentation.

To train and evaluate the UNet-HoVerGNN model, navigate to the src directory and run the main.py script with the desired dataset and number of classes. Command-line arguments --dataset and --num_classes override the default values specified in config.py.

Example usage:

cd src
python -m venv .venv
source .venv/bin/activate
# ".venv/Scripts/activate" (Windows)
pip install -r requirements.txt

python main.py --dataset MoNuSAC --num_classes 5 \
               --data_path /path/to/data \
               --output_path /path/to/output
               
python main.py --dataset PanNuke --num_classes 6 \
               --data_path /path/to/data \
               --output_path /path/to/output

python main.py --dataset CoNSeP_Tiled --num_classes 8 \
               --data_path /path/to/data \
               --output_path /path/to/output