CanineCutaneousTumors

This repository contains code examples for working with our publicly available dataset on microscopic whole slide images (WSIs) of canine cutaneous tumors. The dataset can be downloaded here via the website of The Cancer Imaging Archive.

Tissue Segmentation

We provide two Jupyter notebooks for training and applying a segmentation network to the dataset:

• Training: segmentation_training.ipynb
• Inference: segmentation_inference.ipynb

Tumor Classification

We provide two Jupyter notebooks for training and applying a classification network to the dataset:

• Training: classification_training.ipynb
• Inference: classification_inference.ipynb

Slide Inference

We provide a Jupyter notebook for WSI inference and performance evaluation:

• slide_inference.ipynb

Visualization

We provide two SlideRunner PlugIns for visualization of the segmentation and classification results.

Pre-trained Models

We provide two pre-trained models for patch segmentation and classification. These can be found in the models folder.

Annotation Conversion

We provide six python modules to convert various annotation formats into one another. These can be found in the annotation_conversion folder.