Skip to content

Latest commit

 

History

History
14 lines (9 loc) · 1.06 KB

README.md

File metadata and controls

14 lines (9 loc) · 1.06 KB

TIAger Algorithm

This repository contains the code for the TIAger team's algorithm that was submitted as an entry to the TIGER challenge. The associated paper for this code can be found here.

Currently this repository takes H&E-stained whole slide images (WSIs) from breast cancer as input and outputs a predicted TILs score for the slide. This is done by first segmenting the tumour-stroma areas. Following this, the pipeline will generate a bulk tumor segmentation and detect TILs within the stroma of the tumour bulk. Finally, it will determine the proportion of this area that is TILs in order to geenrate a TILs score between 0 and 100.

The weights for this model can are licensed under CC BY-NC 4.0, and can be found here.

Authors

This code is made by Adam Shephard and Mostafa Jahanifar, as part of the TIA Centre, University of Warwick, Coventry, UK. It is based on the code developed by the TIGER challenge organisers.