This GitHub repository contains trained models to evaluate TDLU involution on new image data. The models are described in "Deep learning assessment of breast terminal duct lobular unit involution: towards automated prediction of breast cancer risk" by Wetstein, S.C. et al.

Our automated method measures TDLU involution using three independent models for acini detection, TDLU segmentation and adipose tissue (fat) segmentation. These three models can be found in this repository. Code to evaluate these models on new image data is also available.

The training weights files for the three models were too large to be uploaded to GitHub but can be downloaded here: https://tue.data.surfsara.nl/index.php/s/RTM6PkUlVEAeIyn