HookNet-Breast


Logo for HookNet-Breast

About

Image Version:
38bb424b-71fb-45ec-98ea-fd2f45445713
Last updated:
March 1, 2022, 8:20 p.m.

Interfaces

This algorithm implements all of the following input-output combinations:

Inputs Outputs
1
  • Generic Medical Image (Image)
  • Generic Overlay (Heat Map)
  • Generic Overlay (Heat Map)
  • Model Facts

    Summary

    HookNet algorithm for the segmentation of histopathology breast tissue including ductal carcinoma in situ, invasive ductal carcinoma, invasive lobular carcinoma, non-malignant epithelium, fat, and other breast tissue.

    HookNet is a multi-resolution semantic segmentation deep learning model that combines context and details via encoder-decoder branches. Results show that HookNet can simultaneously deal with subtle differences at high-resolution and contextual information. Furthermore, it has the potential to be helpful for any application where context and details are essential.

    Requirements

    Please take the following limitations into account when submitting data to this algorithm: All input images should be in .tif format and contain magnification levels that correspond to a 0.5, and 8.0μm pixel spacing (± 0.25). Furthermore, a tissue mask at 0.5μm pixel spacing (± 0.25) should be provided as well.

    paper: https://www.sciencedirect.com/science/article/pii/S1361841520302541

    code: https://github.com/DIAGNijmegen/pathology-hooknet

    Mechanism

    This algorithm has been developed for histopathology H&E stained breast slides. The algorithm expects two inputs: 1) the H&E slide, 2) a tissue-segmentation mask (which can be obtained via this algorithm (https://grand-challenge.org/algorithms/tissue-background-segmenation/))

    Validation and Performance

    Uses and Directions

    This algorithm was developed for research purposes only.

    Warnings

    Common Error Messages

    Information on this algorithm has been provided by the Algorithm Editors, following the Model Facts labels guidelines from Sendak, M.P., Gao, M., Brajer, N. et al. Presenting machine learning model information to clinical end users with model facts labels. npj Digit. Med. 3, 41 (2020). 10.1038/s41746-020-0253-3