---

About

Model Facts

Summary

This algorithm is designed to automatically detect artifacts in whole-slide images (WSIs). Leveraging deep learning techniques, it focuses on identifying six common artifacts such as ink, folds, out-of-focus areas, markers, bubbles, and dust.

Mechanism

This algorithms takes a whole-slide image as input, and produces as output a segmentation map with multiple labels predicted, one for each type of artifact detected in the slide. The used method is based on DeepLab-V3 with an EfficientNet-B2 backbone, trained with manually annotated regions from both H&E-stained and IHC-stained whole-slide images. The method is therefore applicable to both H&E and IHC slides.

The result can be downloaded or viewed online. Results will contain:

1. Artifact-free tissue
2. Tissue folds
3. Ink
4. Air bubbles
5. Dust
6. Pen marker
7. Out-of-focus

Additionally, the model returns a quality score based on the percentage of tissue covered by artifacts.

Validation and Performance

Class	Dice
Folds	0.68
Ink	0.82
Bubbles	0.9
Dust	0.81
Pen marker	0.97
Out-of-focus	0.74
Artifact-free	0.91

Uses and Directions

This algorithm was developed for research purposes only. The terms of usage for Grand Challenge apply to the uploaded slides. To use the algorithm, users need to request access, and only requests from verified users will be accepted.

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