GrandQC: A comprehensive solution to quality control problem in digital pathology
2024

GrandQC: A Tool for Quality Control in Digital Pathology

Sample size: 318 publication 10 minutes Evidence: high

Author Information

Author(s): Weng Zhilong, Seper Alexander, Pryalukhin Alexey, Mairinger Fabian, Wickenhauser Claudia, Bauer Marcus, Glamann Lennert, Bläker Hendrik, Lingscheidt Thomas, Hulla Wolfgang, Jonigk Danny, Schallenberg Simon, Bychkov Andrey, Fukuoka Junya, Braun Martin, Schömig-Markiefka Birgid, Klein Sebastian, Thiel Andreas, Bozek Katarzyna, Netto George J., Quaas Alexander, Büttner Reinhard, Tolkach Yuri

Primary Institution: Institute of Pathology, University Hospital Cologne

Hypothesis

Can GrandQC improve the quality control of digital pathology slides?

Conclusion

GrandQC significantly enhances the performance of image analysis algorithms by providing high-precision tissue and artifact segmentation.

Supporting Evidence

  • GrandQC allows for high-precision tissue segmentation with a Dice score of 0.957.
  • The tool can segment tissue without artifacts with a Dice score of 0.919–0.938.
  • GrandQC was validated using a dataset of 600 whole-slide images from five different pathology departments.
  • Two experienced analysts graded tissue detection quality with an average score of 9.44 out of 10.
  • GrandQC can be used as a benchmark for assessing slide quality across different pathology departments.
  • The tool is open-sourced along with a large annotated test dataset for further research.
  • GrandQC improves the performance of downstream image analysis algorithms.
  • Artifact detection accuracy was validated with a dataset of 318 slides.

Takeaway

GrandQC is a tool that helps doctors check the quality of tissue slides, making sure they are clear and accurate for analysis.

Methodology

The study developed and validated a tool for quality control in digital pathology using deep learning algorithms for tissue and artifact detection.

Potential Biases

Some misclassifications were noted, particularly in highly pigmented tissues.

Limitations

GrandQC was not trained to work with immunohistochemistry slides and may not address staining variability.

Participant Demographics

Slides from 19 international pathology departments were used.

Digital Object Identifier (DOI)

10.1038/s41467-024-54769-y

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication