A new approach assessing the quality of whole slide images by discovering specific issues [post]

Ying Gao, Yuchen Song, Changhong Liang, Zaiyi Liu
2020 unpublished
Background : Since whole slide images (WSIs) are widely used in clinical diagnosis, assessing the quality of these images has generated considerable interest. We divided the evaluation process into different sections focusing on specific issues and developed an approach based on computer technology to assess the quality of WSIs and provide details corresponding to specific issues.Methods : We focused on three issues in this work: position deviation, cover slip misplacement and wrong focus. Our
more » ... pproach revolves around these issues. We collected two datasets from Guangdong Provincial People's Hospital to develop the approach, discovered three issues and provided details corresponding to these issues.Results : We designed experiments based on two datasets to validate the effectiveness of the approach. The experiments show that using our approach, existing issues can be discovered and corresponding details are provided at a high accuracy. In the training set, the accuracy of the discovered three issues reaches 0.998, 0.987 and 0.988. In the validation set, the accuracy of discovering all issues reaches 0.914.Conclusions : We proposed an approach focusing on the following specific image issues: position deviation, cover slip misplacement and wrong focus. Using this approach, issues were discovered with high accuracy, and reports were produced that provide details regarding existing issues discovered by our approach. However, a dataset of WSIs with more issues, including stain concentration and poor microtomy, should be established, and methods discovering these issues must be developed in the future.
doi:10.21203/rs.3.rs-20180/v1 fatcat:an45wdnvo5eq5motqwekbygdte