Developing a Validated Tool for Evaluating Whole Slide Images
Dana Grzybicki MD; University of Pittsburgh School of Medicine; Robb Wilson MA; University of Pittsburgh School of Medicine; Leslie Anthony MA; University of Pittsburgh Medical Center; Russell Silowash BS; University of Pittsburgh School of Medicine;
Content:
Whole slide imaging (WSI) technologies are becoming increasingly popular within pathology practices. Since their introduction, several vendors have produced whole slide scanners with a variety of scanning and viewing features. With the deployment of such devices in hospitals, it is essential to acquire images that are diagnostically equivalent to glass slides in terms of quality. There are, however, no validated image quality assessments mentioned in the pathology literature for rating the quality of whole slide images. One of the goals of the IMITS Telepathology Projects evaluation team is to establish such an assessment tool.
Technology:
N/A
Design:
Thirty difficult prostate biopsy foci were analyzed using WSI and light microscopy. An assessment form was developed that captured rater diagnosis, image quality, case complexity, diagnostic confidence, and time to complete case. Five volunteer pathologists completed the assessment for each case during WSI and light microscopy analyses. Internal, external, and construct validity were measured using Spearmans coefficient. Internal validity was determined by examining associations between the numbers of glass or virtual slides examined and time to complete each case. External validity was determined by examining associations between quality ratings assigned to corresponding glass and virtual slides. Construct validity was determined by examining the association between case complexity and diagnostic confidence ratings.
Results:
In regards to internal validity, a statistically significant correlation was present for only one of the 5 subject pathologists. External validity was revealed to be significant for 3 of the 5 pathologists. Construct validity was measured using the association between subjects assessments of case complexity and their diagnostic confidence for each phase of the study. Correlations were statistically significant for 3 of the 5 pathologists.
Conclusion:
Despite having low internal validity, this assessment tool shows promising results for external and construct validity. In order to continue our work validating this tool, we will modify the internal validity parameters to reflect our findings and concentrate more fully on focusing details regarding specific tissue elements (i.e. nuclear, cytoplasmic, and vascular detail). Additional WSI research conducted with UPMC and US Air Force pathologists will be used to further validate this instrument.
