[1965] Validation of Whole Slide Scanning for Use in Real-Time Clinical Frozen Section Consultation

Scott H Bradshaw, Eric Belanger, Manisha Lamba, Kien T Mai, Susan J Robertson, Harman Sekhon. Ottawa Hospital, Ottawa, ON, Canada

Background: Digital slide scanning offers the posibility for a tertiary care hospital to provide an intra-operative consultation service to satellite hospitals without requiring a staff pathologist to be physically present at the satellite hospital. Before such a service could be offered, however, slide scanning would need to be validated to ensure that the additional step does not introduce error in diagnosis or unacceptable delay. The goal of this study is to assess diagnostic accuracy on digital vs glass slides.
Design: 42 frozen section cases selected to represent the range of specimens requiring intra-operative consultation at our institution were digitally scanned. The specimens were chosen with bias toward difficult consults to increase the chance of uncovering potential issues with the slide scanning process. The digital and glass slides were reviewed by four pathologists, with a 30 day washout period between reviews. The distribution of cases is summarized in table 1.

Case Distribution
Case Type#
Margin, GI6
Margin, skin4
Margin, lung4
Margin, other2
Lymph node5
Neutrophil count4
liver met4
Dx, gyne3
Dx, other4
Bx, R/O malignancy3

Results: Table 2 shows all correct diagnoses, discrepancies, and major discrepancies for glass slides and digital images for all reviewers combined. The number of major errors was not significantly different between digital and glass slides. The total major and minor discrepancies trended toward better performance for the glass slides.

Comparison of errors between glass and digital slides
Correct diagnoses162156 
Major errors43 
Total errors6120.225

Conclusions: Analysis of the individual discrepancies shows that reviewers of digital slides were more likely to defer to permanent section or give a diagnosis of 'suspicious' rather than committing to a positive or negative diagnosis, and this tendency led to a greater frequency of minor discrepancies. This may reflect a true degradation in the digitization of the slide, or alternatively, it may be at least partially due to unfamiliarity with the new media and reluctance to commit to a definitive diagnosis. Three out of four reviewers noted discomfort in couting neutrophils using digitally scanned images. Interestingly, despite this, there were no discrepancies (major or minor) relating to cases of neutrophil counts. Future work is necessary to assess if the number of minor discrepancies changes after the reviewers acquire familiarity with the new media. In addition, the time needed to arrive at a diagnosis using digital material relative to glass slides requires assessment.
Category: Quality Assurance

Monday, March 4, 2013 1:15 PM

Proffered Papers: Section G, Monday Afternoon


Close Window