Design

Digital Pathology
Sergio Serrano Figueras
Hospital del Mar-IMIM-UAB, Barcelona
1620 Successful
Validation of Virtual
Microscopy for Surgical Pathology but Not
Cytopathology: Application of the College
of American Patholgy (CAP) Guidelines
MA Arnold, EA Chevener, KKNicol, T Barr. Nationwide
Children’s Hospital, Columbus, OH
Design
• Surgical Pathology (SP): 60 cases
• Cytopathology (CP): 21 cases
• Number of slides: 627
• Number of pathologists: 9
• Cytospin slides and small biopsies captured at
40X, remaining at 20X
• Evaluation of the glass slides then evaluation
of the virtual slides
Results
• Major discrepancies in SP cases: 1.7%
• Major discrepancies in CP cases: 25%
Conclusions
Pediatric SP specimens can be adequately
reviewed when using virtual microscopy.
However, review of cytology specimens require
improved resolution, perhaps including image
capture of multiple focus planes
1629
Impact
of
Configuration
on
Diagnosis
Monitor
Display
Digital
Pathology
IC Cucoranu, AW Parwani, C Mello-Thomas, SE Monaco,
WE Khalbuss, J Duboy I Ahmed, K Espig, A Xthona, L
Pantanowitz. University of Pittsburgh Medical Center,
Pittsburgh, PA; University of Sydney, Sydney, Australia;
Barco Healthcare, Beaverton, Oregon
Design
• 100 pathology glass slides (SP and CP)
• 3 pathologists
• 3 6MP (30 inch) Barco displays
-Calibration custom or native
-Luminosity high or low
3 2MP (24 inch) Barco displays
-Calibration custom or native
-Luminosity high or low
• Evaluation of the virtual slides then evaluation of
the glass slides
Results
• The 24 inch/2MP display performed significantly
better than the 30 inch/6MP display
• Native calibration provided better results than the
custom calibration
• Luminosity did not have a significant impact
Conclusions
Monitor size and calibration can impact the
detection of morphologic parameters necessary to
render diagnosis. The best performance was
obtained when using a 24 inch monitor with native
calibration
1639 Use of Whole-Slide Imaging
System for Frozen Section Diagnosis:
Comparative study between Virtual Slide
and Glass Slide Interpretation
KK Khurana, A-LA Katzenstein, S Wojcik, LM
Drogowski, N Drotar, O El-Zammar, S Landas, R
Corona, V Baxi, M Montalo. SUNY Upstate Medical
University, Syracuse, NY; Omnyx. LLC, Pittsburgh, PA.
Design
• 304 Frozen sections
• 17 Touch preparations
• 10 Squash preparations
• 3 Smear slides
• 4 pathologists
• Evaluation of the virtual slides then evaluation of
the glass slides.
Results
•Scan time per case: 9.1± 6.1min.
•Read time per case.
-Virtual slides: 2.1 min. (1.4-2.4)
-Glass slides: 4.1 min.
•Concordance on categorical interpretation: 100%
Conclusions
The use of virtual slides for frozen section (FS)
diagnosis compares favorably to conventional glass
slide interpretation. The modest increase in FS
turnaround time due to scan time may be partially
offset by reduced review time needed by
pathologists.
1643 Whole-Slide Imaging in the Routine
Diagnosis in Gynecological Pathology
J Ordi, P Castillo, M del Pino, O Ordi, L RodriguezCarunchio, R Millan, C Garcia, J Ramirez. Hospital
Clinic- CRESIB, Barcelona, Spain
Design
• 351 cases (966 glas slides)
• One pathologists analized conventional glass
slides
• One pathologists analized virtual slides obtained
by scanning glass slides at 20X
• Discrepancies were classified, according to a
modified Goldman classification, as major or
minor
• A statistical analysis was performed
Results
• Major discrepancies were observed in 2% of the
biopsies (underdiagnosed or missed lesions in
virtual slides)
• Minor discrepancies were observed in 4.3% of the
biopsies
• Interobserver agreement was at the almost perfect
level (Kappa 0.904, 95%CI; 0.863-0.945)
• Mean time used by technicians: 25 seconds per
slide
• Percentage of slides requiring reescanning: 4.16%
• Rate of scanning failure: 0.61%
Conclusions
Diagnosis of gynecological specimens by
virtual slides is accurate. Routine diagnosis
and digital archiving of gynecological
specimens by whole slide imaging may be
introduced in departments of pathology
1644Whole Slide Imaging Access and
Retrieval
HC Tsang, CS Friedman, L Rivera, K Turner, V Brodsky.
Weill Cornell Medical College, New York, NY.
Design
• Slides scanned during a 14-month period
-Clinical consult cases
-Educational slides
-Research slides (including tumor micro-arrays)
• Users were interviewed regarding their reasons for
viewing the images
Results
• % of slides stored on the server
- 76% Clinical consult cases
- 15% Educational slides
- 9% Research TMA slides
• % of the logon events
- 53% Research TMA slides
- 30% Educational slides
- 17% Clinical consult cases
Conclusions
Research slides were accessed at a higher rate
than other categories