Diagnostic Accuracy and Efficiency of Whole Slide Digital Imaging in Medical Liver Disease
Margaret S Ryan, Maxwell L Smith, Philip J Boyer, James R Burton, Stephen S Raab. University of Colorado, Aurora, CO; Mayo Clinic Arizona, Scottsdale, AZ; Memorial University of Newfoundland, St. John's, NL, Canada
Background: Whole-slide digital imaging (WSDI) technology is beginning to transform the practice of surgical pathology. We evaluate the use of WSDI with simulated native and allograft liver biopsy cases in respect to diagnostic accuracy, efficiency, and clinical implications.
Design: We performed a case-control study in which a hepatic pathologist examined 50 liver core biopsies- 25 control specimens (using a light microscope) and 25 case specimens (using the Olympus VS110 virtual slide scanning system). These were consecutive cases diagnosed by that same pathologist two years previously. All cases included the initial clinical history and standard protocol panel of slides. The case and control groups were matched so that equivalent diagnoses were in each set. The pathologist rendered a diagnosis, confidence level (1= confident; 4= not confident), total time in slide examination, and limitations. The time to prepare cases for simulation was also recorded. A hepatologist then compared the original and review diagnosis to determine concordance.
Results: The number of non-concordant diagnoses in the case and control groups was 7 (2 major and 5 minimal) and 5 (1 major and 4 minimal), respectively. For the control group, the major error was a steatohepatitis case in which rare single necrotic hepatocytes were not recognized on glass slide review. For the case group, the two major errors were due to inadequate microscopic descriptions. The mean confidence level for the case and control group was 1.48 and 1.52, respectively. On average, cases re-diagnosed under the microscope took 4.4 minutes per case. The digital image slides uploaded slowly, leading to an average time of 5.48 minutes per case. The average time to scan a slide was 15 minutes, with 5 minutes of hands-on time per slide. 11% of the total slides were rescanned due to poor initial image quality.
Conclusions: Simulated medical liver disease cases using glass slides and WSDI show similar rates of correlation with original diagnoses, similar confidence levels, and a trend to longer time to completion using WSDI (not statistically significant). However, there was a significant difference in the time required in assembling the cases for simulation largely due to technical software difficulties and repeated scanning to achieve optimal images. For WSDI to continue its transformation of anatomic pathology practice, attention must be given to the technical/operational aspects of scanning.
Monday, March 19, 2012 9:30 AM
Poster Session I Stowell-Orbison/Surgical Pathology/Autopsy Awards Poster Session # 313, Monday Morning