Liver Steatosis Assessment: Correlations among Pathology, Radiology, Clinical Data and Automated Image Analysis Software
Michael Lee, Pelin Bagci, Jun Kong, Miriam Vos, Volkan Adsay, Puneet Sharma, Diego Martin, Alton Farris. Emory University, Atlanta, GA
Background: Assessing the degree of hepatic steatosis plays an important role in many liver diseases and during orthotopic liver transplantation. Macrovesicular steatosis is associated with post-transplant dysfunction, directly proportional to the level of steatosis. However, estimating the degree of steatosis is prone to both intra- and inter-observer variability among pathologists. Image analysis and computerized algorithms have been used with varying degrees of success.
Design: We compared two computerized techniques and contrasted these with radiology measurements (by MRI), and visual steatosis-scoring methods by three pathologists on frozen section slides, permanent slides, and whole slide images (WSIs). The two computerized methods applied to WSIs included a commercial positive pixel count algorithm tuned to quantitate white space and a custom algorithm programmed at our institution. The study cohort included pediatric patients (n=10), adult patients (n=13), and transplant biopsies with corresponding frozen section (n=18) and permanent slides (n=18).
Results: Correlations were observed among measurement methods and clinical parameters with the strongest correlations observed among the pediatric cohort. For all patients, there was a statistically significant correlation between pathology, radiology and each image analysis modality (r = 0.75-0.97, p < 0.0001). A statistically significant relationship was observed in all patients between each method, body mass index (BMI) and albumin but not with alanine aminotransferase (ALT) or aspartate aminotransferase (AST).
Conclusions: The assessment of hepatic steatosis is susceptible to intra- and inter-observer variability among pathologists. Our study demonstrates the utility of image analysis in providing an objective measurement of hepatic steatosis with correlations between qualitative image analysis, pathologist assessments, radiology measurements and several clinical parameters.
Wednesday, March 21, 2012 9:30 AM
Poster Session V # 236, Wednesday Morning