Contribution of Digital Pathology on the Determination of Liver Steatosis Ratio

Objective: The rate of steatosis is the parameter that should be assessed in steatohepatitis and donor liver biopsies before transplantation. The percentage of steatosis, evaluated as quantitative / semiquantitative, differs between observers and intra-observers. Various image analysis methods have been developed in order to reduce this difference. In our study, pathologists examined virtually created digitized images mimicking liver fat. The results' accuracy rates and differences in assessment, that would impact clinical management, were investigated. Methods: Nine pictures with white circles on a pink background, simulating steatosis, were prepared with the 'Kameram' program. The steatosis area was calculated by computer. These pictures were shared with the Istanbul Hepatopancreatobiliary pathology study group. Participants were asked to specify the percentage of steatosis for each image. The results were compared to computer values. Results: Nineteen pathologists participated in the evaluation. Fifteen of the pathologists indicated higher steatosis percentage than the computirised calculated values, either in all pictures or most of the pictures. Only 1 participant reported the same values with the computer. Difference between computer values and pathologists' values ranged from as high as 40% to as low as 20%. When the images were thought to represent the donor liver, only 3 pathologists succeeded in proper clinical management in all cases, while 11 of the pathologists had misdiagnosed clinical management in 1 case, 4 pathologists in 2 cases, 1 pathologist in 3 cases. When it was thought that the pictures represented the steatohepatitis evaluation, all pathologists correctly scored only 2 pictures and at least 2 pathologists gave high scores in 5 of the pictures. Inter-pathologist agreement was moderate to good with assessing steatohepatitis and donor liver (kappa values, respectively: 0.51 and 0.63). Conclusion: In our study, it is observed that pathologists generally evaluate the steatosis rate higher than real calculated values. This can affect the clinic, in both steatohepatitis scoring classification and graft management differently. Computer-aided automated programs, which are increasingly used in routine pathology to overcome the differences in pathologist evaluations and clinical manifestations, also can have a role in the detection of liver steatosis ratio.