LAUSR

The growing imbalance between the need and the availability of liver donors has resulted in an increasing use of extended liver criteria donors, which are more likely to have hepatic steatosis (HS) with a risk of early graft dysfunction in case of severe ( 67%) macrovesicular steatosis. Although accurate evaluation of HS is crucial for optimal donor selection, the diagnostic performance of noninvasive criteria for liver steatosis using clinical, biological, and morphological criteria remains limited and prevents a reliable assessment in daily practice. On the opposite, frozen section analysis of liver biopsy remains the gold standard to evaluate HS, but its systematic use is not realistic in the very specific organizational setting of organ procurement. Altogether, the assessment of the degree of HS currently relies on visual inspection of the liver, often by a junior surgeon, and may lead to inappropriate liver graft selection. In this context, the development of an easy-to-use, handy, and accurate device for noninvasive quantification of HS would obviously be most desirable. As smartphones have become ubiquitous in modern culture and now provide high-quality pictures, a relevant use would be to obtain standardized images with instant processing by a dedicated application that would automatically evaluate liver steatosis. However, the feasibility of correlating imaging data obtained by smartphone cameras during liver procurement with HS as assessed by histological examination still remains to be assessed. In the present pilot study, all grafts procured by the Beaujon Hospital liver procurement team between January 2017 and February 2017 were included. These grafts all underwent biopsy and pictures. The study protocol received a priori approval by the appropriate institutional review committee. First, HS was estimated on triangular frozen section biopsies harvested on the left lobe during liver procurement (in case of suspected severe HS) or after reperfusion during liver transplantation as a routine procedure at our institution. Steatosis was assessed based on the percentage of hepatocytes with macrovesicular steatosis using a quantitative scale up to 100%. According to the widely used scoring approach, HS was categorized as follows: normal (grade 0) macrovesicular steatosis from 0% to 5%; mild (grade 1) macrovesicular steatosis from 6% to 33%; moderate (grade 2) macrovesicular steatosis from 34% to 66%; and severe (grade 3) macrovesicular steatosis >67%. Second, digital pictures of the livers were taken during liver procurement using the same commercially available smartphone (iPhone 6S; Apple Inc., Cupertino, CA). Exposure, white-balancing, and focus area selection modes were commonly nonadjustable. Pictures of the livers were taken in situ before aortic cross-clamp with the highest resolution. The graft was photographed from a distance of approximately 30 cm with automatic flash light; the operating lights were turned away from the graft and maximal externalization of the liver from the abdominal cavity was applied. In the end, only nonblurred, correctly exposed and balanced, perpendicularly axed pictures of the liver without reflection were selected for further analysis. A set of images was then created by manually cropping the original photographs so that the target organ occupies 85%-100% of the frame. A team of dedicated engineers then processed the images using the CaptureProof software (ComputerVision Technology, Abbreviations: ALT, alanine aminotransferase; AST, aspartate aminotransferase; Cs, Spearman’s correlation; BMI, body mass index; GGT, gamma-glutamyltransferase; HS, hepatic steatosis; ICU, intensive care unit.