"LiverColor": Machine Learning in Liver Photographs

"LiverColor": AN ALGORITHM QUANTIFICATION OF LIVER GRAFT STEATOSIS USING MACHINE LEARNING AND COLOR IMAGE PROCESSING

The main goal of this project is to create a machine learning model in order to quantify liver steatosis in liver donor faster, more objective and reliable than histological analysis and surgeons point-of-view.

Surgeons (junior and senior operators) from the HBP & Transplantation Unit took the pictures. They were taken after the laparotomy and before any type of surgical procedure. For each deceased donor case, a total of 5 pictures were taken: one for the left lobe and another for the right one before undergoing a surgical biopsy, two more (one for the left and one for the right lobe) after the histological analysis, near to the site of the surgical biopsy, and finally, one picture after liver perfusion.

Inclusion Criteria:

• Livers from donor donor brain death with informed consent before inclusion in the study was obtained from all participants or families.

Exclusion Criteria:

• Age < 18 years old
• Donor after cardiac death
• Split
• Cholestasis due to a biliary obstruction
• Total bilirubin levels above 2,5 mg/dL
• Glutamic oxaloacetic transaminase (SGOT)/ serum glutamatepyruvate transaminase (SGPT) levels and gamma-glutamyl transaminase (GGT) levels above 400 U/L
• Cirrhotic livers