*Purpose: An accurate estimation of liver fat content is necessary to predict how a donor liver will function after transplantation. This can be challenging at the many hospitals that lack liver pathology expertise. Therefore, we developed a low-cost, mobile artificial intelligence (AI) platform to evaluate liver fat content on donor liver biopsy slides in real-time.

*Methods: We loaded our previously described AI algorithm that labels and quantifies fat globules onto a microcomputer with a GPU. The computer was embedded into a 3D printed case and connected to a 12MP camera and display. The device is capable of acquiring an image of a liver biopsy through a microscope, labeling the fat globules, and outputting the percent steatosis. Three images were taken of each slide (Figure 1a).

*Results: The assessment of steatosis on 33 slides by the device was compared to the whole slide assessment using the same algorithm on a cloud-based platform previously described by our group. There was strong positive correlation (r=0.871) between the two techniques (Figure 1b).

*Conclusions: Conclusion: This all-in-one, point-of-care device could be used on donor procurements to obtain an objective assessment of liver fat content even when a trained pathologist is not available. The correlation between the cloud and device-determined assessment of steatosis is very good and might be improved by increasing the number of images taken and assessed by the device. Hopefully, such a device could streamline appropriate matching of donors and recipients.

« Back to 2022 American Transplant Congress