Researchers at Asan Medical Center have developed a new method that can identify any rejections after a kidney transplant more accurately and quickly by using artificial intelligence (AI).

From left. Professors Kim Nam-gook and Go Hyun-jung

Until now, pathologists have had to analyze the slide of the pathology taken from the kidney. Pathologists could only analyze a portion of the slides, however, as the process of analyzing all the slides was impossible.

To resolve the problem, the team, led by Professors Kim Nam-gook and Go Hyun-jung at the hospital, developed an AI that reads the pathological tissue slides and diagnoses antibody-mediated immune rejection after the kidney transplant surgery.

The team used 200 kidney pathology slides from kidney transplant patients who underwent surgery at AMC from 2009 to 2016 to train its AI system and used an additional 180 slides to validate its efficacy.

As a result, the team’s system showed a 90 percent accuracy compared to that made from pathologists.

Also, as the average reading time was only about 13 minutes, the team expects that after analyzing kidney tissue with its system an additional reading will reduce the likelihood of diagnostic errors and the time it takes to diagnose.

To determine the immune-competence of donor and recipient before the kidney transplantation, hospitals perform several tests such as histocompatibility in advance. However, there is no way to completely predict antibody-mediated immunity rejection, which is one of the rejection reactions after the transplantation.

Therefore, if the hospital suspects antibody-mediated immune rejection after surgery, they harvest the patient's kidney tissue and count the number of peritubular capillaries using a specific immunostaining technique.

If the number of stained capillaries is higher than a certain level, it is highly likely that the patient is suffering from kidney transplant rejection.

So far, the pathologist had to analyze the tissue hundreds with the naked eye. As there are so many capillaries, however, it took too long to look at and had the possibility of inaccurate readings.

Although hospitals use immunosuppressive drugs to prevent rejection, it is not possible to eliminate the possibility. Therefore, it is essential to go into additional treatment immediately.

“Based on the results of this study, we will be able to use the AI algorithm to diagnose rejection more quickly and accurately after kidney transplantation,” Professor Go said. “If appropriate treatment is applied early, the success rate of renal transplant surgery will become much higher.”

Professors Kim also said, “We have seen the possibility of developing more efficient and precise AI in the field of pathology with the study. Based on accumulated know-how, we plan to expand the scope of AI technology in the pathology field.”

Scientific Reports published the results of the study.

Copyright © KBR Unauthorized reproduction, redistribution prohibited