A recently developed AI-based model for measuring renal cortical volume can easily and accurately predict renal function loss after kidney donation, according to a new study.
The model is expected to revolutionize the existing complex and time-consuming method of kidney evaluation and help older donors, in particular, make safer donation decisions.
Sang-Il Min, Professor of Transplantation and Vascular Surgery at Seoul National University College of Medicine,including Eun-Ah Cho, Professor, Chung-Ang University Hospital, Lee Ju-Han, Professor,Severance Hospital, and Jin-Sung Kim, Oncosoft announced the results of a multicenter retrospective cohort study using AI-based CT images to measure renal cortical volume and its association with renal function decline after kidney transplantation, based on data from 1074 living donors who underwent kidney donation surgery from 2010 to 2020.
Developing AI-Powered Automated Segmentation Models
Previous studies have reported a correlation between renal cortical volume and renal function, but existing measurement methods are complex and time-consuming, making them difficult to apply in clinical practice.
To solve this problem, the team developed an “AI-based automated segmentation model”. The AI model is designed to automatically measure the cortical volume of the kidney by analyzing CT images before donation. Since the renal cortex is an important part of the kidney,accurately measuring its volume is crucial for predicting kidney function.
To validate the accuracy of the model, the team compared the AI-measured cortical volumes with manual measurements.
In our validation, the accuracy of the AI model was very high, with a Dice similarity coefficient (a metric that evaluates the overlap between two images, with closer to 1 indicating more similarity) of 0.97 and a Hausdorff distance (which measures the maximum error between the predicted and actual boundaries, with smaller values indicating more accuracy) of 0.76 mm. This means that the AI model was able to measure the actual renal cortex volume very accurately.
Using the AI model, the team analyzed the association between renal cortical volume and post-donation kidney function(estimated glomerular filtration rate, eGFR), as measured by the AI model. The eGFR is a measure of the kidney's filtration capacity, with lower values indicating decreased kidney function.
Generalized additive models (GAMs) were used to analyze changes in renal function over time after donation.
The results showed that older donors (>60 years of age) tended to have a greater decline in kidney function after donation compared to younger donors (<60 years of age). Specifically, the decline in eGFR was statistically significantly greater in older donors (P = 0.041),indicating that older donors experienced more decline in kidney function.
However, donors with larger preoperative renal cortical volume tended to have less post-donation decline in renal function. This difference was statistically significant (p<0.001),especially in older donors.
This suggests that older donors with greater renal cortical volume may be better able to maintain renal function after donation.
The researchers emphasized that measuring renal cortical volume using AI could be an important marker for predicting loss of kidney function after donation.
Changes in kidney function after kidney donation.
The model, if incorporated into the donor screening and evaluation process, is expected to contribute significantly to improving the success rate of kidney transplant surgery and enhancing donor safety, he added.
“This study, which demonstrates the clinical utility of measuring renal cortical volume using artificial intelligence, is an important step forward in the evaluation and prognosis of kidney donors,” said Sang-Il Min, Prof. ”It is particularly significant because it provides a more precise way to predict renal function loss in elderly donors than previously possible, enabling safer donation decisions.”
The study was published online in the International Journal of Surgery (IF=12.5), the world's leading authority on surgery.
