Back
For best output, select "Paper Size" as "A4" and "Margin" as "0" or "None".
To save or print to PDF, please select Print Destination > Save as PDF, enable Background Graphics under "More Settings", then click "Save".
During the congress, E-Posters will be accessible to all participants on the congress website 24/7, as well as in the E-poster stations in the congress center.
Preparing your E-Poster
Please review the E-Poster format requirements carefully when preparing your E-Poster. Should your E-Poster not meet the mentioned requirements, it may not be displayed as described above.
E-Poster Submission Deadline
Please prepare and upload your E-Poster no later than March 14, 2026 11.59PM CET. After this date, you will no longer be able to prepare and upload your E-poster and it will not be displayed and accessible on the congress website.
Please follow the instructions below to input your abstract title.
Abstract titles should be brief and reflect the content of the abstract.
Renal ultrasound visualizes cortical and medullary structures and is widely used worldwide, including in low-resource settings, because it is noninvasive, low-cost, and portable. In chronic kidney disease (CKD), beyond size reduction, qualitative morphological changes such as irregular renal contour and loss of corticomedullary differentiation are seen, but standardized, objective quantification has not been established. CKD is often regarded as pathologically accelerated aging of the kidney, and many of its morphological features resemble age-related changes. Building on this premise, we used deep learning to quantify such changes as “kidney age” and examined its biological and clinical validity.
We curated 142,401 abdominal ultrasound exams from 27,109 individuals (2007–2024) at Science Tokyo Hospital, each paired with an eGFR within ±3 months, and extracted long-axis kidney images. To avoid patient-level leakage, we split the data by patient into a development cohort (training/validation/test; used for model training and tuning) and an independent clinical evaluation cohort of 5,498 patients (nephrology outpatients, biopsy recipients, and patients with serial ultrasounds) used for pathology, prognosis, and treatment-response analyses. We trained an ImageNet-pretrained EfficientNet regression model with a single long-axis renal ultrasound image as input and the chronological age at the ultrasound date (years) as the target. At inference, the model output—the age predicted from an image—is referred to as “kidney age.” We defined ΔAge = kidney age − chronological age and tested its associations with renal pathology, kidney outcomes, and treatment response.
On the test set, age estimation achieved R²=0.528 with a mean absolute error of 7.0 years. Occlusion sensitivity indicated that predictions mainly relied on the renal cortex, consistent with known cortical predominance of aging-related changes. In cases with biopsy shortly after ultrasound, ΔAge correlated with interstitial fibrosis grade, supporting biological validity. During follow-up, greater ΔAge predicted faster eGFR decline; in multivariable Cox models adjusted for sex, age, diabetes, hypertension, proteinuria, and baseline eGFR—with the event defined as a 30% eGFR decline—ΔAge remained an independent predictor of renal outcomes. In the IgA nephropathy subset, higher ΔAge was associated with less reduction in proteinuria after steroid therapy. Estimates were consistent across planes, time points, devices, and operators, with an intraclass correlation coefficient of 0.99.
AI-derived kidney age and ΔAge from renal ultrasound align with pathology, predict kidney function decline, and help stratify treatment response. Given their high reproducibility, these measures may serve as noninvasive digital biomarkers suitable for clinical use.
