Abstract: TH-PO632
Deep Learning Retinal Image Analysis for the Detection of CKD and Cardiovascular Risk Factors in the General Population
Session Information
- Hypertension and CVD: Epidemiology, Risk Factors, Prevention
November 03, 2022 | Location: Exhibit Hall, Orange County Convention Center‚ West Building
Abstract Time: 10:00 AM - 12:00 PM
Category: Hypertension and CVD
- 1501 Hypertension and CVD: Epidemiology‚ Risk Factors‚ and Prevention
Authors
- Sexton, Donal J., Irish Longitudinal Study on Ageing (TILDA), Trinity College Dublin, Dublin, Ireland
- Romero-Ortuno, Roman, FRAILMatics group, TILDA, Trinity College Dublin., Dublin, Ireland
- Knight, Silvin P., FRAILMatics group, TILDA, Trinity College Dublin., Dublin, Ireland
- Dahyot, Rozenn, Maynooth University, Maynooth, Kildare, Ireland
- Kenny, Rose Anne M., Irish Longitudinal Study on Ageing (TILDA), Trinity College Dublin, Dublin, Ireland
- Karaali, Ali, Irish Longitudinal Study on Ageing (TILDA), Trinity College Dublin, Dublin, Ireland
Background
Retinal blood vessel patterns provide an opportunity to personalize an individuals risk assessment for CKD and cardiovascular risk factors (CVRF). In this study we propose a deep learning (DL) based prediction tool that uses retinal images from the Irish Longitudinal Study on Ageing (TILDA) to detect the existence of CKD in community dwelling individuals aged 50 years and over.
Methods
TILDA is a stratified random sample of the general population of Ireland. N=4569 participants underwent a detailed health assessment including retinal photography. We developed a convolutional neural network architecture inputting a single retinal image per participant for the prediction of CKD & CVRF. Binary cross entropy was used as a loss function. Analyses were conducted on the FRAILMatics HPC “Tinney”.
Results
See Table 1 & Image 1 for results.
Conclusion
A DL retinal image algorithm has good discrimination for CKD, eGFR and CVRF in community dwelling individuals. The prediction emphasis of our DL algorithm focuses on slightly different structures within the retinal image to predict serum creatinine versus serum cystatin.
Results of Deep Learning Retinal Image Analysis for CKD
| Classification Neural Network (Image only) | AUC |
| CKD by Serum Creatinine | 0.68 |
| CKD by Serum Cystatin C | 0.71 |
| Regression Neural Network (image only) | MAE |
| Age | 6.84 yrs |
| Systolic BP | 16.55 mmHg |
| eGFR CKDEPI Creatinine | 13.9 ml/min |
| eGFR CKDEPI Cystatin C | 14.8 ml/min |
Attention Map demonstrating where within the retinal image the algorithm is focusing to predict CKD by Cystatin C. Red-Yellow-Blue in descending order of importance.