Abstract: TH-PO788
Radiomic Texture Features in CT Images of Kidneys in Ventilated Deceased Donors Predict Delayed Graft Function
Session Information
- Transplantation: Clinical - 2
October 24, 2024 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Transplantation
- 2102 Transplantation: Clinical
Authors
- Ali, Fayzan, Saint Louis University School of Medicine, St Louis, Missouri, United States
- Baldelomar, Edwin, Washington University in St Louis School of Medicine Mallinckrodt Institute of Radiology, Saint Louis, Missouri, United States
- Charlton, Jennifer R., University of Virginia Department of Pediatrics, Charlottesville, Virginia, United States
- Wahl, Richard L., Washington University in St Louis School of Medicine Mallinckrodt Institute of Radiology, Saint Louis, Missouri, United States
- Marklin, Gary F., Mid-America Transplant, St. Louis, Missouri, United States
- Bennett, Kevin M., Washington University in St Louis School of Medicine Mallinckrodt Institute of Radiology, Saint Louis, Missouri, United States
Background
Kidney transplants are a life-saving resource for ~25,000 patients each year. 80% of allografts come from deceased donors. The success of kidney transplants is affected by variable organ quality and recipient factors. However, there is no method to reliably predict which kidneys will be successful. We investigated the use of contrast-enhanced Xray-CT images of the kidneys in ventilated deceased donors, acquired during assessment for cardiovascular function. We hypothesized that radiomic features from CT images of deceased donors under ventilation, scanned before the kidneys are removed, associate with transplant outcomes measured by delayed graft function (DGF) after transplant in recipients.
Methods
We acquired CT images of 102 (65M, 37F) ventilated, brain-dead donors at a single organ procurement organization (1/2019-6/2019). Scans were acquired with a Siemens 64-slice Somatom and Optiray 350 contrast. The primary outcome was DGF defined as need for dialysis within a week after transplant, obtained from the Scientific Registry of Transplant Recipients. Donors of all transplanted kidneys were included. Donors whose kidneys were not transplanted were excluded. Texture features (n=64) were extracted from three image slices of each kidney using Lifex. ANOVA was performed with a Bonferroni correction to determine significant prediction of DGF (p<0.01).
Results
Two image texture features, Grey Level Co-occurrence Matrix (GLCM) Contrast and GLCM Cluster Prominence, predicted DGF. These features measure heterogeneity and may reflect microvascular ischemia. Significantly higher GLCM Contrast and Cluster Prominence were observed in both cortex and medulla in kidneys with DGF (p<0.01).
Conclusion
Two radiomic features in CT images of kidneys in ventilated deceased donors were highly predictive of DGF. With further validation, this approach could provide a sensitive, individualized measure of deceased donor kidneys to improve functional assessment resulting in transplantation. It could also enable studies of optimal donor management to improve allograft function.
Funding
- Private Foundation Support