Abstract: SA-PO0607
TraceOrg: Automatically Measuring Kidney, Liver, and Cyst Volumes in ADPKD
Session Information
- Cystic Kidney Diseases: Clinical Research
November 08, 2025 | Location: Exhibit Hall, Convention Center
Abstract Time: 10:00 AM - 12:00 PM
Category: Genetic Diseases of the Kidneys
- 1201 Genetic Diseases of the Kidneys: Monogenic Kidney Diseases
Authors
- Xiong, Qing, Weill Cornell Medicine, New York, New York, United States
- He, Xinzi, Cornell University, Ithaca, New York, United States
- Scalco, Elisa, Consiglio Nazionale delle Ricerche, Rome, Lazio, Italy
- Pasini, Siria, Istituto di Ricerche Farmacologiche Mario Negri, Milan, Lombardy, Italy
- Zhu, Chenglin, Weill Cornell Medicine, New York, New York, United States
- Moghadam, Mina C., Cornell University, Ithaca, New York, United States
- Sattar, Usama, Weill Cornell Medicine, New York, New York, United States
- Davoudi, Vahid, Weill Cornell Medicine, New York, New York, United States
- Bazojoo, Vahid, Weill Cornell Medicine, New York, New York, United States
- Hu, Zhongxiu, Weill Cornell Medicine, New York, New York, United States
- Shih, Sophie Suzuki, Weill Cornell Medicine, New York, New York, United States
- Blumenfeld, Jon D., Rogosin Institute, New York, New York, United States
- Min, Robert J, Weill Cornell Medicine, New York, New York, United States
- Sabuncu, Mert, Cornell University, Ithaca, New York, United States
- Caroli, Anna, Istituto di Ricerche Farmacologiche Mario Negri, Milan, Lombardy, Italy
- Prince, Martin R., Weill Cornell Medicine, New York, New York, United States
Background
Kidney, liver and cyst volumes are necessary for diagnosis, classification and management of autosomal dominant polycystic kidney disease (ADPKD) but challenging to measure accurately and reproducibly. We developed a web-based deep learning platform to automatically and robustly measure kidneys, liver and cyst volumes in ADPKD.
Methods
MRI and CT scans from ADPKD patients (n=400) and subjects without ADPKD (n=257) were annotated to train a 3D hybrid model combining U-Net and transformer elements for segmenting kidneys, liver and cysts. Implemented as a web-based calculator at www.traceorg.com, this model produces labels, volumes and Mayo Clinic Image Classification. Automatic browser anonymization of DICOM images ensures protection of personal information. Internal validation was conducted on 20 MRIs for kidney and liver segmentations, 20 MRIs for cyst segmentations and it was compared to existing segmentation models (TotalSegmentator, MRAnnotator and Kline et al. Model for cysts). External validation was performed on one single-center dataset (n=58), one multicenter dataset (n=73), and CRISP MRIs (n=26).
Results
TraceOrg internal validation performance achieved high mean Dice scores of 0.99 (kidneys), 0.98 (liver), 0.87 (renal cysts) and 0.82 (liver cysts) outperforming existing models for ADPKD subjects. External validation showed strong performance with Dice scores of 0.92 (kidney), 0.88 (liver), 0.83 (renal cysts) and 0.93 (liver cysts) for the single-center and 0.95 (kidney), 0.81 (renal cysts) for the multicenter dataset. Compared to CRISP stereology volumes, the mean absolute percent difference was 8.8% (kidneys), 12% (renal cysts) and 6.6% (liver).
Conclusion
TraceOrg automatically measures kidney, liver and cyst volumes from abdominal scans in ADPKD and is publicly available as a web-based tool. Future work will extend this technology to additional organs, tissues, imaging modalities, and longitudinal TKV tracking.
Funding
- Other NIH Support