Related Experiment Video
Updated: Sep 27, 2025
Use of Ultra-high Field MRI in Small Rodent Models of Polycystic Kidney Disease for In Vivo Phenotyping and Drug Monitoring
Published on: June 23, 2015
Deployed Deep Learning Kidney Segmentation for Polycystic Kidney Disease MRI.
Akshay Goel1, George Shih1, Sadjad Riyahi1
1Departments of Radiology (A.G., G.S., S.R., S.J., H.D., R.H., D.R., K.T., M.R.P.), Internal Medicine (J.D.B., I.B., I.C.), and Pathology and Laboratory Medicine (H.R.), Weill Cornell Medicine, 525 E 68th St, New York, NY 10021.
This study introduces an AI tool for precise automated total kidney volume measurement in autosomal dominant polycystic kidney disease (ADPKD) patients using MRI scans. The deep learning model significantly reduces expert time for kidney segmentation, improving disease monitoring.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Nephrology
Background:
- Autosomal dominant polycystic kidney disease (ADPKD) is a genetic disorder characterized by kidney cyst development.
- Total kidney volume (TKV) is a crucial biomarker for assessing ADPKD disease severity and progression.
- Manual measurement of TKV from MRI is time-consuming and subject to inter-observer variability.
Purpose of the Study:
- To develop, validate, and deploy a deep learning model for automated TKV measurement in ADPKD patients.
- To assess the accuracy and efficiency of the AI model compared to manual segmentation.
- To evaluate the model's performance on internal and external datasets.
Main Methods:
- A U-Net architecture with an EfficientNet encoder was utilized for deep learning-based segmentation.
- The model was trained and validated on 213 T2-weighted MRI studies from 129 ADPKD patients.
- Performance was evaluated using Dice Similarity Coefficient (DSC) and Bland-Altman analysis, with external and prospective validation.
Main Results:
- High accuracy was achieved with external validation (DSC 0.98) and prospective validation (DSC 0.97).
- Bland-Altman analysis showed minimal bias in TKV estimation (2.6% and 3.6% differences).
- The AI model reduced expert contouring time by 51% in prospective cases.
Conclusions:
- The deployed AI pipeline accurately segments polycystic kidneys for TKV estimation.
- Automated segmentation using deep learning significantly reduces the time required for expert contouring.
- This AI tool offers a promising solution for efficient and accurate ADPKD monitoring.
