Applying a Deep Learning Model for Total Kidney Volume Measurement in Autosomal Dominant Polycystic Kidney Disease

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Nephrology

Background:

• Autosomal dominant polycystic kidney disease (ADPKD) is a leading cause of hereditary end-stage renal disease.
• Total kidney volume (TKV) is a key metric for assessing ADPKD severity and prognosis.
• Manual TKV measurement is time-consuming, labor-intensive, and prone to errors.

Purpose of the Study:

• To develop and evaluate a deep learning model for automated TKV measurement using MRI data.
• To compare the performance of the deep learning model against manual measurements by medical professionals.
• To assess the impact of image orientation (axial vs. coronal) on TKV measurement accuracy.

Main Methods:

• Trained a U-net deep learning model on MRI data from 30 ADPKD patients and 10 healthy controls.
• Utilized both axial and coronal MRI image sections for TKV calculation.
• Extracted DICOM images, performed augmentation and labeling, and employed post-processing for TKV estimation.

Main Results:

• The deep learning model achieved TKV measurements comparable to medical professionals.
• The axial-section model demonstrated significantly lower measurement variability (3.95%) compared to the coronal-section model (21.6%).
• The axial-section model achieved higher accuracy, indicated by Dice Similarity Coefficient (0.89) and Jaccard coefficient (0.86).

Conclusions:

• Deep learning models can accurately measure TKV in ADPKD patients.
• Axial-section MRI images yield more reliable TKV measurements with the developed AI model.
• This AI approach offers a more efficient and potentially more accurate method for TKV assessment in ADPKD.