A renal vascular compartment segmentation method based on dynamic contrast-enhanced images.

Area of Science:

• Medical Imaging
• Renal Physiology
• Computational Anatomy

Background:

• Renography is crucial for kidney function assessment using compartment models.
• The vascular compartment is vital in two- and three-compartment models.
• Automated vascular compartment segmentation methods are lacking.

Purpose of the Study:

• To develop an automated method for vascular compartment segmentation in renography.
• To enhance the accuracy and reproducibility of kidney function analysis.

Main Methods:

• Utilized multi-phase scan images and reconstructed feature images.
• Employed time-density curve features of each voxel in contrast-enhanced images.
• Distinguished vascular space from other areas based on contrast uptake dynamics.

Main Results:

• Segmentation accuracy was validated using glomerular filtration rate (GFR) analysis.
• The Patlak-Rutland technique within a two-compartment model was used for GFR evaluation.
• A high correlation (0.919, P<0.001) was observed between reference and model-derived GFR in 11 subjects.

Conclusions:

• The automated method eliminates subjective phase selection inherent in manual segmentation.
• This approach ensures consistent and reproducible segmentation results for kidney data.
• Improved vascular segmentation aids in more reliable kidney function assessment.