Structural and practical identifiability of contrast transport models for DCE-MRI

Area of Science:

• Medical Imaging
• Biophysics
• Pharmacokinetics

Background:

• Compartment models are essential for analyzing dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data.
• These models quantify blood flow and transport, offering diagnostic and prognostic insights.
• Ensuring accurate and repeatable model parameter estimation is critical for reliable analysis.

Approach:

• This study investigates the structural and practical identifiability of nested compartment models commonly used in DCE-MRI analysis.
• Both artificial and real DCE-MRI data were utilized to assess the influence of noise on parameter estimation.
• The impact of increasing data noise on parameter identifiability was systematically analyzed.

Key Points:

• All analyzed compartment models were found to be structurally identifiable.
• Practical identifiability was demonstrated to be highly dependent on data characteristics, particularly noise levels.
• Parameter identifiability can be significantly improved by enhancing data quality.

Conclusions:

• The identifiability of compartment models in DCE-MRI is robust to specific tissue types and contrast agent enhancement patterns.
• Data quality is a paramount factor influencing the practical identifiability of model parameters.
• Strategies to improve data quality are essential for reliable quantitative analysis of DCE-MRI data.