Impact of (k,t) sampling on DCE MRI tracer kinetic parameter estimation in digital reference objects

Area of Science:

• Medical Imaging
• Biophysics
• Quantitative MRI

Background:

• Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is crucial for assessing tissue characteristics.
• Accurate estimation of tracer-kinetic (TK) parameters is vital for clinical interpretation.
• Data sampling strategies significantly impact the precision of TK parameter estimation.

Purpose of the Study:

• To evaluate the impact of (k,t) data sampling on the variance of TK parameter estimation in whole-brain DCE-MRI.
• To assess sampling strategies in the context of TK model constraints and without other constraints.
• To investigate the influence of undersampling on TK estimation accuracy.

Main Methods:

• Generated three realistic brain-tumor digital reference objects.
• Employed various sampling strategies: uniform, variable density, zone-based, lattice, pseudo-random, and pseudo-radial.
• Performed 4-fold to 25-fold undersampling with 50 time frames, assuming a fully sampled first frame and known arterial input function.
• Estimated TK parameters using indirect and direct methods, evaluated via Cramér-Rao bound and Monte-Carlo simulations across relevant signal-to-noise ratios.

Main Results:

• Lattice-based sampling demonstrated the lowest standard deviations (SDs) for TK parameters, followed by pseudo-random, pseudo-radial, and zone-based strategies.
• Pseudo-random sampling yielded 19% higher averaged SDs than lattice-based sampling.
• Zone-based sampling showed substantially increased SDs at undersampling factors exceeding 10.
• Cramér-Rao bound analysis indicated minimal differences between uniform and variable density sampling for lattice and pseudo-random schemes up to 25-fold undersampling.

Conclusions:

• Lattice sampling generally provided the lowest SDs in TK parameter estimation.
• Differences between sampling schemes were not substantial at low undersampling factors.
• The variations between lattice-based and pseudo-random sampling strategies were within the error range from other sources, even at 25-fold undersampling.