Modeling Bias Error in 4D Flow MRI Velocity Measurements

Area of Science:

• Medical Imaging
• Fluid Dynamics
• Biomedical Engineering

Background:

• 4D flow MRI is crucial for cardiovascular research, but velocity measurements can be inaccurate.
• Bias errors arise from intra-voxel velocity variations and partial volume (PV) effects.
• Accurate quantification of these errors is needed to improve 4D flow MRI reliability.

Purpose of the Study:

• To develop and validate a model for estimating bias error in 4D flow MRI velocity measurements.
• To quantify the impact of intra-voxel velocity distribution and PV effects on measurement accuracy.
• To assess the model's performance in synthetic, in vitro, and in vivo scenarios.

Main Methods:

• A bias error model was developed using a 3D Taylor Series expansion for intra-voxel velocity and Richardson extrapolation for PV/numerical errors.
• The model was tested on synthetic Womersley flow, in vitro, and in vivo 4D flow MRI data from a cerebral aneurysm.
• Model validity requires a vessel diameter of at least 3.75 voxels and a signal-to-noise ratio > 5.

Main Results:

• The model accurately estimates bias error in non-PV voxels.
• In PV voxels, bias error is significantly higher (order of magnitude), with estimation accuracy between 67.3% and 108%.
• In vivo bias error doubled from diastole to systole, indicating cardiac cycle variation.

Conclusions:

• The developed bias error model effectively quantifies 4D flow MRI measurement accuracy.
• The model can aid in planning 4D flow MRI scans by predicting potential errors.
• Understanding and correcting bias error is essential for reliable cardiovascular flow analysis using 4D flow MRI.