Improving Blood Flow Visualization of Recirculation Regions at Carotid Bulb in 4D Flow MRI Using Semi-Automatic Segmentation with ITK-SNAP

Figures

• 

  Figure 1. Flow chart of the hemodynamic analysis process of 4D flow MRI (first row) and CFD (second row). Segmentation using ITK-SNAP was included in the 4D flow MRI workflow.

• 

  Figure 2. The whole blood viscosity profiles for each subject according to shear rate. Subject-specific blood viscosity profiles are shown with different lines.

• 

  Figure 3. (a) Three analysis planes of left carotid artery from subject 1 (left panel). The velocity waveforms at the bifurcation planes provided by 4D flow MRI without ITK-SNAP (dashed line), 4D flow MRI with ITK-SNAP (red line), and CFD (blue line). Results from 4D flow MRI with ITK-SNAP and CFD revealed a close similarity of shape. (b) Three analysis planes of left carotid from subject 5 (left panel). The velocity waveforms at the bifurcation planes provided by 4D flow MRI without ITK-SNAP (dashed line), 4D flow MRI with ITK-SNAP (red line), and CFD (blue line). The velocity waveforms were variable among the 3 approaches.

• 

  Figure 4. Bland–Altman plot analysis of 14 carotid bifurcation velocity data (n = 606) comparing results of mean blood flow velocity over the cardiac cycle for all 4D flow MRI with ITK-SNAP and CFD analysis planes. The solid line indicates the mean of the difference between 4D flow MRI and CFD. The dashed lines represent the limits of agreement between the two methods, from −1.96 to +1.96 of standard deviation (limit range from −0.095 to 0.060 m/s).

• 

  Figure 5. The 3D blood flow characteristics at carotid bulb using 3 different approaches. (a) 4D flow MRI without ITK-SNAP) and (b) 4D flow MRI with ITK-SNAP: the visualization of recirculation blood flow at carotid bulb is clearly improved by using ITK-SNAP (white arrow labeled 1); (b) and (c) computational fluid dynamics: CFD): the major flow and secondary flow structures observed by 4D flow MRI with ITK-SNAP and CFD coincide (black arrow labeled 2).

• 

  Figure 6. (a) The 3D flow patterns for seven left carotid arteries at peak systole observed by 4D flow MRI with ITK-SNAP (first row) and CFD (second row). (b) The 3D flow patterns for seven right carotid arteries at peak systole observed by 4D flow MRI with ITK-SNAP (first row) and CFD (second row).

• 

  Figure 7. (a) The 3D vector fields derived from 4D flow MRI with ITK-SNAP (left) and CFD (right). Several streamlines were disconnected at the segment of an arterial wall within the carotid bifurcation domain in 4D flow MRI (white arrows labeled 1). Disagreement in secondary flow directions at the peripheral area of recirculation regions between the two techniques (white arrows labeled 2). (b) The 3D vector fields derived from 4D flow MRI with ITK-SNAP (left) and CFD (right). Generally, 4D flow MR images of secondary flow directions (black arrows) have similarities with CFD simulation.