Subject-Specific Calculation of Left Atrial Appendage Blood-Borne Particle Residence Time Distribution in Atrial Fibrillation

Figures

• 

  Figure 1. (A) One representative 3D reconstructed left atrial appendage (LAA) surface geometry from our cohort from each of the four categories proposed by Di Biase et al. (2012). (B) Generic flow waveform over one cardiac cycle used to generate subject-specific waveforms based on subject's cardiac output and heart rate (adapted from Smiseth et al., 1999). To avoid tracers exiting left atrial inlets, the waveform is shifted such that the backflow occurs at the end of the cardiac cycle around t = 0.6 s. (C) Sample geometry with LAA filled with tracer (colored red). All inlets (four pulmonary veins), the outlet (mitral valve), and the location of LAA are displayed. (D) A representative left atrium and LAA geometry meshed with tetrahedrons. A section through the appendage region shows a finer mesh inside the appendage. Furthermore, it shows the use of prismatic layers at the wall boundary.

• 

  Figure 2. (A) Normalized residual error, RE(i), plotted as a function of the number of principal components (i) used to reconstruct the LAA appearance for three subjects. As i increases, RE(i) decreases, reaching a value of zero when i = 16, corresponding to a perfect reconstruction. LAA appearance complexity index (LAA-ACI) is defined as the area the area under the RE(i)–i curve; larger LAA-ACI corresponds to a more complex LAA appearance. (B) Geometrical features of these three LAAs, including LAA ostium, tip of the LAA, LAA lobes (shown by circles), and LAA centerline bend (shown by curved arrows). The rank ordering of these three LAAs based on the LAA-ACI (most complex to simplest) is subject #3, subject #1, and subject #2.

• 

  Figure 3. (A) Spatially averaged LAA tracer concentration, C(t), plotted as a function of time for a representative LAA. Inset: Tracer concentration contours for selected time, illustrating the tracer washout from most of the LAA, except for the tip of the LAA. (B) E(t), the residence time distribution function quantifying the normalized rate of tracer washout across the LAA ostium, as a function of time for three representative subjects. Data for the first 10 s are shown to highlight the early washout. (C) Box plots showing the mean residence time, tm, for each of the four traditional LAA shape groups. There is large variability in tm within each traditional LAA shape group.

• 

  Figure 4. (A) Relationship between the two indices of LAA residence time distribution (RTD) function: mean residence time, tm, and aymptotic LAA concentration, C∞. The Spearman rank correlation analysis indicates that LAA tm and C∞ are highly correlated, suggesting that only one of these indices is sufficent to characterizie the LAA RTD function. The three representative subjects shown in Figures 2, 3 are identified. (B) Relationship between LAA appearance complexity index (LAA-ACI) and LAA tm. The Spearman rank correlation analysis indicates a weak (although statistically significant) correlation, suggesting that these two variables do not provide the same infoamtion. (C) Relationship between LAA-ACI and CHA2DS2-VASc score, analyzed using Spearman rank correlation analysis, showing a weak and insignificant correlation. (D) Relationship between LAA tm and CHA2DS2-VASc score, analyzed using Spearman rank correlation analysis, showing a weak and insignificant correlation.