Universal atrial coordinates applied to visualisation, registration and construction of patient specific meshes

Abstract

Integrating spatial information about atrial physiology and anatomy in a single patient from multimodal datasets, as well as generalizing these data across patients, requires a common coordinate system. In the atria, this is challenging due to the complexity and variability of the anatomy. We aimed to develop and validate a Universal Atrial Coordinate (UAC) system for the following applications: combination and assessment of multimodal data; comparison of spatial data across patients; 2D visualization; and construction of patient specific geometries to test mechanistic hypotheses. Left and right atrial LGE-MRI data were segmented and meshed. Two coordinates were calculated for each atrium by solving Laplace's equation, with boundary conditions assigned using five landmark points. The coordinate system was used to map spatial information between atrial meshes, including scalar fields measured using different mapping modalities, and atrial anatomic structures and fibre directions from a reference geometry. Average error in point transfer from a source mesh to a destination mesh and back again was less than 0.1 mm for the left atrium and 0.02 mm for the right atrium. Patient specific meshes were constructed using the coordinate system and phase singularity density maps from arrhythmia simulations were visualised in 2D. In conclusion, we have developed a universal atrial coordinate system allowing automatic registration of imaging and electroanatomic mapping data, 2D visualisation, and patient specific model creation.

Keywords: Atria; Coordinates; Mapping; Personalized computational modeling.

Graphical abstract

Fig. 1

Schematic: (A) The LA and RA were segmented from LGE-MRI data using MUSIC software. The colour bar indicates the number of standard deviations above the mean intensity of the blood pool for the LGE-MRI intensity values. (B) The endocardial segmentations were then meshed to create closed surface triangulations, and trimmed at the valves and veins (LA shown in blue; RA shown in red). (C) For each of the LA and RA surfaces, two atrial coordinates were defined. For the RA, these were a lateral-septal TV coordinate (α_RA, shown in septal-lateral view), and an IVC-SVC coordinate (β_RA, shown in lateral-septal view). For the LA, these were a septal-lateral coordinate (α_LA, shown in posteroanterior view), and a posterior-anterior coordinate (β_LA, shown in posteroanterior view). (D) These coordinates were used to map atrial structures from the original reference mesh to the target patient specific mesh. (E) Vector fibre data were also mapped between geometries. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Selection of five landmark points: Two points were chosen on the LA roof; one on the LA septum; and two on the RA lateral-septal boundary. The LA roof points were chosen at the junction of the LA body with the LSPV and RSPV, at the highest posterior location for the LA. The LA septal point was chosen to be just anterior of the FO location. The RA lateral-septal points were chosen in a similar way to the LA roof points; that is at the junction of the RA body with the IVC and SVC, at the highest lateral location for the RA.

Fig. 3

Calculation of Dirichlet boundary nodes used for Laplace solves. (A) Identify region boundaries at the RSPV-LA and LSPV-LA junctions and MV (black lines). (B) Calculate the lateral and septal paths, ensuring the LAA is on the anterior wall (shown in black, with the posterior wall in white and superior veins in green). The intersections of these paths with the RSPV-LA and LSPV-LA junctions are marked (by 1 and 4, respectively). (C) Calculate the geodesic path between the LSPV and RSPV user-defined markers (purple line). The intersections of this paths with the RSPV-LA and LSPV-LA junctions are marked (by 2 and 3, respectively). (D) Use the four points (intersection of lateral, septal and roof boundaries with LSPV-LA and RSPV-LA junctions; points 1–4) to divide the superior vein openings into anterior and posterior segments (purple and green, respectively). (E) Determine the anterior-posterior boundary choice (purple line) such that the superior veins are assigned as anterior (black region). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Laplace calculations and UACs. (A) The Laplace field generated by solving for boundaries of 0 at the TV, and 1 along the lateral-septal boundary. (B) The Laplace field generated by solving for boundaries of 0 at the IVC path, and 1 along the SVC path. (C) Mesh partitioned into lateral and septal regions. (D) The lateral portion of an isoline (value 0.7) for the Laplacian solve in (A) was used to rescale the coordinate in (B). (E) UACs for the RA. (F) UACs for the LA. Isolines are shown at 0.04 increments (except for A which is 0.08 spacing).

Fig. 5

Average error in point transfer does not depend on geometry surface area. Average error in μm for mapping from an atlas geometry to either the atlas (shown as crosses) or one of twelve different atrial geometries and back again is plotted against endocardial surface area for the LA (blue) and RA (red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 6

Mapping bipolar peak-to-peak voltage: (A) Peak-to-peak voltage measured using the Carto system. (B) Peak-to-peak voltage mapped from (A) using UACs to a mesh created from LGE-MRI for the same patient. (C) Peak-to-peak voltage mapped from (A) using an affine registration technique. Black regions indicate areas without recording values. The LAA is not included in the Carto geometry leading to a large area without data on the MRI.

Fig. 7

Patient specific geometries: (A) The original bilayer model displayed in posteroanterior and anteroposterior view. RA endocardial structures are displayed as epicardial, for visualisation purposes. (B) 2D UAC representation with regions labelled for the RA (left) and LA (right). (C) Meshes derived from MRI data for four patients.

Fig. 8

Fibre mapping: The original bilayer geometry is shown on the left column, with fibres for the: (A) LA epicardium, posteroanterior view; (B) LA epicardium, anteroposterior view; (C) CT and PM; (D) RA epicardial fibres.

Fig. 9

Comparison of fibre mapping techniques: Endocardial (A–C) and epicardial (D–F) fibre directions for a human atrial atlas (A$\&$D) mapped to a different patient geometry using either UACs (B$\&$E) or an image based method (C$\&$F). Differences in activation time for the endocardium and epicardium with pacing from the roof or MV respectively are shown in (G) and (J). Activation time maps corresponding to (B, C) are shown in (H, I) for roof pacing, and for (E, F) in (K, L) for MV pacing.

Fig. 10

Point sensitivity. (A) Boundary nodes used for the laplace solves for the baseline case, a case with the LSPV and RSPV markers moved to the middle of the veins (roof 1), a case with the markers at the base of the veins (roof 2), and a case with the septal marker moved to be posterior of the FO. These changes visibly change the boundary condition locations. (B) LA UACs. (C) Epicardial posterior fibres. (D) Epicardial anterior fibres.

Fig. 11

Phase singularity plots visualised in 2D: Normalised phase singularity density maps for three patients in: (A) Posteroanterior view; (B) Anteroposterior view; (C) LA 2D representation; (D) RA 2D representation. Orientation as per Fig. 7.