Technical Note: A 3‐D rendering algorithm for electromechanical wave imaging of a beating heart

Abstract

Purpose Arrhythmias can be treated by ablating the heart tissue in the regions of abnormal contraction. The current clinical standard provides electroanatomic 3-D maps to visualize the electrical activation and locate the arrhythmogenic sources. However, the procedure is time-consuming and invasive. Electromechanical Wave Imaging is an ultrasound-based non-invasive technique that can provide 2-D maps of the electromechanical activation of the heart. In order to fully visualize the complex 3-D pattern of activation, several 2-D views are acquired and processed separately. They are then manually registered with a 3-D rendering software to generate a pseudo-3-D map. However, this last step is operator-dependent and time-consuming. Methods This paper presents a method to generate a full 3-D map of the electromechanical activation using multiple 2-D images. Two canine models were considered to illustrate the method: one in normal sinus rhythm and one paced from the lateral region of the heart. Four standard echographic views of each canine heart were acquired. Electromechanical Wave Imaging was applied to generate four 2-D activation maps of the left ventricle. The radial positions and activation timings of the walls were automatically extracted from those maps. In each slice, from apex to base, these values were interpolated around the circumference to generate a full 3-D map. Results In both cases, a 3-D activation map and a cine-loop of the propagation of the electromechanical wave were automatically generated. The 3-D map showing the electromechanical activation timings overlaid on realistic anatomy assists with the visualization of the sources of earlier activation (which are potential arrhythmogenic sources). The earliest sources of activation corresponded to the expected ones: septum for the normal rhythm and lateral for the pacing case. Conclusions The proposed technique provides, automatically, a 3-D electromechanical activation map with a realistic anatomy. This represents a step towards a non-invasive tool to efficiently localize arrhythmias in 3-D.

Cardiac Resynchronization Therapy Response Assessment with Electromechanical Activation Mapping within 24 Hours of Device Implantation: A Pilot Study

“…QRS durations were measured on 12-lead electrocardiography before and after implantation and are listed in Table 1. Finally, all but nine patients were successfully imaged in the four views required for 3D-rendered isochrone generation, 16 while for the remaining nine, at least the four-chamber view was obtained. This was due to time constraints, patient intolerance and discomfort, or a poor acoustic window for one of the other views.…”

“…Two-dimensional EWI scans were acquired on a Vantage Research system (Verasonics, Kirkland, WA) using a 2.5-MHz P4-2 phased-array transducer in four standard apical echocardiographic planes: four-, 3.5-, two-, and three-chamber views. 16,17 Briefly, the EWI high-frame rate 2-sec ultrasound sequence emits a single diverging wave at 2,000 Hz covering a 20-cm-deep and 90 -wide field of view. Motion is estimated in one dimension with crosscorrelation tracking on the radiofrequency data, 18 and infinitesimal incremental/interframe axial strain, namely, electromechanical strain (on the order of 0.01%) is derived throughout systole using a least squares estimator.…”

“…19 The activation timings, defined as the positive-to-negative strain zero-crossing locations (change from relaxation to contraction), are manually selected on the incremental axial strain curves of the myocardial points after QRS onset 10,20 and later displayed on color-coded two-dimensional (2D) activation maps (milliseconds), with red denoting early activation and blue late (Supplemental Figures 1B, 1C and Supplemental Figures 2.1B, 2.1C, 2.2B and 2.2C). Once all four multi-2D isochrones are created and coregistered, a linear interpolation of the activation time points is performed circumferentially to subsequently generate the 3D-rendered activation maps according to the algorithm in Nauleau et al 16 As a reference, a voxel on the resulting map has an approximate size of 1.7 Â 1.7 Â 0.06 mm (in the x elevational, y lateral, and z axial directions, respectively). Acquisition of the EWI ultrasound scans lasted about 15 min in the postprocedural area, while the full offline processing pipeline required approximately 90 min to run per patient.…”

“…Similarly, we computed the mean activation times from the 3D-rendered isochrones, but on the entire RV and LV wall voxels this time (Figure 2B). In this case, the 3D-rendering algorithm 16 automatically splits the volume into three parts (RV free, septal, and LV lateral walls).…”

Noninvasive localization of cardiac arrhythmias using electromechanical wave imaging

“…After obtaining the multi-2D isochrones in the four apical standard views, the LV median axis or longitudinal apex-to-base rotation axis of the probe (dotted black lines in Fig. 2E) was automatically detected on each view (50). Relative positions of the four 2D imaging planes were assumed to be organized as in the theoretical case with preset probe rotation angles: 60° clockwise between the four-chamber and two-chamber view; 30° clockwise between the four-chamber and 3.5-chamber view; and, last, 60° counterclockwise between the four-chamber and three-chamber view.…”