Tracking the motion of intracardiac structures aids the development of future leadless pacing systems

Leadless pacemakers preclude the need for permanent leads to pace endocardium. However, it is yet to be determined whether a leadless pacemaker of a similar design to those manufactured for the right ventricle (RV) fits within the left ventricle (LV), without interfering with intracardiac structures.


| INTRODUCTION
About one million cardiac pacemakers are implanted every year worldwide. 1 Conventional transvenous cardiac pacemakers with active fixation leads are commonly used to augment the electrical conduction of the heart and treat conditions such as symptomatic bradycardia and high degree atrioventricular block. 2 Since their first implantation in 1950s, pacemaker therapy has evolved considerably.
Reduction in generator size, increased battery longevity, quality of pacemaker leads, and algorithmic responsive programming have all improved pacemaker implantation and management. 1,2 Despite technological advances in conventional transvenous pacemaker therapy, 10% of patients are still affected by surgical complications during the implantation of the device. [3][4][5] The leads are prone to dislodgement, fracture, or insulation failure and can also lead to infection, cardiac perforation, venous occlusion, and tricuspid regurgitation. 3 One treatment requiring transvenous lead implantation is cardiac resynchronisation therapy (CRT), where the left ventricle (LV) is paced via an additional transvenous lead in the coronary sinus (CS). However, CRT has been found to fail in 30%-40% of patients with suboptimal lead placement as one of the reasons for nonresponse. 6 Furthermore, an additional 8%-10% of eligible patients do not receive CRT because of anatomical constraints, such as complex CS anatomy and difficulty in navigating the cardiac valves. 7 The pursuit of leadless pacing options has long been of interest to reduce the complications of traditional CRT. 8 Stimulation of the LV endocardium, which is not constrained to the epicardial coronary venous anatomy, may provide superior hemodynamics and improved CRT response. 9 Nonetheless, the current commercially manufactured leadless pacemakers are limited to perform single-chamber ventricular pacing and their recommended placement in the right ventricle (RV) has not been free of complications. Although comparable to dual-chamber transvenous pacemaker systems, they have shown to result in worsening tricuspid regurgitation and a reduction in ventricular function at 12 months of follow-up primarily due to their mechanical interference on the tricuspid valve or its subvalvular apparatus. 10,11 Since leadless self-contained pacemakers can be ideal for endocardial CRT, developing devices that can physically access all regions of the LV endocardium is desirable. 6,12 Endocardial CRT requires a compact device, which does not interact with the endocardium or intracardiac structures during systole, as collision can cause direct mechanical trauma and lead to cardiac puncture and perforation, which can lead to longer hospital stays, tamponade, or even death. 3,13 Assuming that manufacturers may want to use similar designs to the RV leadless pacemakers for endocardial CRT, it is worthwhile to assess the ability of these designs to fit in the LV safely.
RV leadless pacemakers were not designed for LV pacing and should not be proposed for use in the LV. However, they demonstrate plausible designs for future leadless systems. In this study, we determine whether a leadless pacemaker of a similar design to those commercially manufactured for the RV can access all regions of the LV endocardium without interfering with intracardiac structures. The two designs in our study are the Nanostim Device from Abbott Medical Inc and Micra Transcatheter Pacemaker System from Medtronic PLC. We also estimate optimal theoretical device designs that maximise access in the LV and target regions of the endocardium, which are associated with improved CRT response. 6,12 All our assessments are done by retrospectively analysing motion of the LV endocardial wall, the anterior and posterior papillary muscles, the chordae tendineae, and the mitral valve annulus from a data set of 30 patients with a pacemaker already in situ, who were then upgraded to CRT.

| METHODS
In this section, we first introduce the clinical data set used in our experiments. We then present an image registration technique optimised for tracking cardiac motion, followed by modelling of the intracardiac structures. We finally conclude with a description of the collision detection algorithm used to assess the possible regions within endocardium for the safe placement of a pacemaker design and describe a process for finding an optimal dimension to reduce the risk of collision. Retrospectively ECG-gated image reconstruction was used to generate 10 images per cardiac cycle. All patients were selected for an upgrade to CRT and preprocedural CCT scans were acquired in RV pacing.

| Processing data set scans
Initially, CCT datasets were converted from their original DICOM 1 format into a series of NIfTI 2 images. The dimension of images varied between data sets, but the majority were 512 × 512 voxels with a 0.32-0.48-mm isotropic in-plane resolution. The three-dimensional (3D) stacks had 121-365 slices with a through-plane thickness of 0.8-2 mm. The Hounsfield unit spanned from −1024 to 3071. Each data set consisted of 10 images equally spaced throughout the cardiac cycle with a temporal resolution spanning 72-120 ms depending on the patient's heart rate.

| LV endocardium motion estimation
Tracking motion can essentially be defined as the nonrigid registration of image sequences. To track the LV endocardium motion, we used an intensity-based temporal sparse free-form deformation registration algorithm 14

| Modelling intracardiac structures
A reference image was selected from the first frame of the CCT, representing the LV at the end-diastolic phase and the peak of the R-wave in the ECG. The blood pool of the LV including the papillary muscles was then segmented from this reference image by a clinical expert. Segmentation was performed using a grey-value based region growing tool. The grey values were determined from all point positions plus/minus a margin of 30 Hounsfield units. The 2D region growing segmentation was applied to 5-10 of the long-axis slices.
These slices were interpolated to label the LV cavity in 3D with the option for manual correction. After achieving a full segmentation, a marching cubes process generated a smooth 2D endocardial surface from the segmentation. The average resolution of generated surfaces was 1.0 ± 0.2 mm.
The mitral valve annulus and papillary muscles were also manually marked in the end-diastolic frame ( Figure 1)

| Collision detection
Since the leadless pacemakers were modelled using a cylinder, the collision detection algorithm could be simply devised as a function that determined whether a 3D test point from the en-

| RESULTS
In this section, we report the collision results of two commercially manufactured designs, followed by our analysis of finding an optimal size for the pacemakers with and without considering current technological restrictions. The patient demographic characteristics used in our experiments are summarised in Table 1.

| Pacemaker collision models
The Nanostim LCP and Micra TPS models were placed at each vertex of the LV endocardium mesh for each patient and the collision de-  Note: Values are presented as mean ± SD or as n (%).

| Pacemaker collision models in the right ventricle
Leadless pacemakers were originally developed to avoid interaction with the tricuspid valve in the RV. With the background that leadrelated adverse consequences can be mitigated by leadless pacing, Beurskens et al. 10 reported on the impact of leadless pacemakers on the valvular structure. They found that an RV septal position of the leadless pacemaker was associated with increased tricuspid valve regurgitation, compared with an apical position (odds ratio 5.20; p = .03).
They assumed the fivefold increase was due to displacement of the papillary muscles, entanglement with the chordae tendineae, or direct interaction with the leaflets by the leadless pacemakers. Their findings concluded that longer devices might have greater effects on the valve, shorter ones on the muscles, and a further distance from the proximal end of the pacemaker to the valve decreased the risk of dysfunction. 11 To extend our study, we applied a similar collision simulation approach to generated models of the RV from the same datasets, as seen in Figure 5.

| DISCUSSION
In this study, we assessed the physical ability of leadless pacemakers with similar dimensions to those commercially manufactured for the RV to access regions of the LV endocardium without interfering with intracardiac structures. We also estimated optimal dimensions for theoretical pacemakers to maximise access in the LV endocardium with and without considering current technological limitations.
We found that Nanostim LCP design fits within the apical regions and Micra TPS in the apical, mid anteroseptal, and mid-anterior regions of the LV. We further discovered that decreasing the height of Micra TPS to 20 mm and increasing its diameter to 7 mm will maintain device volume but expand the accessible regions within LV endocardium from 36% to 41%. We confirmed that reducing the volume of Micra TPS can increase the access range from 36% to more than 50% of the endocardial surface, potentially allowing for more optimal pacing locations. At volume set as 472 mm 3 , height as 10 mm, and diameter as 7.7 mm, we predicted complete access across the mid-wall of the LV endocardium. However, the main clinical limitation in deploying the current RV leadless designs in the LV lies in their large dimensions and the risk of stroke from systemic thromboembolism. 16 Previous studies have shown that these devices have a range of endothelialisation, from partial at 19 months to full at 4 months, which can increase the risk of complications. 17,18 An imperative determinant of successful CRT for heart failure is the position of the LV pacing lead. 6,12 LV intracardiac structures limit the placement of currently manufactured designs to primarily apical regions. However, LV leads positioned in the apical region have been associated with an unfavourable outcome in CRT. 12 The myocardial wall is thinner at the apical region, which can also potentially increase the risk of cardiac perforation and lead to tamponade. 13 Figure 2 shows that current designs are unable to target most of endocardial surface, which can be associated with improved CRT response. 6 LV pacing based upon the pacing activity of a co-implanted pacing device, implantable cardioverter defibrillator or CRT system. While an initial study was halted due to safety concerns of these devices, 19 a subsequent study using a second-generation system on a limited cohort of patients showed promising results. 20

| LIMITATIONS
Collision with intracardiac structure was detected using the preprocedural CCT derived motion. This was performed under intrinsic activation or with RV pacing. Ideally, collision would be detected during biventricular pacing. However, post-implant, retrospectively ECG-gated CCT data sets were not available. Implanting an endocardial pacemaker is likely to increase cardiac contraction, increasing the likeliness of collision. Therefore, the current results can be taken as an upper bound on the areas, which endocardial pacemakers can access.
The methods for precise modelling of intracardiac structures required certain practical assumptions and anatomic mapping decisions. Our models of the mitral valve chordae tendineae and papillary muscles were based on previously validated studies. [21][22][23] Furthermore, our collision model did not account for the intracardiac motion of the pacemaker and assumed it would maintain a consistent perpendicular position against the endocardial wall during cardiac contraction. A possible approach to account for this motion would consist of defining the region of potential collision as a semisphere RAZEGHI ET AL | 2437 that encloses the maximum area covered by the motion of the pacemaker. This would have, however, made the computational cost of the study intractable.

| CONCLUSIONS
Leadless pacemaker therapy was developed to address the limitations of standard lead-based pacing. RV leadless pacemakers were not designed for use in the LV. Besides, a left ventricular pacing system with physical dimensions similar to RV pacemakers cannot target most regions of the endocardium without interfering with intracardiac structures. The next generation of self-contained leadless pacemakers needs to be able to synchronise independently and become more compact to be implanted safely within the LV endocardium.