Amir H. Sadeghi [email protected], Joris F. Ooms, Wouter Bakhuis, Yannick J.H.J. Taverne, Nicolas M. Van Mieghem, and Ad J.J.C. BogersAuthors Info & Affiliations

Abstract

In patients with paravalvular leakage (PVL) and high (re)operative risk, transcatheter PVL closure (TPVLC) by means of a closure device is a valuable treatment option (1). Because PVL dimension and location are highly patient-specific, detailed preprocedural planning is required to determine TPVLC eligibility. Currently, echocardiography and computed tomography (CT) are the primary modalities to determine PVL severity and location (2). However, these only marginally incorporate the 3-dimensional (3D) geometry. Patient-specific, CT-derived, 3D computational models (3DCMs) in an immersive virtual reality (VR) environment could potentially provide a detailed representation of the PVL orifice, allowing multidimensional measurements and virtual device fitting. In this retrospective analysis, we assessed 3DCMs of PVL defects in an immersive VR environment to evaluate its feasibility, applicability, and added value in TPVLC planning.

All patients who underwent TPVLC of either an aortic or mitral PVL and who had received a multimodality preprocedural workup (transesophageal [TEE] and transthoracic [TTE] echocardiography, CT, and 3DCM) at our center were included. The study was approved by the Erasmus Medical Center medical ethics committee (MEC2021-0130). CT protocols and generation of 3DCMs (using Mimics Innovation Suite, Materialise) were described earlier (3).

Before each procedure, TTE/TEE, CT, and 3DCMs of each lesion were reviewed, and a PVL closure device (Amplatzer Vascular Plug III, Abbott Vascular) was selected using sizing charts provided by the manufacturer. A VR viewer (CardioVR, developed in-house together with MedicalVR) was used for evaluation of 3D-VR models. Previously acquired 3DCMs were loaded as 3D object files into CardioVR retrospectively, and PVL dimensions were reviewed by an investigator (blinded to other imaging results) on a VR headset. Feasibility, applicability, and added value of CardioVR was evaluated by comparing the VR-based sizing results to CT, TEE, and 3DCM.

A total of 7 PVLs in 6 patients were analyzed. Indications were ≥moderate-severe PVL (2/6 patients) combined with hemolysis (4/6 patients). Figure 1 provides a schematic, scaled representation of the dimensions measured with each modality, the ultimately deployed plug(s), and the retrospective VR measurements. The 3DCM and VR assessments allowed measurements of proximal (atrial/aortic) and distal (ventricular) side of the lesion, resulting in 2 sets of measurements. This was not possible with 2D-CT and TEE. Impaired spatial resolution prohibited measurements in 3 of 7 and 1 of 7 leaks for TEE and 2D-CT, respectively. Mean differences between 2D-CT and 3DCM were −0.4 ± 1.4 mm for minimal and 0.1 ± 1.3 mm for maximal diameter. For TEE, no difference was obtained due to missing data. Mean differences between 3DCM and VR measurements were −0.5 ± 3.2 mm, −0.7 ± 0.8 mm, −1.4 ± 1.5 mm, and −0.3 ± 0.9 mm for proximal-maximal, proximal-minimal, distal-maximal, and distal-minimal, respectively. No in-hospital mortality or major complications were observed. On predischarge TTE, PVL severity was ≤mild in 5 of 6 patients and remained stable at 90 days’ follow-up.

Several interesting observations were made in this pilot study. There was meaningful variability between TEE and CT measurements (Figure 1) with TEE often yielding smaller dimensions. In 3 of 7 cases, TEE was unable to visualize any geometrical defect. In terms of PVL dimensions, VR measurements showed comparable results with CT/3DCMs because these 3 modalities were derived from the same CT scan. When comparing inserted plug dimensions with individual TEE/2DCT/3DCM/VR dimensions, it becomes apparent that the final plug closest approximated the CT-derived measurements, confirming CT as the decisive modality in sizing.

Several added benefits of 3D modalities can be noted. 3DCMs allowed assessment of both the proximal and distal orifice and corresponding lesion depth. Compared with 2D-CT and TEE, 3DCM and VR enabled instant visualization of the PVL and surrounding anatomy in the x-, y-, and z-axes. In addition, immersive VR facilitated evaluation with in-depth perception and free 3D manipulation in space.

Currently, 3D visualization has several drawbacks: the creation and VR visualization of 3DCMs requires dedicated segmentation software, high-performance computers, VR workstations, and technicians. Moreover, it can be a time-consuming process in complex cases. Semiautomated processes are expected to reduce preparation times. Study-specific limitations include small sample size, absence of a gold standard, and unavailability of full cardiac cycle CT scans. In conclusion, we have demonstrated feasibility and applicability of immersive VR technology as a potential modality to evaluate PVL dimensions for TPVLC. CT-derived 3DCMs review in VR may complement conventional imaging in planning TPVLC, exceeding training and educational purposes. Additional research is required to evaluate its impact on procedural outcome.