Open-Source Toolkit for Interactive Planning of Transcatheter Mitral Valve Replacement Using Multi-Modality Imaging

Mitral regurgitation(MR) is the most prevalent form of mitral valve disease in developed countries. Transcatheter mitral valve replacement (TMVR) has emerged as a therapeutic option for the treatment of severe MR and avoids the morbidity and mortality associated with open heart surgery.[1] One challenge to successful application of this therapy is the matching of a growing selection of TMVR devices to the unique dimensions and geometry of an individual patient. Cardiac computed tomography (CT) and recently three-dimensional echocardiography (3DE) are increasingly used for this planning process, in hopes of avoiding complications such as left ventricular outflow tract(LVOT) obstruction, or perivalvar leak.[2] However, the complexity of cardiac anatomy can make 2-dimensional and even 3-dimensionsal (3D) derived linear measurements difficult to translate into improved appropriate device selection, which in turn has led to exploration of CT-derived virtual and 3D printing-based modeling.[3, 4] While CT-based structural modeling software is now commercially available, and there is mitral annular modeling available in commercial 3DE-based platforms, to our knowledge there is no configurable modeling platform of 3DE modeling which allows the insertion of a library of existing and customizable virtual devices into 3DE images to see how devices would actually fit.

In order to potentially inform patient candidacy, appropriate device selection, and the design of new devices for TMV therapies, we created a modeling and visualization framework to insert virtual TMV models based on emerging and available devices into both segmented and volume-rendered 3DE images using custom code within an open-source framework(Figure 1).[1] We further demonstrate analogous functionality in CT images, similar to existing commercial tools, but with the ability to place more complex and configurable virtual devices rather than basic cylinders. We have implemented this functionality into custom modules in the SlicerHeart extension for 3D Slicer(www.slicer.org), an open-source image processing platform.[5, 6] Our newly developed modules include Valve Annulus Analysis, Valve Quantification, and Transcatheter Atrioventricular Valve Simulator. The workflow begins with import of 3DE and CT images of adult sized, adolescent patients for visualization and segmentation (Figure 1, Video 1), followed by an annular quantification workflow to define the mitral and aortic annuli, and calculate annular metrics (Figure 1, Video 2).[5, 6] We then created the functionality to place parametric models of emerging TMV devices, all of which can be further customized in their size and shape(Video 3). TMV devices were placed within volume rendered 3DE and CT images and image derived segmented models in order to visualize device fit(Figure 2, Video 3). In order to quantitatively assess the potential for LVOT obstruction we created a complementary tool for visualizing and quantifying the LVOT area in a plane orthogonal to the aortic outflow with the virtual device in place(Supplemental Figures 1 and 2, Video 4).

Figure 1.

Visualization, Modeling, and Annular Quantification. A. Volume rendering of 3DE (left atrial view) and cardiac CT (left ventricular view) of two adult sized, adolescent patients in 3D Slicer; B. Image segmentation and annular quantification using 3DE (left atrial view) and CT (left ventricular view) images. 3DE = 3D echocardiographic, CT = computed tomography, A = anterior, P = posterior, AL = Anterior-Lateral, PM = Posterior-Medial.

Figure 2.

Virtual Transcatheter Mitral Valve Device Placement within Volume Rendered and Segmented 3DE and CT images. A. Apical tether device(red) placed within volume rendering of 3DE of a mitral valve(left atrial view) and radial force device (blue) placed within volume rendering of cardiac CT(left ventricular view); B. Annular winglet device (red) placed within 3DE-derived segmented model and apical tether device (blue) placed within CT-derived segmented model. 3DE = 3D echocardiography, CT = computed tomography.

We believe that the ability to model and quantify anatomy and insert realistic and configurable virtual devices to assess real device fit has the potential to inform 3DE based pre-operative planning for a growing number of TMV devices, similar to evolving tools based on CT imaging.[3] This more detailed modeling, using actual device shapes rather simple measurements, or basic cylinder models, could eventually be incorporated into commercial imaging platforms, allowing virtual “testing” of device placements. Virtual device modeling within 3DE images may be especially useful for application of transcatheter atrioventricular valve therapies in heterogeneous populations of children with congenital heart disease where the radiation of multiphase CT may be undesirable. 3DE also provides high frame rate relative to CT imaging, which may be helpful to reliably capture the full range of anatomic conformations of the mitral annulus and left ventricular LVOT throughout the cardiac cycle. However, CT reliably creates high-resolution images, is not dependent on optimal sonographic windows, and provides additional information about tissue composition, which may be essential when considering implantation within mitral annular calcification. While we have demonstrated application of simple precursors of these 3DE-based tools in a small cohort of children requiring surgical implantation of a serially expandable stent based valve[7], further validation of these more complex and maturing tools, as well as comparison to existing CT-based tools, is needed.[2] In particular, future incorporation of the anteriorly deflected anterior leaflet of the mitral valve into the simulation will be important to determine the true “Neo-LVOT” area (created by the combination of the device and deflected anterior leaflet) as well as exploration of the potential need for procedures such as the intentional laceration of the anterior leaflet.

This study was approved by the institutional review board at the Children’s Hospital of Philadelphia. We have made this modeling and device insertion functionality part of modules within the free, and open-source SlicerHeart extension available within the 3D Slicer Extension manager. The 3D Slicer platform and documentation are available at http://www.slicer.org. Open-source code and documentation are available at https://github.com/SlicerHeart.

Supplementary Material

1

Supplemental Figure S1. Left Ventricular Outflow Tract Area Quantification Workflow Using 3D Echocardiography(3DE) and Cardiac Computed Tomography (CT).

2

Supplemental Figure S2. Visual Comparison of Two Device Placements and Effect on Left Ventricular Outflow Tract Area Using a 3D Echocardiogram(3DE)-derived Model.

3

Video 1: Overview of Visualization of 3DE and CT. We describe the visualization of a 3D echocardiogram (3DE) of a 14-year-old, 147 kg male with rheumatic mitral disease as well as a computed tomography (CT) image of 14-year-old 55 kg female with structurally normal anatomy. Digital Imaging and Communications (DICOM) CT images are natively supported in 3D Slicer. 3DE images were converted and imported as previously described.[5, 6]

4

Video 2: Overview of Annular Modeling and Quantification. We demonstrate creation of annular models of mitral valve in the images of the two adolescent, adult sized patients. We then demonstrate placement of anterior(A) and posterior(P) annular markersmarkers, followed by designation of the anterior lateral commissure (ALC) and posterior medial commissure (PMC). Finally, we demonstrate automatic quantification of annular metrics. 3DE = 3D echocardiography, CT = computed tomography.

5

Video 3: Insertion of Virtual Devices in Segmented and Volume Rendered Images. We demonstrate the configuration and insertion of a library of virtual devices into the image-derived models described in Videos 1 and 2. 3DE = 3D echocardiography, CT = computed tomography.

6

Video 4: Quantification of the Left Ventricular Outflow Tract (LVOT). We demonstrate the visual placement of a line centered in the LVOT after device placement followed by visualization and quantification of the LVOT area. 3DE = 3D echocardiography, CT = computed tomography.