Four-Dimensional Computed Tomography-Guided Valve Sizing for Transcatheter Pulmonary Valve Replacement

Summary

This study assessed a new methodology with a straightened model generated from the four-dimensional cardiac computed tomography sequence to obtain the desired measurements for valve sizing in the application of transcatheter pulmonary valve replacement.

Abstract

The measurements of the right ventricle (RV) and pulmonary artery (PA), for selecting the optimal prosthesis size for transcatheter pulmonary valve replacement (TPVR), vary considerably. Three-dimensional (3D) computed tomography (CT) imaging for device size prediction is insufficient to assess the displacement of the right ventricular outflow tract (RVOT) and PA, which could increase the risk of stent misplacement and paravalvular leak. The aim of this study is to provide a dynamic model to visualize and quantify the anatomy of the RVOT to PA over the entire cardiac cycle by four-dimensional (4D) cardiac CT reconstruction to obtain an accurate quantitative evaluation of the required valve size. In this pilot study, cardiac CT from sheep J was chosen to illustrate the procedures. 3D cardiac CT was imported into 3D reconstruction software to build a 4D sequence which was divided into eleven frames over the cardiac cycle to visualize the deformation of the heart. Diameter, cross-sectional area, and circumference of five imaging planes at the main PA, sinotubular junction, sinus, basal plane of the pulmonary valve (BPV), and RVOT were measured at each frame in 4D straightened models prior to valve implantation to predict the valve size. Meanwhile, dynamic changes in the RV volume were also measured to evaluate right ventricular ejection fraction (RVEF). 3D measurements at the end of the diastole were obtained for comparison with the 4D measurements. In sheep J, 4D CT measurements from the straightened model resulted in the same choice of valve size for TPVR (30 mm) as 3D measurements. The RVEF of sheep J from pre-CT was 62.1 %. In contrast with 3D CT, the straightened 4D reconstruction model not only enabled accurate prediction for valve size selection for TPVR but also provided an ideal virtual reality, thus presenting a promising method for TPVR and the innovation of TPVR devices.

Introduction

Dysfunction of the right ventricular outflow tract (RVOT) and pulmonary valve abnormalities are two of the most frequent consequences of severe congenital heart disease, for example, patients with repaired tetralogy of Fallot (TOF), certain types of double outlet right ventricle (DORV), and transposition of the great arteries^1,2,3. The majority of these patients face multiple operations throughout their lives and along with advancing age, the risks of complexity and comorbidities increase. These patients may benefit from transcatheter pulmonary valve replacement (TPVR) as a minimally invasive treatment⁴. To date, there has been a steady growth in the number of patients undergoing TPVR and several thousands of these procedures have been performed worldwide. Compared with traditional open-heart surgery, TPVR requires a more accurate anatomical measurement of the xenograft or homograft from the right ventricle (RV) to pulmonary artery (PA), as well as the repair of pulmonary and RVOT stenosis via transannular patch, by computed tomography angiography (CTA) prior to intervention and to ensure that the patients are free from stent fracture and paravalvular leak (PVL)^5,6.

A prospective, multicenter study demonstrated that a multidetector CT annular sizing algorithm played an important role in selecting the appropriate valve size, which could decrease the degree of paravalvular regurgitation⁷. In recent years, quantitative analysis has been more and more applied in clinical medicine. Quantitative analysis has enormous potential to enable objective and correct interpretation of clinical imaging and to verify that patients are free of stent fracture and paravalvular leak, which can enhance patient-specific therapy and treatment response evaluation. In previous clinical practice, it was feasible to reconstruct CT imaging from three planes (sagittal, coronal, and axial) with two-dimensional (2D) CT to obtain a visualization model⁸. Contrast-enhanced electrocardiogram (ECG)-gated CT has become more important in the evaluation of RVOT/PA 3D morphology and function, as well as in the identification of patients with a suitable RVOT implantation site that is capable of maintaining TPVR stability throughout the cardiac cycle^9,10.

However, in the contemporary standard clinical and preclinical settings, the acquired 4D CT data are usually translated into 3D planes for manual quantification and visual evaluation which cannot show 3D/4D dynamic information¹¹. Furthermore, even with 3D information, the measurements obtained from multiplanar reconstruction (MPR) have various limitations, such as poor quality of visualization and lack of dynamic deformation due to the different directions of blood flow in the right heart¹². Measurements are time-consuming to gather and prone to mistakes, as 2D alignment and sectioning can be imprecise, resulting in misinterpretation and distensibility. Currently, there is no consensus on which measurement of RVOT-PA could reliably provide accurate information about the indications and valve sizing for TPVR in patients with dysfunctional RVOT and/or pulmonary valve disease.

In this study, the method for measuring RVOT-PA using a straightened right heart model via a 4D cardiac CT sequence is provided to determine how best to characterize the 3D deformations of RVOT-PA throughout the cardiac cycle. The spatio-temporal correlation imaging was completed by including the temporal dimension and, therefore, were able to measure variations in RVOT-PA magnitude. Additionally, the deformation of the straightened models could positively impact TPVR valve sizing and procedural planning.

Protocol

All cardiac CT data were obtained from GrOwnValve preclinical trials with the approval of the legal and ethical committee of the Regional Office for Health and Social Affairs, Berlin (LAGeSo). All animals received humane care in compliance with the guidelines of the European and German Societies of Laboratory Animal Science (FELASA, GV-SOLAS). In this study, the Pre-CT from sheep J was chosen to illustrate the procedures.

1. Perform 3D cardiac CT in sheep

1. Intravenous anesthesia
   1. Tranquilize sheep (3 years, 47 kg, female, Ovis aries) with premedication of midazolam (2 mg/mL, 0.4 mg/kg), butorphanol (10 mg/mL, 0.4 mg/kg), and glycopyrronium bromide (200 mcg/mL, 0.011 mg/kg) by intramuscular injection.
   2. Check the physical condition of the sheep when they became docile, 15 min after the injection.
   3. Place an 18 G catheter with injection port aseptically in the cephalic vein with perfusion lines jointed to a T-connector for anesthesia and contrast agent.
   4. Anesthetize the sheep by intravenously injecting propofol (20 mg/mL, 1-2.5 mg/kg) and fentanyl (0.01 mg/kg). Check for symptoms of tranquilization like jaw relaxation, loss of swallowing, and ciliary reflex. Intubate the sheep with a 6.5 mm - 8 mm tracheal tube, and place a gastric tube into the stomach for gastric fluid aspiration followed by intravenous injection of propofol (20 mg/mL, 1-2.5 mg/kg) and fentanyl (0.01 mg/kg).
   5. Achieve total anesthesia by injecting propofol (10 mg/ml, 2.5-8.0 mg/kg/h) and ketamine (10 mg/mL, 2-5 mg/kg/h) intravenously, in preparation for cardiac CT.
2. Cardiac CT
   1. Transfer the sheep from the Research Institutes for Experimental Medicine (FEM) to the CT room of the German Heart Center Berlin (DHZB) after the preparations. Scan all sheep in the prone position after securely fixing them on the CT bed with 3 bandages on the arms, abdomen, and legs.
   2. Perform cardiac CT on a 64-slice dual-source multidetector CT system with ECG-gating using the following parameters. Set the standard acquisition technical parameters as follows: gantry rotation time 0.33 s, 100-320 mAs per rotation, 120 kV tube voltage, matrix 256 with a 16-bit depth, deviation effective x-ray dose 15.5± 11.6 mSv, slice thickness 0.75 mm.
   3. Achieve contrast enhancement by administering 2 -2.5 mL/kg of iodinated contrast agent at the rate of 5 mL/s via the T- connector on the arm.
   4. Perform the 4D CT scanning protocol in sequential. Divide the entire cardiac cycle into 11 frames from 0% to 100% with 10% of R-wave to R- wave (RR) interval covering the cardiac cycle. Carry out an end-diastolic phase at approximately 70% of the RR-interval for analysis for the 3D series. Obtain sagittal, coronal, and axial data in each frame of 4D CT, as well as in 70% 3D series.
   5. Use a bolus tracking method for contrast bolus timing in the region of interest on the main pulmonary arteryto achieve ideal synchronization. Do not administer beta-blocker in any sheep.
   6. Transfer the sheep back to the FEM and stop the perfusion of propofol, and ketamineafter scanning. The sheep regained consciousness 10 - 20 min after the extubation. Anesthesiologists and veterinarians oversaw the entire anesthesia treatment until the sheep were completely awake and able to move freely.

2. Open-source 3D reconstruction software application settings and extension installments

1. Click Edit in the top menu to modify the application settings after launching 3D reconstruction software.
   1. Click on DICOM, then Acquisition Geometry Regularization, and select Apply Regularization Transform in the DICOM Scalar Volume Plugin section. Select Volume Sequence as the preferred multi-volume import format in the Multi Volume Importer Plugin section.
   2. Click on Views, select Small Axes. In the Orientation marker, select Thin Ruler.
   3. Restart the 3D slicer software to save the application settings.
2. Click the Extension Manager in the toolbar to open the extensions page.
   1. Find the required extensions and left-click to install them. Use the following extensions in this study: Sequence Registration, Slicer Elastix, Sandbox, Slice Heart, Slicer IGT, Slicer VMTK, DICOM web Browser, Intensity Segmenter, Markups To Model, Easy Clip, mp Review, Slicer Prostate, and VASSTAUgorithms.
   2. Restart the 3D slicer software to confirm the installation of the selected extensions.

3. Load cardiac CT data into 3D slicer from the DICOM files

1. Use one of the two steps described below to load the cardiac CT data into 3D slicer from the DIOCM files (Figure 1).
2. Import CT data: Add cardiac CT data (the Pre-CT from sheep J was selected to illustrate the procedures) into the application's database by switching to the DICOM module and dragging-and-dropping files to the application window.
3. Load CT data: Load data objects into the scene by double-clicking on items (In sheep J, the EKG- Ao asc 0.75 126f 3 70% is the 3D sequence at the end-diastolic phase, and Funkion EKG- Ao asc 0.75 126f 3 0- 100% Matrix 256 is the 4D sequence as an 11-frame volume sequence by cardiac cycle).
4. Left-click the Eye icons in the data tree to show the 3D and 4D sequences from the axial, sagittal, and coronal views in the 2D viewers.
5. Left-click the Slicer layout icon on the top toolbar and select Four-Up or Conventional layout.
6. Click on the Links icon in the top left corner to link all three viewers, and on the Eye icon to display the slices in 3D Viewer.
7. Click on the Save icon and save all the data loaded into the 3D slicer in a selected destination to build a dataset for the segmentation and volume editing.

4. Create 4D beating heart volume and beating right heart volume

1. Select Volume Rendering in the modules drop-down menu, then select the 4D sequence in the Volume drop-down menu.
2. Select CT-Cardiac3 in the Preset drop-down menu to display the 4D heart. Adjust the cursor below the Preset drop-down menu to show the heart only.
3. Click on Sequence Browser in the modules drop-down menu to select and display the 4D sequence. The beating heart is in the scene. Drag the 4D heart into the 3D scene to observe the heart from various directions.
4. Select the Enable and Display ROI functions in the Crop options below the shift bar to crop the 4D volume of the beating heart in order to better observe the structures of the heart.
5. Create the 4D beating heart volume as outlined above. Select Segment Editor in the modules drop-down menu, then click the Scissors effect with the Fill Inside operation to cut one single frame.
6. Click on the Mask Volume effect and apply it to link the segmentation to the 4D heart as a masked volume. The input volume and output volume in the mask volume effect are the 4D sequences.
7. Select the Scissors effect with the Erase inside operation to remove the bones and other unexpected areas. Select the Islands effect with the Keep Largest Island operation to remove small areas.
8. Choose the Erase effect with the 1-3% Sphere Brush to remove the tissues at the aortic arch with attachments to the main pulmonary artery, as well as the tissue between the ascending aorta and the superior vena cava. After each step, apply the Mask Volume effect to mask the 4D volume.
9. Repeat steps 4.7 - 4.8 to carry on removing the areas until the right heart model is shown in the 3D scene.
10. Click on the Sequence Browser and go to the next frame. Use the Scissors effect with the Erase Inside operation to cut any area in the 3D scene; the right heart model will automatically appear in the contemporary frame. Apply the same method to the rest of the frames until the entire 4D sequence has been segmented.
11. Click on the Sequence Browser button to display the right heart 4D volume.
    ​NOTE: When removing the left anterior descending coronary artery in some frames as well as the bifurcation of the left coronary artery, it will remove a tiny portion of the right ventricle. Because of this, it is highly recommended to keep a tiny piece of these coronaries to maintain the right ventricular volume in each frame.

5. Create straightened models from the 4D sequence

NOTE: It is highly recommended to build each 10% of the cardiac cycle frame in a single 3D slicer folder, otherwise there will be too many data trees aligned in the DATA module, making it inefficient to create the straightened models. To get the single 3D slicer folder of each 10% frame, it needs to load the 4D sequence several times, choose every frame and save them in a single folder.

1. Create right heart segmentations for each frame by selecting the Segment Editor module in the toolbar. Add two segmentations for each 10% frame of the 4D sequence, and name them accordingly, e.g., 60% segmentation and Other.
2. Select the Paint effect tool in the Segment Editor module with Editable Intensity Range which depends on the CT images to paint the right heart with the sequence superior vena cava, right atrium, right ventricle, and pulmonary artery.
3. Click on Other Segmentation, use the paint tool to paint other areas to trace the boundaries of the right heart in general.
4. Select the Grow From Seeds effect, select Initialize and Apply to apply the effect. Click on the Show 3D button in the Segment Editor module to display the 3D model of the contemporary frame.
5. Repeat steps 4.7 - 4.8 to remove or improve the 3D model according to the CT images in the three directions. Remove the left and right branches of the pulmonary artery at the bifurcation. The right heart 3D model will then show the 3D scene in each frame.
   ​NOTE: It is highly recommended to paint the boundaries of the right heart with a 1% - 2% diameter sphere brush at the attachments between the pulmonary artery and coronary arteries, as well as the pulmonary artery and the superior vena cava.
6. Clone the segmentations in the DATA tree as a backup, name the segmentations, for example, 10% Segmentation Original and 10% Segmentation for Straightened Model.
7. Add a centerline to the right heart model as described below.
   1. Select Extract Centerline in the modules drop-down menu.
   2. Select Segmentation in the surface drop-down menu in the Inputs section of the extract centerline module. This creates a segmentation, such as 10% segmentation for straightened model as a segment. Click on Create New Markups Fiducial in the endpoints drop-down menu. Click on the Place a Markup Point button to add endpoints on the top plane of the SVC and the end plane of the main pulmonary artery.
   3. Select Create a New Model as a Centerline model and Create New Markups Curve as a centerline curve in the Tree of the Outputs menu. Click on Apply to show the centerline right heart model.
   4. Click on the DATA module, then right-click on the Centerline Curve to edit its properties. Click on the Eye icon to display the control points, and in the Resample section set the number of resampled points to 40 to lower the computer load.
8. Create a straightened model
   1. Select Curved Planar Reformat in the modules drop-down menu.
   2. Shift the cursor after Curve resolution and Slice resolution to 0.8 mm, set the Slice Size to 130140 mm which was according to the range of the right ventricle displayed on the images, and then select Create a New Volume as Output Straightened Volume.
   3. Click on Apply to obtain the straightened volume.
   4. Select Volume Rendering in the module drop-down menu to show the straightened volume. Select the Straightened Volume in the volume drop-down menu and click on the Eye icon. Select CT-Cardiac3 as the Preset, move the Shift cursor to show the straightened right heart volume in the 3D scene.
   5. Column the straightened volume in the DATA tree in the name of straightened volume for segmentation, and right-click to segment this straightened volume.
   6. Select the Threshold effect in the segment editor module to color the desired straightened right heart and click Apply to apply the operation. Select the Mask Volume effect to mask the straightened volume by choosing the Straightened Volume for Segmentation, volume as Input Volume and Output Volume and click Apply to apply the operation.
   7. Click Apply to apply the same operation as outlined above in steps 4.7- 4.8 to keep the straightened right heart segmentation only. Check the straightened right heart volume and 3D model of the straightened right heart segmentation in the 3D scene.
   8. Click Apply to apply the same operation outlined above for other frames to obtain the straightened right heart volume rendering and straightened segmentations and save them in the folder of each frame.

6. Export the figures and STL files

1. Export the figures of the straightened volume rendering by clicking on the Capture and name a scene view effect on the toolbar and saving the scenes in 3D view.
2. Export the STL files of the straightened 3D segmentations by clicking on the Segmentation module.

7. Perform five planar measurements

1. Perform a five planar measurement of the perimeter, cross-sectional area, and circumference in the straightened models from the 4D sequence and right ventricular volume measurements in the straightened model as described below.
2. Apply the following five planar settings: Plane A: at the main pulmonary artery 2 cm offset from the plane of the sinotubular junction; Plane B: at the sinotubular junction; Plane C: at the sinus; Plane D: at the base of the leaflet; Plane E: at RVOT 1 cm offset from D.
3. Add all the above five planes into the straightened models in each frame by holding the Shift key on the keyboard and using the crosshair function in the toolbar to the five planes. Click on the Create and Place module in the toolbar to select the Plane effect.
4. Select the Line effect to measure the perimeters, select the Closed Curve effect to obtain the circumferences and cross-sectional area. Copy the data to build the dataset.
5. Perform right ventricular volume measurements in the straightened model as described below.
   1. Column the straightened segmentation in each frame obtained from the 4D sequence, and label the segmentation according to the matching frame for volume measurement.
   2. Select the Segment Statistics module in the module drop-down menu. Select the X% Segmentation for volume measurement after Segmentation and Scalar Volume in the inputs menu. Select Create New Table as the Output Table and then click on Apply to apply the operations to get the volume table.
   3. Copy the volume data to create the volume measurement dataset for each frame of the straightened segmentation.

8. 3D multiplanar reconstruction (MPR) measurements and right ventricular volume measurement from the 3D sequence (the best-reconstructed phase at the end of diastole)

NOTE: In this study, the sheep J Pre-CT was chosen to illustrate the MPR measurement procedures.

1. Load the diastolic 3D sequence as illustrated in the following steps. Select the downward arrow next to the crosshair effect, choose Jump Slices- Offset, Basic+ Intersection, Fine Crosshair, and The Slice Intersections for crosshair settings.
2. Shift + left-click to drag the crosshair to the plane, for instance, the sinus. Press Ctrl+Alt to adjust the crosshair to the desired position in the axial, sagittal, and coronal scenes perfectly in the center of the targeted position.
3. Select the Line effect to perform the measurements in each plane as illustrated in step 7.4. Copy the data to build the 3D MPR measurement dataset.
4. Click on the Segment Editor module to create a right ventricular segmentation as outlined above in step 5.8.6.
5. Click on the Segment Statistics module to perform the right ventricular volume measurement as outlined above in step 7.5.2.
6. Copy the volume information to build the diastolic 3D right ventricular volume dataset.

9. Calculation for stented heart valve selection

NOTE: In this section, the measurements of the sinotubular junction were used to illustrate the procedure.

1. Calculate the mean of the long axial (d₁) and short axial perimeters (d₂) = (d₃), followed by the mean of d₁, d_2, and d₃ to obtain d₄, as shown in formulas (1) - (2).
   
   
2. Divide the calculation of the cross-sectional area (S₁) by π to obtain d₅ followed by the square root of d₅ to obtain d₆, and then the mean of d₅ and d₆, as shown in formulas (3) - (5).
   
   
   
3. Divide the circumference (C₁) by π to obtain d₈, as shown in formula (6).
   
4. Obtain the overall general diameter d₉ by calculating the mean of d₄, d_7, and d₈, as shown in formula (7).
   
5. Apply formula (8) to calculate the best choice of valve size (h).
   
   NOTE: The stented heart valve is available in diameters 30 mm, 26 mm, and 23 mm. The valve size (h) shows the match as a percentage for the three diameters, namely an ideal match as 10-20%, big for implantation as 30% and above, and small for implantation below 10%.
6. Import the 3D and 4D data into a versatile statistics software to build the trend diagrams of the measurements in the five planes and export the diagrams in TIFF format. Import all the figures into graphics software for organization.

Representative Results

In sheep J, the 4D total heart and right heart models were successfully generated from the 4D cardiac CT sequence which showed the deformation throughout the entire cardiac cycle. For better visualization, the whole deformation of the beating heart and right heart is exhibited in every direction in Figure 3 - Figure 4 and in Video 1 - Video 2.

The straightened right heart models were obtained following the mask volume in each 10% of the segmentation to illustrate the deformations of the right heart in a straightened model in sheep J Pre-CT (Figure 5).

Five planes were added in the desired locations to perform the measurements as shown in Figure 2A, as well as the MPR measurements in 3D reconstruction software and not the conventional method of cropping the 4D volume in sheep J Pre-CT shown in Figure 2B. The changes in cross-sectional area, perimeter, and circumference were obtained in different phases of the cardiac cycle to generate the tendency diagrams as shown in Figure 6. Original data from 4D CT measurements and 3D CT measurements are shown in Supplementary file 1. In sheep J, 4D CT measurements from the straightened model resulted in the same choice of valve size for TPVR (30 mm) as the MPR measurements from the end-diastolic series, with the advantages of remarkable virtual reality and reliable results. There were significant differences in the measured cross-sectional area (RVOT: 3.42 cm² in 4D versus 4.28 cm² in 2D, BPV: 2.96 cm² in 4D versus 3.92 cm² in 2D), and circumference (RVOT: 76.1 mm in 4D versus 87.06 mm in 2D, BPV: 67.65 mm in 4D versus 75.73 mm in 2D) in RVOT and the basal plane of the pulmonary valve. The right ventricular ejection fraction of sheep J from the pre-CT was 62.1%.

Figure 1. User interface in 3- dimensional reconstruction software. Toolbar, data tree, and other functional menus of the 3- dimensional reconstruction software are shown for operating the program. Please click here to view a larger version of this figure.

Figure 2. Five planes in the straightened model for measurement and multiplanar reconstruction measurements in the 3-dimensional sequence (end-diastolic phase). (A) Plane a: main pulmonary artery, 20 mm offset from plane b; plane b: sinotubular junction; plane c: sinus of the pulmonary valve; plane d: bottom of pulmonary valve; plane e: in the right ventricular outflow tract, 10 mm offset from plane d. (B) MPR measurements in the 3D sequence of the end-diastolic phase at five planes: 10 mm offset from the bottom of the pulmonary valve, bottom of the pulmonary valve, sinus of the pulmonary valve, sinotubular junction, and main pulmonary artery (20 mm offset from sinotubular junction). Please click here to view a larger version of this figure.

Figure 3. 4-dimensional heart deformations throughout the cardiac cycle. Total heart deformations of sheep J pre-computed tomography shows the shape changes from 0% to 100% of the cardiac cycle. Please click here to view a larger version of this figure.

Figure 4. 4- dimensional right heart deformation throughout the cardiac cycle. Right heart deformations of sheep J pre- computed tomography shows the shape changes from 0% to 100% of the cardiac cycle. Please click here to view a larger version of this figure.

Figure 5. Straightened right heart deformation of the sheep J pre- computed tomography throughout the cardiac cycle. Straightened right heart deformations of sheep J pre- computed tomography shows the shape changes from 0% to 100% of the cardiac cycle. Please click here to view a larger version of this figure.

Figure 6. Changes in circumference, average diameter, cross-sectional area, and right ventricular volume throughout the cardiac cycle. (A) Changes in circumference during the cardiac cycle at the five planes. (B) Changes in average diameter (calculated using formula 1 in step 9.1) during the cardiac cycle at the five planes. (C) Changes in the cross-sectional area during the cardiac cycle at the five planes. (D) Change in right ventricular volume during the cardiac cycle. Please click here to view a larger version of this figure.

Video 1. 4- dimensional total heart deformation. Throughout the cardiac cycle, the 4-dimensional entire heart reconstruction can be visualized in every direction. Please click here to download this Video.

Video 2. 4- dimensional right heart deformation. The beating heart (superior vena cava, right atrium, right ventricle, and pulmonary artery) can be visualized in every direction throughout the entire cardiac cycle. Please click here to download this Video.

Supplementary file 1. The table presents the original data from 4D CT measurements and 3D CT measurements generated by following the protocol described including the parameters from the pulmonary artery, right ventricular volume, and the measurements of the aorta from sheep J pre- computed tomography. Please click here to download this File.

Discussion

To date, this is the first study to illustrate a patient-specific measurement of the morphology and dynamic parameters of RVOT-PA with a straightened cardiac model generated from a 4D CT sequence, which can be applied to predict the optimal valve size for TPVR. This methodology wasillustrated using sheep J Pre-CT imaging to obtain the dynamic deformations, right ventricular volumes, right ventricular function, and magnitude of RVOT/PA change from the RVOT to the pulmonary trunk in five planes at every 10% reconstruction of the cardiac cycle. Compared with 3D imaging, the straightened models not only predicted the same valve size as the MPR measurements from the end-diastolic 3D images but also allowed for a more intuitive model to extract the desired information about the right heart. According to the findings of a previous study¹³, the proposed method allows for a better understanding of in vivo loading conditions in patients with dysfunctional RVOT and/or pulmonary valve disease, as well as the development of new TPVR devices that are morphologically adapted to the different RVOT anatomies of patients requiring TPVR and may exhibit improved mechanical performance in the long run. However, the current methodology of quantitative measurement for a pre-interventional evaluation of TPVR is based on MPR measurements in the 3D sequence, which could result in unexpected errors during evaluations based on the anatomical curve of the RVOT and PA. Furthermore, detailed information can be lost in the 3D models generated from the 4D sequence in terms of the heart's overall movement¹⁴.

In this study, a 4D beating heart model was created to observe and visualize the heart's total deformation throughout the cardiac cycle by using a mask for the 4D volume of the segmentation in 3D reconstruction software and not the conventional method of cropping the 4D volume in sheep J. This method can provide an accurate and efficient way of building a 4D model as a 3D reconstruction from a 3D sequence to visualize the heart and select the valve size. Furthermore, the same method was used to reconstruct the right heart model as a dynamic model from the segmentations in each 10% of the cardiac cycle segmented using the Grow From Seeds effect in 3D reconstruction software. The 4D right heart model can visualize the entire anatomical morphology throughout the RR interval, based on which cardiologists can develop a patient-specific strategy for TPVR. In addition, the 3D straightened right heart models obtained from the 4D sequence in each 10% of the cardiac cycle can furnish a precise, morphological, and functional quantification of the right heart, especially in the five planes applied for the stented heart valve selection. Prior to creating the straightened models, a manual and exact 3D segmentation of the right heart from each 10% cardiac cycle is required. When doing right heart segmentations, after the volume from one frame has been masked, the 3D segmentation in the current frame will emerge automatically by using the Scissor function for the undesirable structures. In order to retain the whole volume of RVOT, a tiny piece of the left coronary artery must be kept in the segmentations. To create a straightened model, it is pivotal to add a centerline into the original right heart model to ensure the quality of the straightened model and decrease the computational load. The straightened right heart model accurately reflected all the correlations of the cardiac anatomy, including perimeters, circumferences, and cross-sectional areas, allowing a subsequent extraction of morphological information and direct measurements in a holistic fashion. In this study, the measurements from the 4D straightened model resulted in the same choice of valve size (30 mm in diameter) as the 3D measurements in MPR, but with the advantages of remarkable virtual reality and reliable results in sheep J. It also enables the collection of data on right ventricular volumes during the whole cardiac cycle, which can be then applied to calculate the right ventricular ejection fraction.

Previous clinical studies have shown significant differences in the measured cross-sectional areas of RVOTPA between static and dynamic section planes secondary to large 3D displacements and rotations¹⁵. In sheep J Pre-CT, the significant differences in measured cross-sectional areas and circumferences in the RVOT plane and basal plane of the pulmonary valve were also observed in the RVOT: 3.42 cm² in 4D versus 4.28 cm² in 3D, BPV: 2.96 cm² in 4D versus 3.92 cm² in 3D, and RVOT circumferences: 76.1 mm in 4D versus 87.06 mm in 3D, BPV: 67.65 mm in 4D versus 75.73 mm in 3D. To obtain data for the measurements, the five dynamic planes were applied instead of fixed planes; here, the sinotubular plane and the basal plane of the pulmonary valve were chosen as the lines of reference. These five planes included all the space that can be utilized to deploy the stented heart valve. The RVOT plane exhibited the largest deformation throughout the cardiac cycle in the five planes, highlighting the need for a versatile TPVR device that enables adaptability to various anatomies and retains the designed geometry of the stented heart valve for long-term durability without fracture and migration. The nitinol stent with shape memory is a promising candidate for mounting a tri-leaflet valve for future TPVR. For the clinical application, especially for the patients who have had transannular patch repair or TPVR, it would need more efforts to reconstruct the anatomy as there are artifacts from the adhesion between the pericardium and myocardium, stent, and the deformed anatomy. It needs higher resolution CT data, well-developed reconstruction software, and abundant experience of CT analysis to translate this method for clinical use. But this method can be used for large animal trials as well as for the peri-operative evaluation for patients with Tetralogy of Fallot, isolated pulmonic stenosis who haven't had any open-heart surgeries or interventional therapies.

The described method for the 4D straightened model can enable accurate and visual identification and calculation of all segments of the heart from the RVOT to PA, which can help not only cardiologists to obtain a precise pre-interventional evaluation, but also cardiac engineers to innovate novel TPVR devices for future applications.

The main limitation of the methodology for the 4D straightened model measurement in this study is that the data were obtained from only one sheep pre-CT without a large sample population. Additionally, post-implantation CT imaging was not performed to follow up on the valve size and structural changes in the right heart. Lastly, for the patients who have had transannular patch repair or TPVR, it is more difficult to reconstruct the anatomy as there are artifacts from the adhesion between the pericardium and myocardium, stent, and the deformed anatomy.

Conclusion
In contrast with 3D CT, the straightened 4D reconstruction model not only enabled an accurate prediction of valve size selection for TPVR, but also provided ideal virtual reality in sheep J, and therefore will be a promising method for TPVR and the innovation of TPVR devices.

Materials

Name	Company	Catalog Number	Comments
Adobe Illustrator	Adobe	Adobe Illustrator 2021	Graphics software
Butorphanol	Richter Pharma AG	Vnr531943	0.4mg/kg
Fentanyl	Janssen-Cilag Pharma GmbH	DE/H/1047/001-002	0.01mg/kg
Glycopyrroniumbromid	Accord Healthcare B.V	PZN11649123	0.011mg/kg
GraphPad Prism	GraphPad Software Inc.	Version 9.0	Versatile statistics software
Imeron 400 MCT	Bracco Imaging	PZN00229978	2.0–2.5 ml/kg
Ketamine	Actavis Group PTC EHF	ART.-Nr. 799-762	2–5 mg/kg/h
Midazolam	Hameln pharma plus GMBH	MIDAZ50100	0.4mg/kg
Multislice Somatom Definition Flash	Siemens AG	A91CT-01892-03C2-7600	Cardiac CT Scanner
Propofol	B. Braun Melsungen AG	PZN 11164495	20mg/ml, 1–2.5 mg/kg
Propofol	B. Braun Melsungen AG	PZN 11164443	10mg/ml, 2.5–8.0 mg/kg/h
Safety IV Catheter with Injection port	B. Braun Melsungen AG	LOT: 20D03G8346	18 G Catheter with Injection port
3D Slicer	Slicer	Slicer 4.13.0-2021-08-13	Software: 3D Slicer image computing platform