Mitral paravalvular leak 3D printing from 3D-transesophageal echocardiography – proof of concept and preparation flowchart

doi:10.21203/rs.3.rs-1647198/v2

Download PDF

Article

Mitral paravalvular leak 3D printing from 3D-transesophageal echocardiography – proof of concept and preparation flowchart

https://doi.org/10.21203/rs.3.rs-1647198/v2

This work is licensed under a CC BY 4.0 License

Version 2

posted

You are reading this older preprint version

Read the latest preprint version →

Paravalvular leaks (PVLs) can be detected in almost 15% of patients after mitral valve prosthesis implantation. This complication can result in congestive heart failure and hemolysis. Despite advancements in non-invasive imaging, percutaneous closure of PVLs is not always successful. Therefore, efforts are made to improve treatment outcomes by using 3D-printed models of defects as pre-procedural support for interventional cardiologists.

Retrospectively, 3D-TEE recordings of 8 patients with clinically significant mitral PVLs were analysed. Qlab Software was used to export DICOM images of each PVL channel, including surrounding tissue. Image segmentation was performed in 3D Slicer, a free, open-source software package used for imaging research. Models were printed to actual size with the polyjet Stratasys Objet 30 printer with a transparent, rigid material. Duration of model preparation and printing, as well as total cost were calculated. Mean total time of model preparation was 430,5 ± 196 min.

3D-printing from 3D-TEE is technically feasible. Both shape and location of PVLs are preserved during model preparation and printing. It remains to be tested if 3D-printing would improve outcomes of percutaneous PVL closure.

paravalvular leak

3D-printing

echocardiography

Studies in which routine transesophageal echocardiography (TOE) was used to examine patients who underwent surgical valve replacement procedures revealed that paravalvular leaks (PVL) can be detected in approximately 10% of subjects who received an aortic prosthetic valve and in 15% of patients with a mitral prosthesis 6 months after surgery [1].

Despite the fact that most PVLs are asymptomatic and are accidentally detected during echocardiographic examinations, 1–5% of patients may have serious symptoms that require treatment [2]. Patients with clinically significant PVLs often present with signs and symptoms of congestive heart failure (90% of cases) that may be accompanied by hemolysis in 33–75% of cases [3]. PVLs are also considered to be a risk factor for infective endocarditis due to the turbulent pattern of blood flow through these intracardiac channels.

Historically, surgical intervention was the only procedure available for treatment of clinically significant PVLs. Nowadays, due to progress in interventional cardiology techniques, surgery is primarily reserved for cases that cannot be treated percutaneously (rocking valves, PVLs in active endocarditis), because surgical intervention is associated with high mortality [4]. Percutaneous treatment of aortic and mitral PVLs is a complex procedure that requires careful pre-procedural planning and good peri-procedural cooperation between an operator and echocardiographer. There is no gold standard for either the interventional techniques or the ideal imaging modalities that should be used to ensure the highest success rates.

Both 2D and 3D transesophageal echocardiography (2D-TEE and 3D-TEE) are most commonly used for diagnosis and peri-procedural management of patients with PVLs. Prior to the procedure these modalities are used to detect both number and location of PVLs as well as grade severity of regurgitation. During the procedure, TEE (especially 3D-TEE) is critical in assessment of PVL size and shape as correct measurements warrant selection of a proper closure device.

The challenges associated with PVL visualization in TTE or cardiac computed tomography (CCT) are related to a type of prosthesis (biological or mechanical), position of prosthesis and location of PVLs. Most important limitations of TEE are: color artifacts, acoustic shadowing and reverberations. Under CCT imaging, the PVLs are often poorly visualized due to artifacts from prostheses, dense sutures or calcifications. Despite advancements in echocardiographic equipment and excellent quality of images, it may not be possible to predict the final result of intervention after closure devices have been placed in the PVL channel. This phenomenon can have several reasons. The closure device may shift after being released from the delivery shaft. It is also possible to improperly select a closure device based on echocardiographical measurements alone, because this imaging modality may not adequately visualize the course and angulation of the PVL channel - in such case the device has to be removed and exchanged for a different one which makes the procedure longer and more costly. Incomplete seal of the PVL channel worsens patient’s prognosis and increases chances of post-procedural hemolysis.

Therefore, we performed a feasibility study on 3D printing of PVLs from 3D-TEE data aiming to facilitate percutaneous PVL closure procedures. 3D printing is often employed to visualize complex anatomy to plan medical procedures. However, preparation of 3D models from echocardiographic data is more challenging than from CT or MRI due to lack of full compatibility between echocardiographic DICOM format and software used to prepare stereolitography models. The 3D printing allows the operator to interact with the physical model of the defect and surrounding structures prior to the procedure in order to better appreciate the anatomy. The aim of our study was to create a simple and repeatable method that would lead from 3D-TEE to printing of a 3D PVL model comparable with initial 3D-TEE datasets.

Study design

We retrospectively analyzed 3D-TEE of 8 patients with mitral PVLs, acquired prior to percutaneous PVL closure procedures, who were admitted to our unit between 2015–2020. We included subjects with at least one mitral PVL who had a good quality 3D-TEE recording of the whole mitral prosthesis. This study was performed in line with the principles of the Declaration of Helsinki. Statement The concept of this research was presented in front of an appropriate Ethics Committee (Medical Silesia University Bioethical Committee) and was found to be exempt from ethics approval. (Decision number: PCN/0022/KB1/82/20/21)

3D-TEE image acquisition

3D-TEE imaging was performed with an iE33 or EPIQ7C ultrasound system (Philips Medical Systems, Andover, MA) with a fully sampled matrix transducer (X7-2t). All examinations were performed according to recommendations of the European Association of Cardiovascular Imaging (EACVI) under conscious sedation with diazepam administered intravenously.

At least three sets of the following 3D datasets were acquired: 1) ECG-gated, zoomed, full volume 4-beat recording of the whole mitral prosthesis during breath hold, 2) ECG-gated, zoomed, full volume 4-beat recording of the whole mitral prosthesis during breath hold with color doppler, 3) ECG-gated, zoomed, single-beat recording of the whole mitral valve in High Volume Rate (HVR) mode, 4) ECG-gated, zoomed, single-beat recording of the whole mitral valve in High Volume Rate (HVR) mode with color doppler. Sector size was carefully chosen to achieve the highest possible frame rate. Images with significant stitching artifacts were excluded.

Image transfer and segmentation

Images were transferred to the workstation and uploaded to QLab station ver 3.8.5. Quality of images was assessed and recordings with the best quality were identified. Selected non-color 3D volumes were exported in Cartesian DICOM (3DDCM) format. Next the Cartesian DICOM files were uploaded into Slicer [7], which is a free, open-source software package used for imaging research. To convert Cartesian DICOM to a standard DICOM, Philips DICOM Patcher module in Slicer was used (this step requires that the SlicerHeart and Sequences extensions are installed). Figure 1 shows DICOM data prior to conversion with a DICOM patcher module. Semi-automatic segmentation of volumetric data was performed using thresholding with subsequent manual correction. Diastolic frames were selected for best PVL channel visualization. Created model was cropped to remove expandable and excessive parts in order to lower the cost of printing and improve visibility of the region of interest. Built-in smoothing algorithms were applied. The final result was exported as a .stl file which was ready for 3D printing.

3D printing

Stl files were imported into the 3D printer software. All models were printed to actual size with the polyjet Stratasys Objet 30 printer. Rigid printing (IORA Model) and support (IORA Support) materials, produced by Isquared, were used. Printing accuracy of the Stratasys Objet 30 printer, provided by the manufacturer, for models printed with rigid materials is based on actual size of the model and is reported as maximal size deviation from the original size of the model - for model dimensions under 100 mm maximal deviation is ± 100µm and above 100 mm the deviation is ± 200µm or ± 0.06%, whichever is greater. Material usage and printing time were determined automatically. After printing was complete, support material was separated from the model with a high pressure water jet.

We calculated how much time was spent on each of the following steps: 1) Conducting a TEE examination and data transfer from the echocardiographic machine to Q-Station software, 2) Cartesian DICOM patching, segmentation and model preparation in 3D Slicer, 3) Printing of a 3D model. In addition, the amount of material used to prepare each model was established based on data reported by the printer software. For better visualization red and blue paint was used to highlight the structures of interest: mitral prostheses were painted blue and the tissue surrounding the PVL channel was painted red.

Statistical analysis

Continuous variables are expressed as mean and standard deviation or median and range depending on distribution of variables. Shapiro-Wilk test was used to check for normal distribution of variables.

Model preparation times for each model are presented in Table 1. Mean time of TEE recording and data transfer was 42,25 ± 6 min. Mean time of segmentation was 60,62 ± 12 min. Median time of printing was 284 (range 492) min. Mean total time of model preparation was 430,5 ± 196 min. Figure 2 shows photos of all printed models shot from an angle that provided the best visualization of the PVL.

Table 1

Model preparation time (in minutes)
	Model a	Model b	Model c	Model d	Model e	Model f	Model g	Model h
Recording and transfer	41	45	40	47	50	38	35	42
Segmentation	75	55	60	45	80	50	55	65
Printing	340	183	278	193	675	534	191	290
Total	456	283	378	285	805	622	281	397

In our work we demonstrated that printing of mitral PVL models based on data obtained from 3D transesophageal echocardiography is technically feasible. PVLs of various sizes and locations around the mitral annulus were selected, in each case a model was created and then successfully printed. Models were compared with original 3D datasets with and without color doppler. Both location and shapes of PVLs were preserved.

Up to date there is no established methodology for printing of mitral paravalvular leaks from echocardiographic data.

The key step in 3D model preparation is known as segmentation. During this process the structure of interest within acquired data is selected and labeled, which can be done in an automatic, semi-automatic or manual manner. The international standard to store, transmit, process and display medical data is DICOM. In contrast to other imaging modalities commonly used for medical purposes, such as CCT or MRI, the ultrasound data in the DICOM object is encrypted in vendor-specific formats. Therefore DICOM data acquired during echocardiographic examinations cannot be directly used as source data for segmentation and requires specific conversion. Nowadays, only Philips Q-Station software allows to convert DICOM data to a format that can be used for segmentation. This is done in two steps. Firstly, DICOM data is exported from the Q-Station software in a proprietary Philips Cartesian DICOM format. During this step anisotropic, frustum-shaped voxels, of which the raw 3D ultrasound data is composed of, are resampled into uniform, cube-shaped voxels. Secondly, Cartesian DICOM files are loaded into the 3DSlicer and are converted again with the SlicerHeart extension to be appropriate for the actual segmentation. Cartesian DICOM data prior to conversion with the SlicerHeart extension is presented in Fig. 1.

We measured the time required to complete each step of a 3D model preparation. The initial step is to perform TEE and then transfer all images to the Q-Station software. This step did not take longer than 50 minutes and is probably the easiest one, especially in centers with extensive experience in PVL treatment and 3D-TEE. Next cartesian DICOM files were loaded into 3D Slicer software, converted and then underwent semi-automatic segmentation with subsequent manual correction. This method of segmentation provides good balance between the quality of the created model and time required to complete the process, since the initial step that involves separating blood from tissue is conducted automatically and manual correction is only applied for the region of interest which in our case was the PVL. Duration of this step is based on user experience, in our case the mean time was 60,62 ± 12 min and longer times should be expected for novice users. The longest part of PVL model preparation is printing which is influenced by parameters such as size of the model or its orientation on the printer tray. Although significant reduction of printing times could be achieved by printing several models at once, we did not attempt this approach in order to allow for credible comparison of printing times between each model.

In all published articles cardiac computed tomography with contrast (CCT) was used as source data for model preparation with good results [8]. The advantage of CCT is that its DICOM data does not require any additional conversion and can be directly used for segmentation purposes. However, it is not possible to perform CCT of adequate quality in every patient with a PVL due to lack of cooperation, atrial fibrillation or artifacts from mechanical prostheses. In addition, CCT may be contraindicated in this population due to chronic renal disease. It should also be mentioned that analysis of CCT data is performed after the scan is completed, therefore no corrective measures can be applied if the scan is of unacceptable quality. TEE on the other hand delivers excellent temporal and spatial resolution with real-time analysis of acquired data performed by an experienced echocardiographer. Therefore, utilizing 3D echocardiography for PVL model preparation may be a preferable option in pre-procedural planning of percutaneous PVL closure. Of note, 3D TEE is a low-cost procedure with few absolute contraindications that is readily available in units that treat patients with PVLs.

It remains to be tested if pre-procedural printing of mitral PVLs would facilitate the procedure and improve technical success rates. Importantly, a recent work in which LAA closure was performed with device sizing based on 3D models prepared from 3D-TEE datasets showed superior outcomes in subjects in whom device selection was in agreement with 3D model sizes [9]. Nevertheless, despite significant technological advancements, echocardiography is not perfect in predicting procedural outcomes in percutaneous PVL closure. PVL channels are often angulated (C-shaped or S-shaped) and this can result in unexpected behavior of closure devices once these are placed in the PVL channel. Our clinical experience shows that even when a PVL channel is properly sized with echocardiography a shift in device position can occur once the device is released from the delivery cable. Thus, being able to test several devices on a 3D model prior to the procedure may lead to fewer peri-procedural complications (such as device embolization), improve success rates and lower the total cost of the procedure since improperly sized devices have to be removed from the patient and discarded.

3D-printing of mitral PVL models from 3D-TEE is technically feasible. Echocardiographic studies, routinely performed for diagnostic purposes, can serve as source data for model creation. Open-source software can be used for image segmentation and model preparation. 3D-printing can also be outsourced and therefore it is not necessary to commit to 3D-printing from the get-go by making a significant investment in a 3D-printer.

Author Contributions: “Conceptualization, Marek Jędrzejek and Michał Kozłowski; methodology, Marek Jędrzejek and Michał Kozłowski.; software, Marek Jędrzejek, Michał Kozłowski, Ewa Peszek-Przybyła.; validation, Marek Jędrzejek, Michał Kozłowski, Piotr Pysz.; formal analysis, Tomek Jadczyk, Grzegorz Smolka, Wojciech Wojakowski.; investigation, Marek Jędrzejek, Michał Kozłowski,.; data curation, Marek Jędrzejek and Michał Kozłowski.; writing—original draft preparation, Marek Jędrzejek and Michał Kozłowski.; writing—review and editing, Marek Jędrzejek, Michał Kozłowski, Tomek Jadczyk.; visualization, Marek Jędrzejek, Michał Kozłowski, Ewa Peszek-Przybyła.; supervision, Grzegorz Smolka, Wojciech Wojakowski.; project administration, Grzegorz Smolka, Wojciech Wojakowski.; funding acquisition, Grzegorz Smolka, Wojciech Wojakowski. All authors have read and agreed to the published version of the manuscript.”

Funding: This work was was co-financed by the National Center for Research and Development as part of the strategic program "Prevention and treatment of civilization diseases" Strategmed II Project "Integrated paravalvular leak closure system" - acronym VALE, contract no. STRATEGMED2/269488/7/NCBR/2015"

Institutional Review Board Consent Statement: Statement The concept of this research was presented in front of an appropriate Ethics Committee (Medical Silesia University Bioethical Committee) and was found to be exempt from ethics approval. (Decision number: PCN/0022/KB1/82/20/21)

Informed Patient consent was waived. Anonymous imaging data was used for this study therefore consent was not required.

Data Availability Statement: The datasets used and analysed during the current study are available from the corresponding author on request.

Conflicts of Interest: The authors declare no conflict of interest.

Ionescu A, Fraser AG, Butchart EGPrevalence and clinical significance of incidental paraprosthetic valvular regurgitation: a prospective study using transesophageal echocardiographyHeart 2003;89:1316-1321.
Pate GE, Al Zubaidi A, Chandavimol M, Thompson CR, Munt BI, Webb JG. Percutaneous closure of prosthetic paravalvular leaks: case series and review, Catheter Cardiovasc Interv, 2006, vol. 68 (pg. 528-533).
Chiam PT, Martinez C, Eiros R, Roubin G, Cohen HA. Clinical outcomes in patients undergoing percutaneous closure of periprosthetic paravalvular leaks, J Am Coll Cardiol, 2011, vol. 58 (pg. 2210-2217).
Emery RW, Krogh CC, McAdams S, Emery AM, Holter AR. Long-term follow up of patients undergoing reoperative surgery with aortic or mitral valve replacement using a St Jude medical prosthesis, J Heart Valve Dis, 2010, vol. 19 (pg. 473-484)
Smolka G, Pysz P, Ochała A, et al. Transcatheter paravalvular leak closure and hemolysis – a prospective registry. Archives of Medical Science. 2017;13(3):575-584. doi:10.5114/aoms.2016.60435.
Pysz P, Kozłowski M, Malczewska M, Adamczyk-Filipek E, Wojakowski W, Smolka G. Prospective registry validating the reproducibility of mitral paravalvular leak measurements in a standardized real-time three-dimensional transesophageal echocardiography algorithm for optimal choice of the closure device. Postep w Kardiol interwencyjnej = Adv Interv Cardiol [Internet]. 2019 [cited 2021 Feb 24];15(2):203–10. Available from: http://www.ncbi.nlm.nih.gov/pubmed/31497053
https://www.slicer.org/
Motwani M, Burley O, Luckie M, Cunnington C, Pisaniello AD, Hasan R, Malik I, Fraser DG. 3D-printing assisted closure of paravalvular leak. J Cardiovasc Comput Tomogr. 2020 Sep-Oct;14(5):e66-e68. doi: 10.1016/j.jcct.2019.03.008. Epub 2019 Mar 26. PMID: 30930181.
Fan Y, Yang F, Cheung GS, Chan AK, Wang DD, Lam YY, Chow MC, Leong MC, Kam KK, So KC, Tse G, Qiao Z, He B, Kwok KW, Lee AP. Device Sizing Guided by Echocardiography-Based Three-Dimensional Printing Is Associated with Superior Outcome after Percutaneous Left Atrial Appendage Occlusion. J Am Soc Echocardiogr. 2019 Jun;32(6):708-719.e1. doi: 10.1016/j.echo.2019.02.003. Epub 2019 Apr 2. PMID: 30948144.

No competing interests reported.

Download PDF

Version 2

posted

You are reading this older preprint version

Read the latest preprint version →

Mitral paravalvular leak 3D printing from 3D-transesophageal echocardiography – proof of concept and preparation flowchart

Status:

Version 2

Abstract

Figures

1. Introduction

2. Materials And Methods

3. Results

4. Discussion

5. Conclusions

Declarations

References

Additional Declarations

Status:

Version 2