DOI: 10.1148/radiol.2422051994
(Radiology 2007;242:490-497.)
© RSNA, 2007
Percutaneous Pulmonary Valve Implantation Based on Rapid Prototyping of Right Ventricular Outflow Tract and Pulmonary Trunk from MR Data1
Silvia Schievano, MEng,
Francesco Migliavacca, PhD,
Louise Coats, MBBS,
Sachin Khambadkone, MD, MRCP,
Mario Carminati, MD,
Neil Wilson, MRCP,
John E. Deanfield, FRCP,
Philipp Bonhoeffer, MD and
Andrew M. Taylor, MD, MRCP, FRCR
1 From the Cardiothoracic Unit, UCL Institute of Child Health & Great Ormond Street Hospital for Children, Great Ormond St, London WC1N 3JH, England (S.S., L.C., S.K., N.W., J.E.D., P.B., A.M.T.); Laboratory of Biological Structures Mechanics, Department of Structural Engineering, Politecnico di Milano, Milan, Italy (F.M.); and Department of Pediatric Cardiology, Istituto Policlinico San Donato, San Donato Milanese, Italy (M.C.). Received December 8, 2005; revision requested January 27, 2006; revision received February 6; accepted March 7; final version accepted June 1. S.S. and L.C. supported by the British Heart Foundation (BHF grants FS/05/039 and FS/04/008, respectively). A.M.T. supported by the Higher Education Funding Council for England (HEFCE).
Address correspondence to A.M.T. (e-mail: a.taylor{at}ich.ucl.ac.uk).
 |
ABSTRACT
|
|---|
Purpose: To determine if magnetic resonance (MR) imaging data can be used to create rigid models that are accurate representations of the right ventricular outflow tract (RVOT) and pulmonary trunk anatomy and if such models can be used to refine the selection of patients for percutaneous pulmonary valve implantation (PPVI).
Materials and Methods: Institutional review board approval and informed patient consent were obtained. Twelve patients' MR data were analyzed and elaborated for input into a rapid prototyping (RP) system. RP models were successfully built and presented to two experienced cardiologists, who were retrospectively asked if they would have attempted PPVI. Their responses were compared with the documented decisions and outcomes of PPVI.
Results: For four subjects, both cardiologists correctly determined, on the basis of MR image or three-dimensional (3D) RP model findings, that PPVI should not have been attempted. Two patients in whom PPVI was attempted were considered to be unsuitable for the procedure after balloon sizing, and in another two patients, implantation was unsuccessful because of device instability. For the four patients in whom PPVI was suitable and the four in whom it was unsuitable, observers 1 and 2 correctly determined suitability for PPVI in four and two patients, respectively, by using the MR images alone. Both observers correctly determined the suitability of five patients by using the 3D models alone.
Conclusion: Using 3D RP models resulted in more accurate selection of patients for PPVI than did using MR images.
© RSNA, 2007
Editor's note: In January 2006 (From the Editor), I announced a new section in Radiology—Innovations. Under this banner, we will publish original research that may possibly have far-reaching implications. Authors interested in having their manuscripts considered for Innovations should first read the From the Editor article to learn of more specifics. Regarding the manuscript by Schievano et al, the positive reviewer comments we received concerning its publication in Innovations included "The use of detailed models for correction of valvular abnormalities in congenital heart disease is highly innovative and new. Clinical validation of this approach requires a multidisciplinary team of specialists, and their reported success in a preliminary clinical feasibility trial is impressive," and "The benefits to patients of the technical approach using models appear superior...."
—Anthony V. Proto, MD, Editor
|
INTRODUCTION
|
|---|
Valvular heart disease is an important cause of morbidity and mortality (1). Pulmonary valve abnormalities are most common in patients with repaired congenital heart disease (2). Pulmonary incompetence or pulmonary trunk and/or conduit stenosis can occur in subjects with this abnormality and is associated with detrimental sequelae (3–5). These abnormalities have led to an increasing proportion of patients who require surgical replacement of the pulmonary valve or conduit (6–8). At present, this is performed surgically during cardiopulmonary bypass, with all of the concomitant problems related to repeat replacement surgery (9).
We recently developed a technique for transcatheter percutaneous pulmonary valve implantation (PPVI) in humans (10,11) that potentially could be used to overcome many of the disadvantages of surgical valve replacement. However, the main difficulty of PPVI is in the assessment of the implantation site before treatment. As a consequence of device design and valve availability, successful PPVI is currently limited to patients with a specific anatomy at the implantation site (diameter < 22 mm). Thus, to safely anchor the valved stent, the exact size, shape, and physical properties of the implantation site need to be known. Moreover, marked variations in the three-dimensional (3D) geometry of the right ventricular outflow tract (RVOT) and pulmonary trunk and/or conduit among individuals make preprocedural assessment and patient selection crucial.
In our practice, information about the anatomy of the RVOT and pulmonary trunk and/or conduit is acquired by using a combination of echocardiography and two-dimensional and 3D magnetic resonance (MR) imaging (11). The 3D anatomy can be defined particularly well with reconstruction of the MR angiography data set. However, in borderline patients (those with complex anatomy and an RVOT around 22 mm), it can still be difficult to appreciate the 3D anatomy presented on a two-dimensional computer screen, and we have had several patients in whom PPVI was attempted but was unsuccessful. Our experience suggests that this is due to both misinterpretation of the 3D MR data and the texture and distensibility of the RVOT. Thus, the purpose of our study was to determine if MR data can be used to create rigid models that are accurate representations of the RVOT and pulmonary trunk anatomy and if such models can be used to refine the selection of patients for PPVI.
|
MATERIALS AND METHODS
|
|---|
Patients and Phantoms
We acquired MR data in 12 patients who had been referred for possible PPVI (seven male, five female; mean age, 20 years; range, 9–39 years). Ten patients had a tetralogy of Fallot; three, a homograft conduit; three, a homograft monocusp patch; and six, a transannular patch. These patients were retrospectively selected from our PPVI program and were chosen to reflect patients with the wide range of RVOT anatomies that we see in our practice. Eight patients went on to undergo catheterization in preparation for PPVI: The outcome was successful for four of these patients and unsuccessful for four. These patients were considered to have borderline morphology. This borderline abnormality accounts for 25% of our cases when we consider patients referred for PPVI in our practice. The remaining four patients, who were turned down for PPVI on the basis of RVOT size and geometry, underwent surgery. The research ethics committees of the Heart Hospital and Great Ormond Street Hospital for Children NHS Trust approved the study, and all subjects (and/or a parent or guardian) had given informed consent for MR imaging and for use of the acquired MR data for future research.
To validate the reconstruction methodology, MR data were also acquired from five phantoms (60-mL syringe, hemispheric roll-on antiperspirant bottle top, universal hospital specimen bottle, small mint box, and plastic square salad container) filled entirely with saline and 1% gadolinium. The various shapes of these phantoms were selected to test the ability of the reconstruction algorithm by using different object curvatures. The phantom size was compared with the human RVOT anatomy to evaluate errors in the same range of dimensions.
MR Imaging
All MR examinations were performed with a 1.5-T unit (Symphony-Maestro; Siemens Medical Systems, Erlangen, Germany) by one author (A.M.T., 9 years experience in cardiovascular MR imaging). A 3D gradient-echo sequence was used to perform MR angiography after the administration of 0.4 mL of gadolinium-based contrast material (gadopentetate dimeglumine, Magnevist; Schering Healthcare, Burgess Hill, West Sussex, United Kingdom) per kilogram of body weight. Imaging parameters were 4.4/2.3 (repetition time msec/echo time msec), 12° flip angle, 1.3–1.5-mm section thickness, 256 x 512 matrix, 400–500-mm field of view, one breath hold, one signal acquired, and no electrocardiographic gating. The raw Digital Imaging and Communications in Medicine data from the MR angiograms were sent to an offline workstation for image elaboration.
Model Creation
Three-dimensional MR data reconstruction.—MR image data were reconstructed by using Mimics software (Materialise, Ann Arbor, Mich). All reconstructions were executed by one operator (S.S., 3 years experience in image processing). Image elaboration for each patient's MR data took 2–3 hours. The Digital Imaging and Communications in Medicine data were imported into the Mimics software for image processing. These data were viewed in two dimensions (transverse, coronal, and sagittal sections) and in 3D after segmentation. Segmentation masks were then used to detect the region of interest—in this case, the RVOT.
Thresholding was the first action performed to create a segmentation mask. The region of interest was selected by defining a range of gray values. The boundaries of this range were the lower and upper threshold values. All pixels with a gray value in this range were highlighted in a mask. To detect the inner arterial wall, two suitable threshold values were chosen.
Next, a region-growing algorithm was used to eliminate noise and separate structures that were not connected. Finally, manual editing functions were used to draw, erase, or restore parts of the image by clicking on single pixels. When the region of interest was completely selected, the software constructed a 3D model of the structure by means of pattern recognition and interpolation algorithms. In this way, it was possible to generate the blood volume without arterial wall thickness (ie, inner wall of RVOT). To create the 3D model of the arterial wall, a virtual wall of 2-mm constant thickness was built around the blood volume, which was subsequently deleted (Fig 1).
View larger version (32K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1: A, Original coronal, and, B, reformatted transverse MR images show segmentation mask (orange) of pulmonary trunk (A) and proximal right pulmonary artery after thresholding, region growing, and manual editing. C, Computer-generated 3D reconstruction of blood volume without arterial thickness (constant 2-mm thickness), anterior view of RVOT, and pulmonary bifurcation. D, E, Reconstructed D, anterior, and, E, lateral views of arterial wall.
|
|
The standard stereolithography solid-to-layer format of the designed structure was exported to the rapid prototyping (RP) system. This format is a triangulated surface representation of a 3D model, in which the triangular elements approximate the surface of the object. In the presence of curved surfaces, this configuration enables better approximation of the object (Fig 2).
View larger version (46K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2: Standard stereolithography solid-to-layer format of RVOT and pulmonary bifurcation; magnified area (outlined in square) is triangulated surface representation of a 3D model. These data are exported to the 3D RP printer.
|
|
RP protocol.—RP is the name given to a host of related technologies used to fabricate physical objects directly from computer-aided-design data sources. This method is unique in that materials are added and bonded in layers to form objects. Imaging data files can be input into the prototyping machine, and 3D models of any geometric complexity can be formed without the need for elaborate machine setup or final assembly. In essence, a prototyping machine can be thought of as a 3D printer.
The 3D physical models were created by using a 3D printer with a drop-on-demand technique and a thermoplastic resin (P1500 polyester; Stratatis, Eden Prairie, Minn). The machine included a nozzle driven by an x-y stage to create outlines of each layer. The thickness of the machine layer was 0.33 mm. The printer software calculated the best orientation for creation of the object and the supports for the structure required during the printing phase. The drop-on-demand ink-jet head assembly deposited fine droplets of resin, and an ultraviolet lamp immediately hardened the material. The third dimension was constructed by means of a displacement through the z-axis of the support stage. Successive layers were added until the object was completed (Fig 3). The fine-layer building capability of the machine ensured a quality finish that required no postprocessing. Building the models took 3–4 hours.
View larger version (25K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3a: (a) Schematic diagram of 3D printer nozzle moving in x-y direction to create outlines of each layer, with movement of support causing displacement through z-axis. (b) Three-dimensional printer. (c) Completed 3D model, including supporting structures.
|
|
View larger version (105K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3b: (a) Schematic diagram of 3D printer nozzle moving in x-y direction to create outlines of each layer, with movement of support causing displacement through z-axis. (b) Three-dimensional printer. (c) Completed 3D model, including supporting structures.
|
|
View larger version (85K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3c: (a) Schematic diagram of 3D printer nozzle moving in x-y direction to create outlines of each layer, with movement of support causing displacement through z-axis. (b) Three-dimensional printer. (c) Completed 3D model, including supporting structures.
|
|
Assessment of Clinical Utility
For each patient, cardiologists (L.C., S.K., N.W., J.E.D., P.B.), imagers (A.M.T.), and cardiothoracic surgeons at our institution made a consensus decision on the basis of clinical history and MR data as to whether PPVI should be attempted. The success or failure of PPVI, if attempted (in eight patients), was then regarded as the end point. If PPVI was not attempted (in four patients), this decision was regarded as the end point. Then, for all patients, we retrospectively presented the 3D MR images alone or the 3D models alone, randomly and in a blinded fashion, to two observers (cardiologists M.C. and N.W., 27 and 21 years experience in interventional cardiology, respectively); they were also blinded to the actual outcome of the procedures. The observers were given the dimensions of the RVOT and pulmonary trunk and the MR report when the MR images were presented and were given a caliper to take their own measurements when the models were presented. They were then asked to state whether they would have attempted PPVI on the basis of the 3D MR image findings alone and on the basis of the 3D model findings alone. These decisions were then compared with the documented decisions and outcomes of PPVI to evaluate the accuracy of 3D MR imaging and 3D modeling for selection of patients suitable for PPVI.
Data and Statistical Analyses
Precision of 3D data reconstruction.—There was potential for operator-introduced error during the thresholding phase of the 3D model reconstruction because the operator had to select the suitable gray threshold values to identify the inner arterial wall. To estimate the possible range of this error, each of five operators (engineers in our department [including S.S.] with 2–5 years experience in image processing) reconstructed the RVOT from the MR data set of one patient. The maximum diameter in the same sectional area of the five reconstructed models was measured and compared. These data were evaluated by using coefficient of variation analysis.
Correlation between computer-generated model and phantom.—The physical properties of each phantom—specifically, the volume of saline and gadolinium (1% dilution) in each object—were measured by one author (A.M.T.) at the time of MR imaging. The volumes of all five computer-generated phantoms were measured by one operator (S.S.) using Mimics software tools and were compared with the original volumes of fluid placed in each phantom. The phantoms were reconstructed as physical models, and their volumes (volume of water required to fill the model) were compared with the volumes of the original phantom to assess the correlation between the phantom and the RP technique. The internal diameters also were measured by using a mechanical caliper for the original and reconstructed phantoms and by using the Mimics electronic caliper for the computer-generated phantoms.
Assessment of patient models.—The narrowest diameter of the RVOT and pulmonary trunk was chosen and measured on both the 3D MR images (with electronic calipers) and the 3D physical models (with mechanical calipers).
Statistical analysis was performed by using 2003 SPSS for Windows, version 12.0.1 (SPSS, Chicago, Ill). All data were analyzed by using Spearman rank correlation and Bland-Altman plots to assess correlations. P < .05 was considered to denote a statistically significant difference.
|
RESULTS
|
|---|
Model Building
Rigid 3D models of all five phantoms and the RVOT in all 12 patients were successfully built (Figs 4–6).
View larger version (125K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 5: Rigid 3D models of five phantoms: syringe (far left), roll-on antiperspirant bottle top (top left), hospital specimen bottle (bottom left), plastic salad container (top right), and small mint box (bottom right).
|
|
Precision of 3D data reconstruction.—The coefficient of variation in diameter measurements due to MR data operator thresholding was 3.4%. For the critical diameter of 22 mm, this would lead to an average operator-induced error of approximately ±0.75 mm, which would be clinically acceptable.
Correlation between computer model and phantom.—In the phantom experiments, there was excellent correlation between the volume of gadolinium-enhanced saline that was imaged and the volume measured on both the computer-generated model and the 3D physical model (r = 0.99, P < .001, for both correlations) (Table 1). There was also excellent correlation between the diameters of each real phantom and the diameters measured on both the computer-generated phantom models and the 3D physical models (r = 0.99, P < .001, for both correlations) (Table 1).
Assessment of patient models.—For the patient studies, there was excellent correlation between the narrowest dimensions measured on the 3D MR images and 3D models (r = 0.97, P < .001) (Table 2). The Bland-Altman plot revealed a small positive systematic bias toward the 3D models (mean difference between model and image dimensions [bias on the plot], 1.9 mm) (Fig 7).
View larger version (10K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7a: (a) Plot of real phantom volume against created model volume (x = 0.99y + 0.64, where x indicates real phantom volumes and y indicates created phantom volumes; r = 0.99; P < .001). Dotted line represents line of correlation. (b) Plot of dimensions measured in patients at MR against dimensions measured in created models (x = 0.92y + 0.68, where x indicates MR dimensions and y indicates model dimensions; r = 0.97; P < .001). Dotted line represents line of correlation. (c) Bland-Altman plot of (model dimensions minus image dimensions) versus average of model and image dimensions. Two outliers, at points 5 and –2, represent 2 standard deviations from the mean. Central dotted line at approximately point 2 is the mean.
|
|
View larger version (10K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7b: (a) Plot of real phantom volume against created model volume (x = 0.99y + 0.64, where x indicates real phantom volumes and y indicates created phantom volumes; r = 0.99; P < .001). Dotted line represents line of correlation. (b) Plot of dimensions measured in patients at MR against dimensions measured in created models (x = 0.92y + 0.68, where x indicates MR dimensions and y indicates model dimensions; r = 0.97; P < .001). Dotted line represents line of correlation. (c) Bland-Altman plot of (model dimensions minus image dimensions) versus average of model and image dimensions. Two outliers, at points 5 and –2, represent 2 standard deviations from the mean. Central dotted line at approximately point 2 is the mean.
|
|
View larger version (11K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 7c: (a) Plot of real phantom volume against created model volume (x = 0.99y + 0.64, where x indicates real phantom volumes and y indicates created phantom volumes; r = 0.99; P < .001). Dotted line represents line of correlation. (b) Plot of dimensions measured in patients at MR against dimensions measured in created models (x = 0.92y + 0.68, where x indicates MR dimensions and y indicates model dimensions; r = 0.97; P < .001). Dotted line represents line of correlation. (c) Bland-Altman plot of (model dimensions minus image dimensions) versus average of model and image dimensions. Two outliers, at points 5 and –2, represent 2 standard deviations from the mean. Central dotted line at approximately point 2 is the mean.
|
|
Clinical Utility
For the four subjects who did not undergo PPVI (Table 3), both cardiologists correctly suggested, on the basis of the 3D MR image or 3D model findings, that PPVI should not have been attempted. Of the eight patients in whom PPVI was attempted, two were considered to be unsuitable for the procedure after balloon sizing and another two had unsuccessful implantation owing to instability of the device. Thus, four patients were suitable for PPVI, and four were unsuitable. For these eight patients, observers 1 and 2 correctly determined the suitability for PPVI (based on the final outcome) in four and two patients, respectively, by using the MR images alone. Their accuracy increased with use of the 3D model alone: Both of them correctly determined the suitability of five patients (Table 3).
Thus, the observers made the same decision regarding PPVI suitability during the cardiac catheterization on the basis of the known clinical outcome in 67% (eight of 12) and 50% (six of 12) of the total number of patients by using conventional MR assessment. Their accuracy improved to 75% (nine of 12) and 75% (nine of 12), respectively, with the availability of 3D models. For both patients in whom PPVI failed, both observers' decisions were 50% (six of 12 cases) correct with use of the MR images alone but 100% correct with use of the models.
|
DISCUSSION
|
|---|
Our study results demonstrate that computer-aided design combined with RP can be used to create accurate models of the RVOT and pulmonary trunk. We found that for borderline patients such models can help clinicians select those who are suitable for PPVI with increased certainty compared with the certainty in determining suitability according to conventional MR findings alone. Although RP is widely used in engineering and manufacturing industries (12), in medicine to custom make prosthetic parts (13,14) for clarification of complex surgical procedures (15,16), and in preliminary studies of the cerebral vasculature (17), to our knowledge, the application of RP to assess great vessel abnormalities with MR data had not been previously reported. A handheld model of the anatomy of interest can be easily viewed from any angle. Furthermore, we believe it is easier to measure the dimensions from models than to measure them from MR data, and this may account for the small systematic bias between the two techniques that we observed. Further studies are necessary to confirm these findings.
There were limitations to this study: First, the MR images used to build the 3D models were not electrocardiographically gated; thus, the acquired data represented an average of the RVOT–pulmonary trunk shape and size measured during the cardiac cycle. This can lead to an underestimation of the maximum systolic dimension of narrowest diameter. This was underscored in our study by the two patients in whom the cardiologists believed PPVI should be attempted but at balloon sizing were considered to be unsuitable for PPVI. This limitation will be addressed when 3D gated MR sequences become more widely available (18,19). In practice, we overcome the error introduced by the dynamic nature of the RVOT–pulmonary trunk morphology by measuring the maximum systolic dimension through the region of interest on cine images. Another method of acquiring such data may be to use electrocardiographically gated multidetector computed tomographic images.
Second, small errors were introduced by the operator performing the computer-aided design, especially during the thresholding operation. These errors were small and within an acceptable range for model building. Similar errors occur when MR images are measured and the operator makes similar threshold judgments by using the electronic calipers.
Third, rigid models are not realistic in terms of mimicking the mechanical behavior of the arterial wall, which is distensible. During virtual implantation of the valved stent into the rigid model, the virtual wall—unlike the real arterial wall—does not undergo deformation. In the future, it may be possible to use different materials that more realistically mimic the mechanical properties of the arterial wall, which could be depicted with 3D cardiac gated MR sequences (18,19). Object transparency also would be useful for better visualization if a preprocedural trial of PPVI was attempted with the model.
The patients who most commonly require pulmonary valve replacement have dilatation of the RVOT and pulmonary trunk. The majority of these patients have an RVOT–pulmonary trunk dimension greater than 26 mm and are therefore not suitable for PPVI. To extend the indications for PPVI, new techniques and devices to reduce the RVOT dimension before PPVI need to be developed, and models of dilated outflow tracts may be used to help develop them (20).
Applying modeling to other valves in the heart may be useful for planning percutaneous interventions in the future (21–23). Modeling the aortic valve, where the position of the coronary artery ostia and the relationship of the anterior leaflet of the mitral valve with the aortic valve are crucial for optimizing the safe deployment of a device, may be particularly useful.
In conclusion, 3D RP is a method of model building that enables complete appreciation of the 3D anatomy of the RVOT and pulmonary trunk and/or conduit. Such models can be used to select patients for PPVI more accurately, compared with the patient selection performed by using 3D MR images, and have the potential to aid in the design of future devices in this field.
|
ADVANCES IN KNOWLEDGE
|
|---|
- Three-dimensional models of the heart and great vessels can be created from the MR data of individual patients.
- These models are accurate depictions of the structures they represent.
- Such models can be used to refine criteria for the selection of patients for interventional cardiac catheterization procedures and surgery.
|
ACKNOWLEDGMENTS
|
|---|
The authors thank Giuseppe Sala, PhD, Silvio Ferragina, MEng, and Antonio Armillotta, PhD, from the Aerospace and Mechanical Engineering Departments of Politecnico di Milano for their technical support in the RP printing.
|
FOOTNOTES
|
|---|
Abbreviations: PPVI = percutaneous pulmonary valve implantation • RP = rapid prototyping • RVOT = right ventricular outflow tract • 3D = three-dimensional
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, A.M.T.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, S.S., A.M.T.; clinical studies, S.S., L.C., S.K., M.C., N.W., P.B., A.M.T.; experimental studies, S.S., F.M., A.M.T.; statistical analysis, S.S., A.M.T.; and manuscript editing, all authors
|
References
|
|---|
- Wren C, O'Sullivan JJ. Survival with congenital heart disease and need for follow up in adult life. Heart 2001;85:438–443.[Abstract/Free Full Text]
- Nollert G, Fischlein T, Bouterwek S, et al. Long-term results of total repair of tetralogy of Fallot in adulthood: 35 years follow-up in 104 patients corrected at the age of 18 or older. Thorac Cardiovasc Surg 1997;45:178–181.[Medline]
- Horneffer PJ, Zahka KG, Rowe SA, et al. Long-term results of total repair of tetralogy of Fallot in childhood. Ann Thorac Surg 1990;50:179–183.[Abstract]
- Chan KC, Fyfe DA, McKay CA, Sade RM, Crawford FA. Right ventricular outflow reconstruction with cryopreserved homografts in pediatric patients: intermediate-term follow-up with serial echocardiographic assessment. J Am Coll Cardiol 1994;24:483–489.[Abstract]
- Murphy JD, Freed MD, Keane JF, Norwood WI, Castaneda AR, Nadas AS. Hemodynamic results after intracardiac repair of tetralogy of Fallot by deep hypothermia and cardiopulmonary bypass. Circulation 1980;62(2 pt 2):I168.[Medline]
- Yemets IM, Williams WG, Webb GD, et al. Pulmonary valve replacement late after repair of tetralogy of Fallot. Ann Thorac Surg 1997;64:526–530.[Abstract/Free Full Text]
- Oechslin EN, Harrison DA, Harris L, et al. Reoperation in adults with repair of tetralogy of Fallot: indications and outcomes. J Thorac Cardiovasc Surg 1999;118:245–251.[Abstract/Free Full Text]
- Hazekamp MG, Kurvers MM, Schoof PH, et al. Pulmonary valve insertion late after repair of tetralogy of Fallot. Eur J Cardiothorac Surg 2001;19:667–670.[Abstract/Free Full Text]
- Parr GV, Kirklin JW, Blackstone EH. The early risk of re-replacement of aortic valves. Ann Thorac Surg 1977;23:319–322.[Abstract]
- Bonhoeffer P, Boudjemline Y, Qureshi SA, et al. Percutaneous insertion of the pulmonary valve. J Am Coll Cardiol 2002;39(10):1664–1669.[Abstract/Free Full Text]
- Khambadkone S, Bonhoeffer P. Nonsurgical pulmonary valve replacement: why, when, and how? Catheter Cardiovasc Interv 2004;62:401–408.[CrossRef][Medline]
- Yan X, Gu P. A review of rapid prototyping technologies and systems. Comput Aided Design 1996;28(4):307–318.[CrossRef]
- Cheah CM, Chua CK, Tan KH. Integration of laser surface digitizing with CAD/CAM techniques for developing facial prostheses. II. Development of molding techniques for casting prosthetic parts. Int J Prosthodont 2003;16(5):543–548.[Medline]
- Salako F, Aubin CE, Fortin C, Labelle H. Feasibility study of patient-specific surgical templates for the fixation of pedicle screws. Stud Health Technol Inform 2002;88:419–422.[Medline]
- Muller A, Krishnan KG, Uhl E, Mast G. The application of rapid prototyping techniques in cranial reconstruction and preoperative planning in neurosurgery. J Craniofac Surg 2003;14(6):899–914.[CrossRef][Medline]
- Wagner JD, Baack B, Brown GA, Kelly J. Rapid 3-dimensional prototyping for surgical repair of maxillofacial fractures: a technical note. J Oral Maxillofac Surg 2004;62(7):898–901.[CrossRef][Medline]
- Knox K, Kerber CW, Singel SA, Bailey MJ, Imbesi SG. Rapid prototyping to create vascular replicas from CT scan data: making tools to teach, rehearse, and choose treatment strategies. Catheter Cardiovasc Interv 2005;65(1):47–53.[CrossRef][Medline]
- Razavi RS, Hill DL, Muthurangu V, et al. Three-dimensional magnetic resonance imaging of congenital cardiac anomalies. Cardiol Young 2003;13:461–465.[Medline]
- Sørensen TS, Körperich H, Greil GF, et al. Operator-independent isotropic three-dimensional magnetic resonance imaging for morphology in congenital heart disease: a validation study. Circulation 2004;110:163–169.[Abstract/Free Full Text]
- Boudjemline Y, Agnoletti G, Bonnet D, Sidi D, Bonhoeffer P. Percutaneous pulmonary valve replacement in a large right ventricular outflow tract: an experimental study. J Am Coll Cardiol 2004;43(6):1082–1087.[Abstract/Free Full Text]
- Boudjemline Y, Bonnet D, Sidi D, Bonhoeffer P. Percutaneous implantation of a biological valve in the aorta to treat aortic valve insufficiency: a sheep study. Med Sci Monit 2002;8(4):BR113–BR116.[Medline]
- Boudjemline Y, Bonhoeffer P. Steps toward percutaneous aortic valve replacement. Circulation 2002;105(6):775–778.[Abstract/Free Full Text]
- Boudjemline Y, Bonhoeffer P. Images in cardiovascular medicine: percutaneous aortic valve replacement in animals. Circulation 2004;109(10):e161.[Free Full Text]
This article has been cited by other articles:
|
|
|
|
 
A. Frigiola, V. Tsang, J. Nordmeyer, P. Lurz, C. van Doorn, A. M. Taylor, P. Bonhoeffer, and M. de Leval
Current approaches to pulmonary regurgitation
Eur. J. Cardiothorac. Surg.,
September 1, 2008;
34(3):
576 - 581.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
M. S. Kim, A. R. Hansgen, O. Wink, R. A. Quaife, and J. D. Carroll
Rapid Prototyping: A New Tool in Understanding and Treating Structural Heart Disease
Circulation,
May 6, 2008;
117(18):
2388 - 2394.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
P. Lurz, L. Coats, S. Khambadkone, J. Nordmeyer, Y. Boudjemline, S. Schievano, V. Muthurangu, T. Y. Lee, G. Parenzan, G. Derrick, et al.
Percutaneous Pulmonary Valve Implantation: Impact of Evolving Technology and Learning Curve on Clinical Outcome
Circulation,
April 15, 2008;
117(15):
1964 - 1972.
[Abstract]
[Full Text]
[PDF]
|
|
|
|
|
|
|
R. Quaden and G. Lutter
Editorial comment: Resection and new stents before valved stent implantation
Eur. J. Cardiothorac. Surg.,
August 1, 2007;
32(2):
295 - 295.
[Full Text]
[PDF]
|
|
|
TOP
ABSTRACT
INTRODUCTION
