If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Cardiothoracic Surgery, Maastricht University Medical Center, Maastricht, The NetherlandsFaculty of Health, Medicine and Life Sciences (FHML), Cardiovascular Research Institute Maastricht (CARIM), Maastricht, The Netherlands
Department of Cardiothoracic Surgery, Maastricht University Medical Center, Maastricht, The NetherlandsFaculty of Health, Medicine and Life Sciences (FHML), Cardiovascular Research Institute Maastricht (CARIM), Maastricht, The Netherlands
The aim of this study was to develop a high-fidelity minimally invasive mitral valve surgery (MIMVS) simulator.
The process of industrial serial design was applied based on pre-set requirements, acquired by interviewing experienced mitral surgeons. A thoracic torso with endoscopic and robotic access and disposable silicone mitral valve apparatus with a feedback system was developed. The feedback system was based on 4 cameras around the silicone valve and an edge detection algorithm to calculate suture depth and width. Validity of simulator measurements was assessed by comparing simulator-generated values with measurements done manually on 3-dimensional reconstructed micro-computed tomography scan of the same sutures. Independent surgeons tested the simulator between 2014 and 2018, whereupon an evaluation was done through a questionnaire.
The feedback system was able to provide width and depth measurements, which were subsequently scored by comparison to pre-set target values. Depth did not significantly differ between simulator and micro-computed tomography scan measurements (P = .139). Width differed significantly (P = .001), whereupon a significant regression equation was found (P < .0001) to calibrate the simulator. After calibration, no significant difference was found (P = .865). In total, 99 surgeons tested the simulator and more than agreed with the statements that the simulator is a good method for training MIMVS, and that the mitral valve and suture placement looked and felt realistic.
We successfully developed a high-fidelity MIMVS simulator for endoscopic and robotic approaches. The simulator provides a platform to train skills in an objective and reproducible manner. Future studies are needed to provide evidence for its application in training surgeons.
This high-fidelity simulator provides residents and surgeons with a reproducible and objective platform to acquire the necessary skills for minimally invasive mitral valve surgery (MIMVS), shorten the learning curve, get acquainted to new techniques, simulate complex cases, and provide objectification of skill assessment, all of which may subsequently enhance the reproducibility, safety, and efficacy of MIMVS.
Nevertheless, surgical volume is mandatory to achieve proficiency and allow concurrent successive repair. This is best demonstrated by the observation that surgeon-specific mitral valve volume is independently associated with a greater probability of repair.
In addition, with the current limited working hours and rapid introduction of new surgical techniques, training is likely to move outside the operating theater, where new educational tools could be used to overcome the learning curve. Simulators can provide a full-spectrum solution for these dilemmas. Simulation-based training may subsequently allow training of residents to achieve proficiency in basic skills within shorter training periods, whereas practicing surgeons can be swiftly trained in novel techniques.
Simulator training has been well described for laparoscopic surgery and proven to be effective by reducing the learning curve of in vivo procedures.
Although high-level evidence of simulator training is mainly reported for laparoscopic surgery, the basic idea of simulation training and its effectiveness remains. In cardiothoracic surgery, simulation-based training has been proposed as an approach to augment proficiency in basic skills and new techniques.
reported the usability of their simulator for endoscopic minimally invasive mitral valve surgery (MIMVS). In addition, all of the aforementioned studies reported the use of low-fidelity simulators and/or models that can be seen as a black box without feedback. In comparison, a high-fidelity simulator may provide objective, real-time, and reproducible feedback, which can be used to guide skills development and assess the level of proficiency. The aim of this study is to demonstrate the development of a high-fidelity minimally invasive mitral valve simulator that provides real-time feedback and a reproducible platform for the training of surgical skills in a repeated and objective manner.
Simulator development was performed according to the concept of serial industrial design. The first step includes a problem and function analysis in which simulator requirements are analyzed and summed. followed by prototype development. This prototype is then tested by experts in the field to direct improvements and changes for its final version. Subsequent to production of the actual simulator, it is important to assess whether simulator provided measurements are accurate (ie, does it measure what it has to measure), performed by calibration and validation. The final, most critical step focuses on determining the simulator's added value to traditional training methods. This study directs its focus towards the first two steps. Research ethics board approval was waived, as this study did not incorporate patient data.
Simulator requirements were obtained by interviewing experienced mitral valve surgeons and by reducing the endoscopic minimally invasive mitral valve repair procedure to its essential components. The resulting list of requirements and their rationale is shown in Table 1.
Table 1Simulator requirements and their rationale
The simulator must contain a manageable size for easy transport and storage.
Multiple repair possibilities
A left ventricular model with papillary muscles, mitral valve, and left atrial model should be created to allow simulation of different repair techniques (eg, placement of neochordae or an annuloplasty ring).
Realistic suturing experience
To facilitate a true suturing experience, manufactured models should mimic in vivo tissue properties.
As simulation requires suture placement, models should be reusable, easily replicable, and replaceable.
To remain close to the in vivo anatomical dimensions, 1:1 models must be created. In addition, to provide a lifelike appearance, a close-to-nature thoracic representation with MIMVS access port should cover the simulator. The created mitral models should be accurately positioned within the simulator to mimic its in vivo angulation and 3-dimensional position.
Objective and real-time feedback about suture depth and suture width is required to provide a direct learning experience. By this method, the end-user will be able to develop and objectively assess their MIMVS skills. Subsequently, diminishing their associated learning curve. Generated feedback could also provide a platform for uniform skill's assessment.
A user interface that allows the user to pre-set the target suture depth and with before any suturing should be created. The user interface should additionally provide objective feedback regarding each suture attempt, get scoring regarding the exercise and be able to visualize the suture track.
Multiple surgical approaches
The simulator should be designed as such that one could train minimally invasive mitral valve surgery through direct vision, endoscopic approach, and robotic-assistance.
The core of the simulator (mitral valve apparatus) should be moveable to mimic the access difficulty.
Before production of the actual simulator, a prototype was developed based on preset requirements.
The geometry of left-sided heart structures can substantially differ among individuals with mitral regurgitation by which no known model is able to resemble all mitral regurgitation variations. Therefore, a characteristic regurgitation model with slightly dilated atrium and mitral valve was chosen. Dimensions were retrieved from the available literature.
The mitral valve, left atrium, and left ventricle with papillary muscles models were subsequently created from dedicated polyurethane-foam, which allows submillimeter precision (Figure 1, A and B). Positive foam models were then used for negative mold production, whereas molds were cast with a dedicated mix of silicones. Silicone for the mitral valve should provide a true-to-nature suturing experience while remaining glass-like transparency and maintaining a distinct level of toughness to prevent the silicone from tearing on suture placement. After multiple tests and third-party consultations, a weight-based silicone mixture of Sorta-Clear 18 (Smooth On Inc, Easton, Pa) and Slacker (Smooth-On Inc, Macungie, Pa) was chosen and molds were cast. The mitral valve was cast in fully transparent silicone, whereas ventricle and atrial models were cast in tissue-pink silicone after addition of a silicone dye. Inner geometry of the mitral model was subsequently colored by hand to depict atrial (pink) and valvular tissue (white) (Figure 1, C).
A hollow, aluminum housing was created as a basis for the feedback system and to fit the silicone models. Four cutouts were made in the housing's straight sides (90° apart) to fit 4 high-resolution cameras (XIMEA GmbH, Münster, Germany). Resolution of the cameras was 1.3 megapixels (1280 × 1024 pixels) at 60 frames per second (Figure 2, A and B).
To be able to detect and measure depth of sutures, a software algorithm and graphical user interface was developed in MATLAB (Release 2015b, The MathWorks Inc, Natick, Mass). This optical analysis algorithm initially differentiates between new and already placed sutures by subtracting a reference (ie, background) image. The algorithm is subsequently able to automatically weigh the camera acquisition in which the newly placed suture can best be seen. To detect suture depth, an edge detection algorithm with high signal-to-noise ratio was used. Following proper edge detection, a binary image was created from which suture depth was calculated in pixels and converted to millimeters (Figure 3). Calculated depth was provided as on-screen feedback using a graphical user interface.
A thoracic representation with endoscopic access port was added to provide a lifelike appearance (Figure 2, C and D). A 4-cm access port was created at the level comparable with the fourth intercostal space. In this way, the simulator is able to closely mimic MIMVS, given the unique surgeon's angle of view and working space for long-shafted instruments. It is known that more-experienced surgeons use smaller ports; however, to be able to provide simulation for the full range of users, a 4-cm port was created. Nevertheless, to simulate more difficult procedures, one may partially tape the access port. Finally, a camera and white LED lightning strips were mounted inside the thoracic cavity to provide and mimic the on-screen endoscopic view. In this way, no external endoscopic station is required for simulation.
The feedback system and thorax model were subsequently mounted on a mobile table to allow easy transport. Target position of the feedback system was derived from an in-house database of computed tomography (CT) images of patients with a right-sided pneumothorax, as is induced during MIMVS. As a last step, all components were assembled, including a computer and monitor to run the feedback-algorithm, provide a real-time on-screen endoscopic view, and provide suture feedback. The resulting prototype simulator is seen in Figure 2.
From prototype to actual simulator
Prototype testing and evaluation was performed by experienced cardiothoracic surgeons. Subsequently, adjustments were made. To ease physical stress and provide a proper ergonomic position, the simulator was mounted on a height-adjustable table. In addition, a set of adjustable screws and bolts were added to permit adjustment of the mitral housing's position. This adjustable configuration allows simulation of different levels of complexity (eg, increased access port – mitral valve distance).
Software was initially developed using MATLAB (Release 2015b, The MathWorks Inc). However, for the second generation, software was written in C++ to allow license-free distribution and the feedback algorithm was sped up to provide direct, real-time feedback. In addition, several novel features were added to the fundamental feedback algorithm and user interface. An algorithm to measure suture width was added, alongside the ability for users to define the intended suture depth and width and its allowable margin of error (Figure 4, A). Entered values are used by the system to provide value-based, objective feedback (eg, “suture width spot on”) and calculate an overall score (Figure 4, B). In addition, users are given the ability to login using their self-chosen user credentials and track their progression over time (Figure 4, C).
To be able to simulate different minimally invasive approaches, 3 trocar-like ports were added to the thoracic model (Figure 5, A and B). These provide the ability to simulate robotic mitral valve repair by for example using the Da Vinci surgical system. However, because most robotic systems possess their own lightning source and camera, a switch was added to the LED strips and the endoscopic view camera was made detachable using magnets. To adhere to the replicability and replaceability of simulator components, the feedback system's housing was produced in more robust manner (Figure 5, C and D) and a 3-dimensional (3D) printed, hard-plastic ventricle model with magnetic, silicone papillary muscles was created (Figure 5, E and F). These papillary muscles can subsequently be replaced when worn out without having to replace the entire left ventricle. In addition, an aluminum mitral valve model mold was manufactured, which provides greater-quality silicone castings (Figure 1, D) and allows inexhaustible reproduction.
Calibration and Validation of Simulator Measurements
Different silicone valves were mounted into the simulator, whereupon simple interrupted annuloplasty sutures were placed. After recording simulator-measurements, the silicone valves including placed sutures were placed into a micro-CT-scanner. Acquired CT scans were exported to a dedicated 3D software system for image processing (Vesalius3D; PS-Medtech, Amsterdam, The Netherlands). Subsequently, 3D-reconstructed models of each silicone valve were created and suture depth and width was measured. Validity of simulator's measurements (ie, depth and width) was assessed by comparing simulator-generated values to those acquired through micro-CT acquisitions.
Evaluation by independent operators
Independent surgeons were given multiple opportunities to test the simulator during various European Association for Cardio-Thoracic Surgery events and courses in the period from 2014 to 2018. Following simulator-training, surgeons were asked to provide evaluation of the MIMVS simulator through a questionnaire.
The questionnaire used a 1- to 5-point Likert-scale to assess the level of agreement to 6 statements. The first statements addressed the look and feeling of the disposable, silicone mitral valve, whether the relation of the mitral valve to the thoracic torso was realistic, and if performing of sutures felt realistic. The last 2 statements assessed whether surgeons found the simulator a good method for training of MIMVS and if the surgeon became more aware of the skills required to perform MIMVS.
Data were analyzed using SPSS 25 (IBM Corp, Armonk, NY). Paired measurements were subject to the Wilcoxon-signed rank test to assess potential differences between measurements. If statistically significant, calibration was required. For calibration, an orthogonal regression analysis was conducted to determine a formula, which may compensate for measurement errors. Following calibration, the Wilcoxon-signed rank test was performed again as a check of rigor for remaining differences. All values were denoted as median and interquartile range. A P value ≤.05 was considered to be statistically significant.
A 1:1 high-fidelity minimally invasive mitral valve simulator was created to practice endoscopic and robotic mitral valve surgery (Figure 5). The simulator encompasses close-to-nature silicone mitral valve and papillary muscle models, which provide a true suturing experience and haptic feedback. Moreover, the developed feedback system allowed users to objectively assess their skills and guide their development. The algorithm determined suture depth and width by placing a rectangle around the edge-detected suture, from which the distances between opposite sides were calculated (Figure 4, D). Simulator and CT measurements were performed for a total of 15 sutures, divided over 8 mitral valve models. The median suture depth was 3.00 ± 1.00 and 3.30 ± 1.10 mm. for simulator and CT measurements, respectively. In addition, no statistically significant difference was found (P = .139). Median suture width was 9.00 ± 2.00 and 11.26 ± 2.50 mm. for simulator and CT measurements, respectively. Wilcoxon-signed rank test revealed a statistically significant difference between both measurements (P = .001). Subsequently, a simple orthogonal regression was calculated to predict CT measured width values (ie, true width values) based on width values measured by the simulator. A significant regression equation was found (F[1,13] = 28.120, P < .0001), with an R2 of 0.684. Predicted CT values are equal to 3.49 + 0.89 × (simulator measured width) millimeters. Simulator width measurements were adjusted by this formula. Median adjusted simulator width measurement was 11.48 ± 1.77 mm. Comparing the adjusted with CT measurements, no statistically significant difference was found (P = .865) and calibration was deemed successful. Figure 6 shows simulator and CT width measurements before and after calibration.
Independent Operator Evaluation
A total of 99 independent surgeons tested the simulator and completed the questionnaire. Twelve were experts in the field of MIMVS, whereas 87 were senior surgeons with variable levels of experience in mitral valve surgery. The statements and their corresponding median ± interquartile range of the 1- to 5-point Likert-scale are reported in Table 2. From this table, it is seen that all surgeons more than agree with all statements.
Table 2The questionnaire statements, which were scored on a 1- to 5-point Likert scale
Response (median ± IQR)
The mitral valve in the simulator looked realistic.
5.00 ± 1.00
The mitral valve in the simulator felt realistic.
4.00 ± 1.00
The relation of the mitral valve to the thorax was realistic.
5.00 ± 1.00
Performing sutures on the valve felt realistic.
4.00 ± 1.00
The simulator is a good method for training minimally invasive mitral valve repair.
5.00 ± 0.00
I am more aware of the skills needed to perform minimally invasive mitral valve repair.
This study demonstrates the development and validation of a transportable, reusable, and high-fidelity MIMVS simulator on which almost all types of surgical techniques, in both endoscopic and robotic approaches could be practiced (Video 1). To enhance reusability, the simulator is made up of static and replaceable components. Static parts encompass the thoracic model, 3D-printed left ventricle, and feedback system. The latter provides objective, real-time feedback on suture depth and width. The margin of error and mitral valve position can be specified and adjusted in advance to adjust the level of complexity. Replaceable mitral valve and papillary muscles are composed of tissue-mimicking silicone that permits reuse of up to 4 procedures with feedback and dozens of procedures without feedback. Total costs of these models were €90 (±110 US$).
The simulator may also be used to acquire basic repair skills used in conventional median sternotomy. This is achieved by opening the lid of the thoracic model that covers the simulator's core.
Simulation-based learning could shorten the typical MIMVS learning curve; however, learning curves may vary markedly among surgeons, subsequently drawing the need for proper monitoring and skill assessment in the initial phase.
Our developed simulator may provide a first step to objectify this skill assessment by providing feedback and an overall score on suture placement. Such measurable indicators may in the future also be used to guide trainees or residents toward a goal of ideal suture width and depth, demonstrate proficiency, and maintain certification within the field of MIMVS. The simulator may also improve the level of overall skill in the surgical field. By practicing on the simulator, trainees get exposed and used to the restricted working space, to working with long-shafted instruments, to endoscopic vision, and several other basic skills.
Despite all these advantages, the field of surgical simulation within cardiothoracic surgery is still quite young, which leaves various challenges to be tackled before full acceptance and implementation can be reached. As commented by Feins,
simulation will bring a conflict between patient care demands and education to the forefront. Meeting patient care demands is already stressed by restricted work hours and will be even further stressed by the introduction of simulator-based training for residents. As a consequence, adequate funding is required to compensate for this loss of resident patient care services. Another persistent concern within the topic of simulation is the question of who will be supervising and mentoring the learning on a simulator.
The simulator is not able to provide feedback on correctness of suture positioning and can only provide information about its depth and width. Consequently, students or residents should first be taught on the optimal suture positioning, followed by hands-on simulator practice under expert supervision, who can provide a frame of reference. The simulator can subsequently guide trainees toward the technical goal of placing a suture at any width or depth.
The authors also developed the process to create patient-specific, silicone mitral valve models from 3D transesophageal echocardiographies. These models can be placed in the simulator for presurgical planning and practice of complex cases. This allows the operating surgeon to choose and test the best strategy for repair.
Pathologic models can also be used for training of residents and novice surgeons to enhance reproducibility, safety, and effectiveness. It has, however, to be noted that if leaflet resection is performed, they are one-time use only.
To maintain cohesion among the growing literature on surgical simulation in cardiothoracic surgery, Trehan and colleagues
developed a simulator classification. The presented simulator is categorized as human patient simulator, while it fuses physical components with a computer interface, can include patient variation (mounting patient-specific valves into the simulator), and has assessment and feedback capabilities.
Simulator measurements of the presented simulator have been calibrated and validated; however, its use for skills training in MIMVS has not been validated yet and should subsequently be the subject of future research. Methods to assess and validate such simulators have not been standardized. However, validation should ideally be performed with a cohort of trainees who lack experience in MIMVS to assess differences in their associated learning curve. This cohort should be divided into those who were and were not exposed to simulator training in addition to traditional training methods. In addition, one may also aim to validate surgical simulation training by comparing the level of skill before and after simulator training, in terms of time needed for correct suture placement, width, depth, etc.
In conclusion, this novel, portable, reusable, and high-fidelity MIMVS simulator for endoscopic and robotic approaches provides residents and surgeons with a reproducible simulation platform to acquire basic skills, shortens their learning curve, helps them become acquainted to newly introduced techniques, and simulates complex cases. Although simulation-based training intuitively seems to have significant advantages, there is little high-level evidence in cardiothoracic surgery. Future research should subsequently focus to validate this type of surgical training.
Conflict of Interest Statement
Peyman Sardari Nia is the inventor of the simulator that is commercialized by Maastricht University Medical Center, Maastricht, The Netherlands. All other authors have nothing to disclose with regard to commercial support.
We gratefully acknowledge R. Verhoeven, R. Van Veen, S. Overeem, and J. Riddenhof from the Technical Medicine of University Twente, Enschede, The Netherlands for their contribution to the development of the first prototype in 2013. We also gratefully acknowledge Instrument Development, Engineering & Evaluation (IDEE), Maastricht University, Maastricht, The Netherlands for their technical support for final design in 2014.
Funding for simulator development was provided by Maastricht University Medical Center, Maastricht, The Netherlands.
The first author invented the simulator to start the endoscopic mitral valve program in 2013 and to maintain the necessary skills during the start-up of the program. The conceptualization, prototyping, and further realization of the project were done before any commercialization. Therefore, the data collection, interpreting and writing the manuscript were not influenced by commercialization. In addition, results, interpretation, and writing of the manuscript were shared by 2 other co-authors, who have no conflict of interest.
The past decade has seen tremendous growth in the role of simulation in cardiothoracic surgery training. Simulators may have a role in teaching and shortening the learning curve associated with complex cardiothoracic procedures.1 In parallel, minimally invasive mitral valve surgery (MIMVS) is increasingly accepted as a feasible and efficacious approach for patients with mitral valve disease. What has been lacking, especially in the field of MIMVS, is a high-fidelity simulator that provides objective feedback on components of the operation.