Colonoscopy procedure simulation: virtual reality training based on a real time computational approach
© The Author(s) 2018
Received: 1 November 2017
Accepted: 8 January 2018
Published: 25 January 2018
Colonoscopy plays an important role in the clinical screening and management of colorectal cancer. The traditional ‘see one, do one, teach one’ training style for such invasive procedure is resource intensive and ineffective. Given that colonoscopy is difficult, and time-consuming to master, the use of virtual reality simulators to train gastroenterologists in colonoscopy operations offers a promising alternative.
In this paper, a realistic and real-time interactive simulator for training colonoscopy procedure is presented, which can even include polypectomy simulation. Our approach models the colonoscopy as thick flexible elastic rods with different resolutions which are dynamically adaptive to the curvature of the colon. More material characteristics of this deformable material are integrated into our discrete model to realistically simulate the behavior of the colonoscope.
We present a simulator for training colonoscopy procedure. In addition, we propose a set of key aspects of our simulator that give fast, high fidelity feedback to trainees. We also conducted an initial validation of this colonoscopic simulator to determine its clinical utility and efficacy.
Colorectal cancer (CRC) ranks as one of the most prevalent and significant causes of morbidity and mortality in the developed world . This disease mostly develops from adenomatous polyps , which develop into CRCs. If their progress is unchecked, CRCs can aggressively penetrate the colonic wall and metastasize to vulnerable locations such as the liver or lungs through vascular and lymphatic spread . As a result, regular screening, early diagnosis and effective treatment is paramount to decrease the mortality and morbidity caused by this disease. Lower gastrointestinal endoscopy (i.e. colonoscopy and sigmoidoscopy) is a key player in the management of patients at risk of CRC. It is used as screening technique alongside faecal occult blood testing (FOBT), double contrast barium enema and computer tomographic colonoscopy (CTC) (among other tests) . It can also diagnose malignancy or pre-malignant adenomatous polyps through biopsy and polypectomy of suspicious polypoid masses. It can also prophylactically treat CRC through polypectomy and periodic surveillance .
It has been suggested that primary care physicians perform colonoscopies in order to better manage the increasingly high burden of disease caused by CRC . It seems reasonably foreseeable that the demand for colonoscopy training will increase not only there, but throughout the developed world. It is therefore desirable and expedient that physicians become proficient in colonoscopy as quickly and as easily as possible. We present details of the colonoscopic procedures, how gastroenterologists are trained and our proposed simulation training techniques as follows.
In colonoscopy, a scope measuring approximately 7–9 mm in diameter is inserted into the patient’s rectum and advanced into the colon up until the caecum/terminal ileum. The endoscopist can either sit or stand while using pushing, pulling, torsion, hooking and sliding movements in order to advance the colonoscope throughout the colon . The colonic mucosa is constantly examined through the use of a camera on the distal tip of the colonoscope. In the case that the view is obstructed, the endoscopist can use suction or air insufflation to clear the view . In the case that the endoscopist encounters a polyp, the endoscopist can biopsy it and use forceps/snare to excise it from the mucosa (polypectomy).
Change in teaching paradigms
There has been a movement away from the traditional ‘see one, do one, teach one’ model to the use of virtual simulators to train colonoscopy. This is for several reasons: (1) colonoscopy is a difficult and time consuming procedure to master, with previous studies suggesting that up to 700 performed procedures are required to gain proficiency ; (2) it is ethically questionable to train colonoscopy on real patients, as a poorly performed colonoscopy can have very serious complications including perforation, bacteriaemia and haemorrhage ; (3) it is an expensive model as the productivity of qualified endoscopist suffers while supervising trainees compared with performing procedures independently. It should be noted that virtual simulators are expensive and are only a valuable investment if proven to be valid and high fidelity ; (4) the traditional model does not offer an objective standard of success in achieving proficiency in colonoscopy. In a clinical environment where the colonoscopist’s skill and style can have wide-ranging variations that influence the quality and safety of their colonoscopy procedures;  colonoscopy simulators can aid in providing objective parameters that can standardise optimum techniques and strategies; (5) the skills gained by using colonoscopy simulators seem to be transferable to real life patient situations  and retainable for at least 4 months . Furthermore, an attempt has been made to establish competency criteria for training using simulators . Due to the above reasons, it is reasonable to encourage this paradigm shift towards simulation environments for training novices where possible.
State of virtual simulators
Virtual simulators provide safe, realistic environments for trainees to learn the skill of colonoscopy. Trainees can understand 3D relationships between anatomical structures in the colon and develop the haptic control and finesse required to operate the colonoscope. Intraoperative events such as perforation or haemorrhage can be managed in a safe and low-stress environment without putting a patient at risk. Furthermore, the progress of trainees can be objectively measured by several parameters including percentage of the mucosa visualised, time taken for caecal intubation, level of patient discomfort and number of intraoperative events/errors. Simulators may also offer the opportunity for experienced endoscopists to rehearse colonoscopy procedures with patient-specific data in order to foresee any potential challenges during the real colonoscopy.
Several virtual reality colonoscopy simulators and diverse haptic interfaces have been proposed and developed. Yi et al. proposed a simulator whose haptic interface allowed for the ‘jiggling’ motion in colonoscopy to straighten the colonoscope and shorten the bowel . De Visser et al. incorporated loop formation into their simulator by modelling tissues surrounding the colon . Samur et al. instrumented a real clinical colonoscope in their haptic interface for improved haptic fidelity . US company Simbionix offers the commercial product GI Mentor 2 which includes features like the scope indicator pain indicator and simulates cases based on real patient data . The CAE EndoVR (previously Accutouch) allows trainees to perform simulation of polypectomy, biopsy and haemostasis . Olympus offers the Endo TS-1 commercial colonoscopy simulator which provides a real time simulation for trainees . King et al. recently developed a low-cost physical simulator with the aim to be more accessible and affordable for residency programs . Wong et al. also mentioned other development of biomedical devices .
Model surrounding organs to allow loop formation and simulate the dynamic nature of the bowel.
For haptic interface, an actual colonoscopy should be used to allow better haptic fidelity and better likeness to real life colonoscopy.
Create cases and histories that vary in difficulty and also have randomly generated cases to prevent trainees from ‘gaming’ the simulator.
Includes translational motion that shortens colon and also yaw and pitch movements to the tip of the colonoscope for better haptic fidelity.
Integrate procedures such as biopsy, haemostasis and polypectomy into the simulation environment as they are necessary for proficiency in colonoscopy.
In this paper, we will present a real-time, valid simulation environment for colonoscopy training. It contains an excellent surface modelling for the colonic anatomy. The colonoscope is represented by a physics-based elastic rod model which discretize into different resolutions that dynamically adapt to the diameter and curvature of the colon. To dramatically reduce computational effort and time, we propose a force correction strategy that adjusts the elastic force between the colonoscope and bowel wall after collision detection. We also allow user-induced perforations so that trainees can manage intra-operative complications in a safe and low-stress environment.
The rest of this paper is presented as follows: details of our simulator are described in “Methods” section, followed by experimental results and discussion of the physical rod model and virtual simulator in “Results” section. Finally, “Discussion” section contains the conclusion and comments on possible improvements in future works on the topic of colonoscopy.
In this section, we firstly outline the structural overview of our proposed colonoscopy simulator. Secondly, we introduce our geometric modelling method for the colon and surrounding structures to give rise to a realistic simulation environment. Thirdly, we describe our colonoscope-colon physic-based system. Finally, we briefly describe the haptic interface and hardware of the simulator.
The bending energy of the rod model is measured by the angle φ i between two adjacent rods t i and ti-1. To simulate intrinsically curved tips with different curvatures, bias angles (e.g. φ1, φ2, φ3, …,) as shown in Fig. 4. In addition, It is need to be predefined a maximum bias angle φ m all joint points in order to let the colonoscope bend gradually.
It can be seen that this method also eliminates artificial perforation of the bowel wall by reducing rod length in small-diameter areas where greater rod lengths could not pass smoothly. However, it is important to note that iatrogenic perforation is a valuable and useful experience for simulation and is preserved in our approach (see Perforation).
Perforation of the bowel wall is a severe consequence of applying excessive force to the colonoscope in colonoscopy and carries various sequelae that can cause serious morbidity and mortality. In order to account for this possibility, we apply a user-defined boundary force F r to define the rupture strength of the bowel wall. When the system total energy’s reaches minimum and collision response is completed, the elastic force applied by the colonoscope to the bowel wall is calculated and compared with F r . If the elastic force is greater than F r ; the bowel wall is perforated.
The hardware component involves a haptic controller as well as a motion tracker. The tracker is responsible for measuring the translation (pushing and pulling) and rotation (twisting) of both catheter and guide wire and thus comprises two modules: rotation detection and translation detection. The rotation module includes encoder raster, encoder disc, and dummy revolving hollow rod which is connected to the encoder disc. When manipulating the hollow rod, the rotation will be detected by the encoder grating, and the signal is simultaneously transmitted to the controller to control instrument model rotating in the colonic model.
To evaluate the simulation accuracy of the proposed physical model of instruments, validation experiments are carried out on a rectal-large bowel phantom which is made of transparent silicon tubing. Therefore, real colonoscopes used in clinical practice can be inserted into the rectal-bowel phantom. It is derived from a real human anatomy and thus provides a realistic environment for evaluating our simulated colonoscopy procedure. The bowel wall of the phantom has the flexibility equivalent to that implemented in our physical model. In our experiments, the phantom is first scanned by computed tomography with a spatial resolution of 0.75 mm and then a semi-automatic segmentation is applied on the 2D scanned images, followed by manual refinement by an expert physician. Finally, 3D surface geometry is obtained by the technique described in earlier papers.
We performed the same operations that navigate the colonoscope to the same position in both the phantom and the simulated model. We performed the same movements several time to guarantee that the simulated displacement of the colonoscope can be reproduced in the physical phantom model experiment. During this process, we took pictures to make comparisons between the simulated model and the phantom model.
Due to the refraction of the transparent silicon material, it is impossible to measure the actual distances of the colonoscope and the bowel walls. However, the relative distance, the shape of the colonoscope and the main collision points in pictures indicate that the behaviors of the virtual instruments in virtual models and the behaviours of the real ones in the real phantom are well matched visually.
In experiment, we adopted different setting of η ranging from 2 to 15 times of the value of elastic coefficient to find a proper feedback coefficient. Then we carried out the 4 simulation positions with different setting of η and got similar results.
The evaluation form 15 experienced third-party gastroenterologists
Bend effect of colonoscope
Haptic experience of manipulation
Translation experience of colonoscope
Rotation experience of colonoscope
Behavior in complex colonic curvatures
Time performance for the system including collision response and frame per second in different number of nodes
Number of nodes
Collision response (ms)
Frame per second
We present in this paper a high fidelity and fast colonoscopy simulator with the aim to train core skills of colonoscopy in a high yield yet safe environment. We also developed a series of strategies for accurate and high resolution anatomical representations, curve skeleton extraction and physics based instrument modelling. Various rendering technique are applied to ensure a real-time visualization by taking full advantage of the GPU computing capability.
The strengths of our system include good quality visual and haptic feedback as well as sophisticated haptic control and maneuvering of the virtual colonoscope inside realistic anatomical models. The trainee can interact with the virtual simulator with a real colonoscope and perform maneuvers such as pulling, pushing, applying torsion and translational motion to shorten the bowel and avoid high curvature formation.
Furthermore, more advanced techniques such as air/water insufflation and polypectomy can also be performed inside the simulation environment. Therefore, the trainee can acquire the necessary skills to identify and, using their clinical judgment, treat polyps found during the procedure.
Time taken for caecal intubation;
Percentage of mucosa visualizer;
‘Pain indicator’ or quantifiable measure of discomfort caused by colonoscope maneveurs;
Adenoma detection rate (ADR) (note: after consulting experienced endoscopists and reviewing publications we could potentially settle on a ‘mock ADR’). Then, using randomly generated cases, we could input adenomas into our generated cases proportionally according to the rate of our ‘mock ADR’.
When a trainee makes an error, the consequences will be reflected in the system. For example, if the bowel wall is perforated, then hemorrhage will occur and be simulated. By this way the trainee can see potential consequences on virtual patient and learn how to avoid and manage intraoperative catastrophic events.
This is a particularly notable advantage that our simulator will have over the traditional ‘see one, do one, teach one’ model. It is extremely difficult if not impossible to ensure that trainees experience a full range of complications because they are not common in real life clinical situations. Therefore, modelling complications such as haemorrhage or perforation could become crucial aspects of our simulator, as the trainee will then be able to learn to manage serious complications without any harm occurring patients. The colon can be modelled using discrete elements to give more flexibility in future studies [24–26].
Furthermore, the current bowel patient database will be updated to include more challenging cases including sharper angulations and thinner colon walls. These bowel characteristics will test the more advanced trainee, ensuring that the simulator does not become boring or stale after multiple use.
Currently, our prototype has been validated in terms of the physical modeling of instruments. However, upon the completion of several additional components that are ongoing development, it will undergo face and content validity from our hospital collaborators and then a comprehensive assessment and validation study including concurrent and predictive validity.
As previously alluded to, the ultimate utility of our simulator is not limited to providing trainees in a highly immersive and realistic environment. We also hope to provide more experienced endoscopists with a predictive tool in which patient-specific data can be inputted in order to plan a high risk or difficult colonoscopy procedure.
JHW conceived and designed the study. TXW and DM preformed the experiment and analyzed the data. JW wrote the manuscript. All authors read and approved the manuscript.
This work was supported in part by National Natural Science Foundation of China (No. 61672510), Shenzhen Science and Technology Program (No. JSGG20150602143414338), and Guangdong Science and Technology Program (No. 2016A020220016).
The authors declare that they have no competing interests.
Availability of data and materials
No public data are used.
Consent for publication
We agree to publish this paper.
Ethics approval and consent to participate
No animals or human are used in this experiment. No ethics approval is required.
National Natural Science Foundation of China (61672510).
Guangdong Science and Technology Program (2016A020220016).
Shenzhen Science and Technology Program (JSGG20150602143414338).
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Torre LA, Bray F, Siegal RL, Ferlay J, Lortet-Tieulent J, Jemal A. Global cancer statistics, 2012. CA Cancer J Clin. 2015;65(2):87–108.View ArticleGoogle Scholar
- Janne PA, Mayer RJ. Chemoprevention of colorectal cancer. NEngl J Med. 2000;342:1960–8.View ArticleGoogle Scholar
- Watanabe T, Itabashi M, Shimada Y, Tanaka S, Ito Y, Ajioka Y, Hamaguchi T, Hyodo I, Igarashi M, Ishida H, Ishihara S, Ishiguro M, Kanemitsu Y, Kokudo N, Muro K, Ochiai A, Oguchi M, Ohkura Y, Saito Y, Sakai Y, Ueno H, Yoshino T, Boku N, Fujimori T, Koinuma N, Morita T, Nishimura G, Sakata Y, Takahashi K, Tsuruta O, Ya maguchi T, Yoshida M, Yamaguchi N, Kotake K, Sugihara K, JSCCR. Japanese Society for cancer of the colon and rectum (JSCCR) guidelines 2014 for treatment of colorectal cancer. Int J Clin Oncol. 2015;20(2):207–39.View ArticleGoogle Scholar
- Lin JS, Piper MA, Perdue LA, ad Elizabeth CMR, Webber M, O’Connor E, Smith N, Whitlock EP. Screening for colorectal cancer: updated evidence report and systematic review for the US preventive services task force. J Am Med Assoc. 2016;315(23):2576–94.View ArticleGoogle Scholar
- Lee S-H, Park Y-K, Lee D-J, Kim K-M. Colonoscopy procedural skills and training for new beginners. World J Gastroenterol. 2014;20(45):16984–95.View ArticleGoogle Scholar
- Patwardhan VR, Feuerstein JD, Sengupta N, Lewandowski JJ, Tsao R, Kothari D, Anastopoulos HT, Doyle RB, Leffler DA, Sheth S. Fellowship colonoscopy training and preparedness for independent gastroenterology practice. J Clin Gastroenterol. 2016;50(1):45–51.View ArticleGoogle Scholar
- Lewis JJ, Rubenstein JH, Singal AG, et al. Complications of colonoscopy. Retour Au Numéro. 2011;74(4):745–52.Google Scholar
- Walsh CM, Sherlock ME, Ling SC, et al. Virtual reality simulation training for health professions trainees in gastrointestinal endoscopy. The cochrane library. New York: Wiley; 2010. p. CD008237.Google Scholar
- Kravochuck S, Gao R, Church J. Differences in colonoscopy technique impact quality. Surg Endosc. 2014;28(5):1588–93.View ArticleGoogle Scholar
- McIntosh KS, Gregor JC, Khanna NV. Computer-based virtual reality colonoscopy simulation improves patient-based colonoscopy performance. Can J Gastroenterol Hepatol. 2014;28(4):203–6.View ArticleGoogle Scholar
- Snyder CW, Vandromme MJ, Tyra SL, et al. Retention of colonoscopy skills after virtual reality simulator training by independent and proctored methods. Am Surg. 2010;76(7):743–6.Google Scholar
- Preisler L, Svendsen MBS, Nerup N, et al. Simulation-based training for colonoscopy: establishing criteria for competency. Medicine. 2015;94(4):e440.View ArticleGoogle Scholar
- Yi SY, Woo HS, Ahn WJ, Kwon JY, Lee DY. New colonoscopy simulator with improved haptic fidelity. Adv Robot. 2006;20(3):349–65.View ArticleGoogle Scholar
- Visser HD, Passenger J, Conlan D, Russ C, Hellier D, Cheng M, Acosta O, Ourselin S, Salvado O. Developing a next generation colonoscopy simulator. Int J Image Graph. 2010;10(2):203–17.MathSciNetView ArticleGoogle Scholar
- Samur E, Flaction L, Bleuler H. Experimental evaluation of a haptic interface for endoscopic simulation. In: World Haptics Conference, vol. 8. IEEE; 2011. p. 545–49.Google Scholar
- Konstantinos T, Lazaros Dimitrios L, Dimitriadis GD. Virtual reality simulators for gastrointestinal endoscopy training. World J Gastrointest Endosc. 2014;6(1):6–12.View ArticleGoogle Scholar
- King N, Kunac A, Johnsen E, Gallina G, Merchant AM. Design and validation of a cost-effective physical endoscopic simulator for fundamentals of endoscopic surgery training. Surg Endosc. 2016;30:4871–9.View ArticleGoogle Scholar
- Wong K, Tu J, Sun Z, et al. Methods in research and development of biomedical devices. Schools Discip. 2014;13(1):196.Google Scholar
- Wu J, Wang H, Zhang P, et al. A preliminary real-time and realistic simulation environment for percutaneous coronary intervention. Biomed Res Int. 2015;2015:183157.Google Scholar
- Wang H, Wu J, Wei M, et al. A robust and fast approach to simulating the behavior of guidewire in vascular interventional radiology. Comput Med Imag Graph. 2015;40:160–9.View ArticleGoogle Scholar
- Alderliesten T, Konings MK, Niessen WJ. Simulation of guide wire propagation for minimally invasive vascular interventions. International conference on medical image computing and computer-assisted intervention. Berlin: Springer; 2002. p. 45–252.MATHGoogle Scholar
- Li S, Qin J, Gao J, et al. A novel FEM-based numerical solver for interactive catheter simulation in virtual catheterization. Int J Biomed Imag. 2011;2011(1687–4188):815246.Google Scholar
- Luboz V, Blazewski R, Gould D, et al. Real-time guidewire simulation in complex vascular models. Vis Comput Int J Comput Graph. 2009;25(9):827–34.Google Scholar
- Gardiner BS, Joldes GR, Wong KKL, et al. Controlling seepage in discrete particle simulations of biological systems. Comput Methods Biomech Biomed Eng. 2016;19(11):1160–70.View ArticleGoogle Scholar
- Wong KKL. Three-dimensional discrete element method for the prediction of protoplasmic seepage through membrane in a biological cell. J Biomech. 2017;65:115–24.View ArticleGoogle Scholar
- Gardiner BS, Wong KK, Joldes GR, et al. Discrete element framework for modelling extracellular matrix, deformable cells and subcellular components. PLoS Comput Biol. 2015;11(10):e1004544.View ArticleGoogle Scholar