In contrast to the rapid advances and breakthroughs in endoscopy, ERCP training has not been changed much. Animal models of mainly pigs and dogs were introduced since the beginning [17, 18]. They pose technical difficulties due to the anatomical differences, high cost, and lack of reuse. For overcoming these problems related to animal models, ex-vivo models using some organs from pigs, and hybrid ex-vivo models combining organs from pigs and chickens, have been introduced [19–22]. Although they have been widely used in ERCP training workshops for providing detailed and specific training programs, they have not completely overcome the challenge of anatomical differences, and limitations in their use, especially related to storage and durability. Computer simulators have some advantages as they pose no ethical problems, no storage limitations, and are durable [23–26]. However, they have the disadvantages of not being realistic due to the lack of haptic feedback and high initial-purchase cost. Also, compared with the mechanical method mentioned below, computer simulators do not use real endoscopes or accessories, and have not received much response so far [27]. The mechanical methods involve artificial duodenum, ampulla, and distal bile ducts in a container, so that only some techniques such as cannulation and stent insertion can be practiced [7, 28]. It is simple and portable, and it has been widely used for basic training. However, the training using this model lacks realism, and cannot provide more sophisticated training.

We were trying to find a solution, which would provide a realistic experience during practicing ERCP, before we made our ERCP training silicon model. Although fluoroscopic guidance carries a burden of radiation exposure and space constraints, it is more important to directly feel the ERCP procedure. We also thought that it would be more effective to exercise the procedures while observing the endoscopic and the fluoroscopic images simultaneously. Also, the model would be designed to make repetitive and semi-permanent reuse possible, and not be limited to simple basic ERCP techniques, but to be utilized for all advanced ERCP techniques as much as possible.

In order to make the experience of being placed around the duodenal ampulla and the feeling of using various ERCP accessories as realistic as possible, 3D modeling data were obtained from human CT images, as per previous reported studies [13–15]. One of important steps in this process was to make the stomach and duodenum. We wanted to make a model where the duodenoscope could be inserted in a sequence that followed from esophagus, stomach, and duodenum, but since silicone material was not elastic and stretchable like the human stomach, we decided to cut the stomach for easy insertion of duodenoscope around the duodenal ampulla and omitted the shortening step. Therefore, insertion of the duodenoscope was not difficult, but the direction of the fluoroscopic image had to be inevitably adjusted because of the difference in direction as compared to an actual human (Fig. 5). Another important step in the designing process was to make the ampulla and CBD in various shapes. If this part could be made into various shapes, it could be implemented for various training models and disease models. In order to resolve this issue, we implemented a new 3D printing method to make these parts in various forms, and to connect them to the main body in a module-type assembly format. Another important step was to implement the sphincterotomy model. For the new model to be appealing and used widely in future, it was necessary that it should be more convenient, cost-effective, and reusable than the previous models. As mentioned for other modules, it had to be easily assembled with the main body, electrically operated, easy to store, and relatively inexpensive. Because the present silicone-material phantom utilizes injection molding technique, it can secure many economic benefits in re-production. Since re-production requires only silicone materials and injection molding efforts, we can save the cost of designing for positive phantom model and negative molder model, and the cost of 3D printing of negative molder as well. Therefore, it is possible to reproduce the same phantom at a price of one-third. Because the present phantom has modular configuration, in addition, it can be implemented with minimal efforts of designing and 3D printing for additional variant designs. While looking for suitable material, the curved Vienna sausage met all the conditions for eligibility and could be used in variable sizes and shapes. Also, it is very similar to the practice in the human body, and creating the sphincterotomy model was a very positive and reaffirming experience.

There were some limitations to this study. First, we did not create a model that had all portions of the esophagus, stomach, and duodenum in the same order as humans. We would like to try this procedure again with using silicon with greater flexibility, which is not available currently. Second, transparent silicone is considered to have the advantage of maximizing portability because it does not need fluoroscopic guidance. However, at present, transparent silicone cannot be used for an ERCP training model because the more silicone is transparent, the more it is harder [29]. We hope highly transparent silicon that overcomes these problems would be developed in future. Third, due to the nature of silicone material, the surface tension is higher than that of the bio-tissue. But, if a softener such as glycerin is injected into the tract, this problem can be solved to some extent. Fourth, if part of the intrahepatic bile duct can be made in various shapes, it can be utilized for various disease models. Fifth, if there is a tube-shaped Vienna sausage, we will have more realistic experience with endoscopic sphincterotomy. It can demonstrate procedure-related perforation and allow us to set a safety margin for endoscopic sphincterotomy. Despite these limitations, our silicon model would be a more attractive and professional training model once some 3D technical limitations are resolved. We are certain that ERCP training will be easy and effective if better silicon models are created by overcoming these problems.