SURGICAL TRAINING AND PERFORMANCE EVALUATION USING VIRTUAL REALITY BASED SIMULATOR

Sangeeta Shrivastava1, R Sudarshan2, M Ramesh3, C S Rajan4
CRJ Prakash Naidu5, N Sitaram6

(1),(2),(5),(6) : Institute for Robotics and Intelligent Systems/ Centre for Artificial Intelligence and Robotics
Rajbhavan Circle, High Grounds, Bangalore, India – 560 001
E-mail: naidu@cair.res.in

(3),(4) : Bangalore Endoscopic Surgery Training Institute and Research Centre, A V Hospital
Patalamma Temple Street, Basavanagudi, Bangalore, India – 560 004
E-mail: drramesh@hotmail.com

Abstract: Evaluating the performance of a surgeon under training is traditionally at the discretion of a senior surgeon. The methods and tools used to evaluate are subjective and could be influenced by either the trainee or the trainer. Laparoscopy is a Minimally Invasive Surgery (MIS) of the abdomen, which is performed with instruments and viewing equipment inserted through small incisions (about 10mm). Using virtual reality technology, a method has been developed to evaluate the trainee using the laparoscopic surgery-training simulator. The individual user can practice laparoscopic surgical procedures by performing the representative operative steps/actions on virtual objects in a virtual world for the purpose of learning and constantly improving the skill of hand-eye coordination and manual dexterity. The exercises and the performance evaluation tools are designed for accurate movement of the tool to the target objects and its interactions with the other objects within the vicinity of the tool path. The simulator used for performance evaluation elucidated in this paper is developed to work on a PC based system. The simulator is designed and developed in active collaboration with laparoscopic surgeons. The paper elucidates the novel features of the simulator, examines the parameters needed to be assessed, their importance in the evaluation and also the results of the preliminary clinical trials which were carried out by the users.

Key words: Laparoscopic surgery, hand-eye coordination, cholecsystectomy, pelvi trainer.

1- Introduction
Over the last decade or so the growing use of laparoscopy has radically reduced the pain and cost of some types of surgery and reduced recovery time for patients. The technique requires tremendous dexterity and hand-eye coordination, hence simulators are valuable training and assessing tools.


Fig 1: BEST-IRIS Laparoscopic Surgery Training Simulator

The traditional training method employed by the Laparoscopic surgeons to train their juniors is by using the actual Laparoscopic equipment on a device called pelvi trainer. The laparoscopic equipment includes the laparoscope (camera), fiber optic light source, surgery tools, TV set, etc. In this trainer the trainees place objects (such as grapes, tomatoes, balloons, etc.) inside a cavity and practice on them to gain the skills required for the actual surgery [1, 2]. Typical exercise would be to peel the skin of a grape or put a rubber band over a set of nails. After training on the pelvi trainer the trainees are asked to practice in animal lab [3] before operating on human beings. The main advantages of this training is that the trainee gets the feel of basic operation steps, natural force feedback from the operated objects, and also the pivot of the surgical tool simulates the position of pivot of the trocar as in actual operation. The disadvantage is that unless the trainer is observing the performance of the trainee, it is difficult for him/her to assess the trainee’s psychomotor skills. This can be overcome by a Virtual Reality based simulator as it is possible to automate the evaluation process. Also, a Virtual Reality based simulator detects movements less than 1 mm which is difficult by human eye.

Laparoscopic simulators for training presently available are MIST VR, VSOne, etc. MIST VR is a commercial product manufactured by an UK based company called Virtual Presence [4]. There are six basic tasks designed for improving the hand eye coordination of the trainee surgeon. Karlsruhe has a similar project going on based on their VEST System One (VSOne) Technology using kismet simulation software [5]. This simulator is developed to be a dedicated system. Most of the simulators rely on specialized hardware. However, MIST VR is a PC based solution. In this paper, the development of an improved Laparoscopic Surgery Training Simulator called BEST-IRIS is presented. It is a PC based system developed jointly by Bangalore Endoscopic Surgery Training and Research Institute (BEST) and Institute for Robotics and Intelligent Systems (IRIS). The simulator is illustrated in Figure 1.

The BEST-IRIS simulator integrates two standard laparoscopic instruments retrofitted to a pair of 4 degrees of freedom electro-mechanical devices, interface electronics, Virtual Reality Software and a PC. The PC monitor displays the movement of the surgical instruments in real-time 3D graphics. The virtual environment displayed on the computer screen is an abstracted representation of the abdominal cavity in the form of simple geometric shapes. Trainees are guided through a set of seven exercises of progressive complexity that imitate techniques employed in surgeries such as laparoscopic cholecsystectomy (removal of gall bladder stones), ovarian cyst puncturing, etc. Each exercise primarily focuses on training a particular skill and also uses the acquired skills from the previous exercises. Different levels of difficulty have been provided in each exercise for the trainee to sharpen his/her skills. In this paper, several aspects of the simulator and its trials are presented. The block diagram for the entire virtual reality simulator is explained in section 2. Section 3 summarizes the seven training exercises designed for training the surgeons. Section 4 depicts the overall configuration of software design and implementation in the form of block diagrams. Section 5 briefly discusses the salient features of the software. Section 6 specifies the parameters considered for assessment of the trainees. Section 7 discusses the clinical trials and its results. The conclusion and plans for future work are summarized in Section 8.

2- Virtual Reality System Integration approach
The BEST-IRIS VR simulator consists of an indigenously developed hardware to approximate a laparoscopic surgery set up prepared for the procedures on an operating table. This hardware is interfaced to the software program running on a Windows based personal computer. Figure 2 shows the block diagram of the total system.


Fig 2: VR System Block Diagram

The hardware is divided into mechanical and electronics modules. Mechanical module is made up of two surgical tools placed in two independent tracking devices, each having four degrees of freedom (DOFs). Initially potentiometers and micro-controllers were used in the electronics module, but the results were unsatisfactory due to low resolution and lack of durability of the potentiometers. The speed and the number of sensors the micro-controller could handle were also limited. Hence this module was reengineered entirely to replace the potentiometers and the micro-controllers with high resolution Optical encoders and FPGAs to get best case resolution of up to 11 microns. A device driver is necessary for the hardware to talk to the software program. It has been developed for Windows 9X/NT/2000. The software module provides the User-Interface and the visual images for the simulator. Its design is described in section 4 of this paper.

3- Training Exercises
The exercises were designed based on the different actions performed during live surgeries. All the different kinds of surgeries performed by Gastro Intestinal surgeons, gynecologists, urologists, pediatric and endocrine surgeons were recorded and the various actions performed during the surgeries were analyzed. The basic actions performed were noted and exercises were designed to mimic these basic actions. Thus, the common basic actions involved in most laparoscopic surgeries are covered in the set of exercises included in the simulator.

The first exercise aims to manipulate the tool in 2D to reach the target square with economy of movement. The trainee gains the skill in hand-eye coordination for movement of the tool in 2D. The second exercise aims to manipulate both the tools in 3D. The trainee learns to coordinate the movement of both the tools to acquire the target by manipulating the right hand tool end effector and place the target object inside a cavity in the left hand tool end effector. The third exercise aims to mimic actions in diathermy. It involves hand-eye-foot coordination, as the surgeon uses a foot pedal to do the action of diathermy. The fourth exercise mimics the action in dissection. The trainee is taught to grasp the vein obstructing the duct using the left tool, move it slightly to see the duct, place the right tool end effector (scissors) at the target and dissect it. The fifth exercise imitates the basic tasks conducted in ovarian drilling for puncturing of cysts. The trainee practices to grasp the organ/object using the left tool, rotate it to reveal the target cyst, place the right tool end effector (piercing tool) at the target and puncture it. The sixth exercise focuses on the act of peeling, which is similar to the peeling of certain unwanted layers deposited on the liver, etc. The seventh exercise is based on one of the most difficult to perform tasks – suturing. The trainee practices the task of winding the thread on a cylindrical target and then ties the surgical knot.

4- Software Design
 

Fig.3: Basic Block Diagram of Software Structure.

Figure 3 shows the basic block diagram of the software. The simulator device drivers and Open Data Base Connectivity (ODBC) drivers are implemented for interfacing the application with the simulator and the database. The design and interconnection of modules is as shown in
Figure 4.


Fig. 4: Software Design.

The data from the sensors in the simulator is polled during the idle cycles of the application and the corresponding attributes are updated. The database is used to store the user information along with the data pertaining to the exercise conducted by the users and the optimal set of values for each exercise.

5- Salient Features of Simulator Software

5.1 - Camera Simulation
The simulator is provided with a virtual camera controlled by the keyboard. Hence the trainee is able to move the camera also during the training. The virtual camera has all the degrees of freedom similar to that of the actual laparoscope, including zoom-in & zoom-out. This is designed to imitate the surgeon’s visibility during the surgery.

5.2 - Database using Open Data Base Connectivity (ODBC)
ODBC is the database section of the Microsoft Windows Open Services Architecture (WOSA), an interface which allows Windows-based desktop applications to connect to multiple computing environments without rewriting the application for each platform. The simulator data base is developed such that it stores all the personal data of the users, their performance in each exercise and optimum values of all the exercises in a Microsoft Access database running on ODBC driver. The performance data includes various metrics measured during the exercises along with date, time and scores.

5.3 - Scalability of software
The software is designed keeping in view futuristic addition of training exercises. As and when the need for training in new procedures is required, it can be independently developed and integrated into the existing software.

5.4 - Simulator User Interface:
The software for simulator was developed as a Single Document Interface (SDI) Application using Microsoft Access database. In addition to the standard Main Frame Window features, a Dialog Bar is added for easy accessibility of user commands. Figure 5 shows the actual Application window.


Fig.5: The Simulator User Interface.

The user interface classes also have methods, which query the database for offline viewing of the trainee performances. It also provides instructions to the user regarding the system and procedural usage.

6 – Parameters for Assessment of the Trainee
The main advantage of VR based training is the automated evaluation of the individual being trained. In this simulator, the trainee is evaluated based on three parameters viz. economy of distance, economy of time and the negative marks to penalize an incorrect action during the training of a particular exercise. These three parameters carry different weightage based on the importance of concerned parameter.

Economy of time is categorized as the least important parameter, because time taken by the surgeon to perform an operation is relatively less critical. But large time delay is unwarranted. Considering both these aspects, the weight assigned is 0.1 or 10%.


Fig. 6: Distance and Time graphs of previous five attempts.


Fig. 7: Tool Path traversed by the trainee which is recorded and displayed after each exercise.

Economy of distance carries a weight 0.2 or 20%. Even though this parameter is not very critical, it has significance because of the biological work environment. The workspace where the operation takes place is inside the body (abdomen) surrounded by delicate organs. If the surgeon moves the tools inside the body for unnecessarily long lengths, there is a greater probability of him/her damaging the healthy tissues in the process.

Provision is made in the software to graphically illustrate the distance and time taken by the trainee during the last five attempts as shown in Figure 6. The weighted score considering both the aspects of distance and duration is automatically computed and is displayed at the graphs. Further, the tool path traversed can be recorded and displayed as shown in Figure 7. These features enable objective post session analysis of trainee performance.

Penalties carry the maximum weight of 0.7 or 70%. The trainee is penalized when there is some wrong action during the training. These mistakes are classified into different categories, again depending upon their relevance to the training. The tolerance of the errors is progressively reduced depending upon the level of training in each exercise; the higher the level, the lower the tolerance.

7- Clinical Trials and Results
An analysis of laparoscopic surgical training on the BEST-IRIS Laparoscopic Surgery Training Simulator was conducted on a study group at Bangalore Endoscopic Surgery Training Institute and Research Centre. The study group consisted of first, second and third year surgical residents. They were made to work on Level I of all the exercises daily for a period of one week by the end of which they had finished practicing 20 times on each exercise. Their scores based on the different parameters being tested were recorded in a database. A progress chart of each individual trainee was maintained to acquire the trends in their performance. A positive trend (scores increasing) ascertained the improvement in skills that can be attributed to the simulator. At the end of the clinical trials, the analysis of their performance was made both on the virtual reality simulator and also on a standard paper cutting exercise in a pelvi trainer. There was a significant improvement in both the paper cutting test and the virtual reality exercise performances in majority of the individuals. The results obtained on the virtual reality simulator were also analyzed to study the learning curve of the subjects. The learning curve for acclimatization to the system was found to be very short (a mean of 2.9 attempts/exercise) which indicates that the system is very user friendly. The learning curve for performing each exercise varied from individual to individual. While analyzing the different metrics individually, it was observed that the study group reduced negative scores (penalties) early in the course of their exercises (a mean of 4.1 attempts/exercise), which meant that the virtual reality exercises helped them to perfect their purposeful movements. There was a large variation in the distance covered in many individuals, though some of them showed a consistent improvement. There was a more consistent improvement noticed with the time taken to perform the exercises. Fifteen percentage of trainees obtained good scores right from the beginning in all the exercises and no improvement was found in their scores at the end of the training session. These residents were experienced in assisting endoscopic surgeries prior to this study. Ten percentage of them did not show a significant improvement in their scores.

Figures 8 and 9 show the graphs of thirteen attempts of two different trainees for exercise 1. The several attempts are represented on the X-axis tagged with the attempt identification number noted by the simulator (“Attempt ID”) and the height of vertical column of graph at each attempt indicates the corresponding Negative Score on a scale of 0 to 10 along the Y-axis. The instructors at BEST evaluated trainees as follows:

a) Trainee X (Figure 8): above average student
b) Trainee Y (Figure 9): average student


Fig. 8: Score Graph of an above Average Trainee.

Fig. 9: Score Graph of an Average Trainee.

The above result can be ascertained from the graphs. Trainee ‘X’ shows a progressive decline in penalties, which has reduced to zero after a few initial attempts (6 attempts). Trainee ‘Y’ has an overall negative trend for penalties but shows certain amount of fluctuation. The fluctuation in the penalties was noticed in a large percentage of trainees. A small percentage (15%) of the trainees showed above average trend. A subjective analysis of the entire study group showed that they gained a lot of confidence while practicing on the virtual reality trainer and also felt that it helped them to fine-tune their skills.

8 – Conclusions and Future Work
The simulator presented in this paper reports about a system to train in laparoscopic surgery with features to automate the evaluation process. The clinical trials have demonstrated logical trends in improvement of the trainee skills by use of the simulator. It can be further enhanced to make it a more immersive and realistic system. Haptics, the ability to simulate touch plays a key role in making VR based simulators more realistic. It is desirable to model the anatomy of the abdomen so that the VR environment can be made as natural as possible [6]. Producing realistic force feedback is a research challenge [7] and work in this direction has been initiated. The simulator can also be developed as a pre-operative planning tool that can allow the laparoscopic surgeon to perceive, practice reasoning and manipulate 3-D representations of the anatomy.

9- Bibliography
[1] http://www.lapsurgery.net.
[2] Graham Upton, “Laparoscopic Surgery Simulation
Realism in a PC”, http://www.dvc400.com/papers.
[3] Michael downes, M.Cenk Cavusoglu, Walter Gantert, Lawrence W. Way, Frank Tendick, “Virtual Environment for Training Critical skills in Laparoscopic Surgery”, MMVR ’98 pp 316-322, JD westwood etal,.eds. IOS Press.
[4] http://www. hmc.psu.edu/simulation/Equipment/
MIST_VR_Trainer.
[5] U. Kühnapfel, Ch. Kuhn, M. Hübner, H. G. Krumm, H. Maab, B. Neisius, “The Karlsruhe Endoscopic Surgery Trainer as an example for Virtual Reality in Medical Education.” Minimally invasive Therapy and Allied Technologies (MIITAT). Blackwell science Ltd. 6(1997) pages 122-125. http://www.www-kismet.iai.fzk.de.
[6] J. Platt, D. Terzopoulos, K.Fleischer and A. Barr, “Flostically Deformable model”, In Siggraph Proceedings (July 1978), ACM, 205-214.
[7] Basdogan, C., C.-H. Ho and M. A. Srinivasan, “Virtual environments in medical training: Graphical and haptic simulation of laparoscopic common bile duct exploration”, IEEE/ASME Transactions on Mechatronics (2001), vol (6): pp 269-285.