SimVis Medical - Virtual Knee Arthroscopy Training System

The Virtual Environment Knee Arthroscopy Training System (VE-KATS) has been developed as a joint project between Orthopaedic Surgeons, Computer Scientists and Psychologists at the University of Hull. The aim is to provide a comprehensive training environment for knee arthroscopy, a form of minimally invasive (keyhole) surgery of the knee joint.

In this procedure, the surgeon operates using a miniature telescope camera and specialised instruments, inserted through small incisions made below the kneecap. This is a difficult task, demanding a great deal of physical dexterity, skill and knowledge on the part of the surgeon.

The simulator incorporates a pair of mock instruments (the arthroscope camera, and surgical probe), which are used in conjunction with a hollow, articulated model of the knee. The position and orientation of the instruments is continuously tracked by the computer, which produces a simulated view from the arthroscope camera.

Mock Instruments

The instruments have been designed to closely resemble those used during normal arthroscopy, and are based on dimensions taken from actual equipment. The magnetic tracking receivers are concealed inside the body of the instrument. The mock arthroscope allows the 'camera' at the rear of the scope to be rotated relative to the body of the scope, to provide the correct view orientation. The cables used by the tracking system have been located at the points where the camera cable, light-cable and tubing normally connect to the arthroscope, so as not to interfere with normal use of the instrument. The direction of view can be placed opposite or adjacent to the light post.

The instruments are tracked with an electromagnetic tracking system (Polhemus 3Space Fastrak), which is also used to measure movement of the knee joint. Three tracking receivers are currently used, while the transmitter is mounted in the upper thigh so that the entire knee can be rotated (eg. movement at the hip) without affecting the co-ordinate origin.

Simulated Arthroscopic View

VE-KATS produces a real-time simulation of the view from the arthroscope (20-30Hz on Pentium II 300MHz, FireGL 1000 Pro Graphics). The offset viewing angle (0, 30, 70 degrees) and field of view are adjustable, and the camera rotation relative to the scope body is included.

An optional anatomical overview (visible in top right of screen) can be used during the early stages of training to show the position of the instruments relative to the knee. It is also possible to view video-clips from real arthroscopies to demonstrate technique, and to illustrate the appearance of actual anatomy.

If you have a suitable MPEG video player, there is a Downloadable Movie you can watch, showing the simulated video from the arthroscope. This doesn't show all the features of the simulator, and we hope to add some more interesting movies over the next few weeks.

Arthroscope Lens Distortion

VE-KATS incorporates a lens distortion model, to reproduce the barrel image distortion caused by arthroscopes. Experiments have been conducted to evaluate the distortion parameters by photographing regular hexagonal grids through the arthroscope and measuring the change in distance between points on the grid. These parameters are then used within the simulator.

Collision Detection

It is important in the course of the operation that the surgeon does not damage the sensitive articular surface of the bones. A novel, real-time voxel-based collision-detection system has been developed, which informs the surgeon when the arthroscope has come into contact with a solid surface such as bone.

Deformable Objects

Aside from the rigid bones of the knee (Femur, Tibia, Patella etc.), there are several structures in the knee that can be deformed, such as ligaments and menisci. The behaviour of these objects is currently implemented using the linear tetrahedral finite element method (FEM). This suffers from a limitation in terms of the complexity of model that can be handled in real-time, since the technique is computationally expensive.

Models have recently been developed using a technique known as Modal Analysis  (again, a finite element method), which allows more complex and detailed models to be used in real-time. This software exists in a stand-alone form at present, and will be integrated into VE-KATS in the future.

Arthroscopic Force Measurement

A standard arthroscopic surgical hook probe (supplied by Smith & Nephew Dyonics) was fitted with a miniature 6-Axis Force / Torque Sensor (ATI Industrial Automation). This was then used to measure operative forces and torques applied by the surgeon, during typical diagnostic arthroscopies. The data acquired will provide valuable information about the range and distribution of forces, to support development of realistic force feedback.

Experiments were performed using a synthetic model of the knee (the Hillway Knee) and using a cadaver pig knee. In the future, pending safety and ethical approval, we intend to repeat the experiments in the operating theatre.

Force Feedback Device

A prototype force feedback device has been developed to investigate the feasibility of including physical sensations alongside the simulated video. This is necessary to prevent penetration of instruments into bone, and to simulate specific tasks such as applying traction to the anterior cruciate ligament. 

The device measures the position of the mock surgical probe with optical encoders (angular resolution 0.036 degrees, linear 0.031mm). Collision detection is performed between the probe and the knee anatomical model, and response forces calculated to prevent penetration of the bones. These forces are then applied to the instrument by DC electric motors (480mNm continuous torque, 3.2N continuous force). The device is limited to three degrees of freedom at present, two rotations in the plane of entry, and one translation along the long axis of the instrument.

Scoring System

An objective scoring system is being developed to assess the performance of trainees during arthroscopy. The trainees are asked to perform specific tasks, such as examining or probing certain anatomical features. The simulator then assesses how well the trainee performed and provides numerical scores.

The system is based upon an objective scoring method for pre-recorded video of conventional arthroscopic training. The work was carried out in conjunction with the Department of Psychology at the University of Hull.