Government & Politics

A vrml simulator for ventricular catheterisation

Description
A vrml simulator for ventricular catheterisation
Published
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
  Eurographics UK, Cambridge, April 1999 A VRML Simulator for Ventricular Catheterisation Nigel W. John, Manchester Visualization Centre, University of ManchesterNicholas Phillips, Department of Neurosurgery, Leeds General InfirmaryReaz Vawda, Institute of Neurology, University College LondonJames Perrin, Manchester Visualization Centre, University of ManchesterAbstract One skill that trainees in neurosurgery need to gain early in their training is anappreciation of the ventricular system in the brain, and how to cannulate it inan emergency. The flow of cerebrospinal fluid can be obstructed in theventricles by many pathological processes leading to a dangerous conditionknown as hydrocephalus. The pressure within the ventricles can rise leading toloss of consciousness. The ventricular system can be cannulated in theoperating theatre, fluid drained and the potentially lethal rise in pressurerelieved. This procedure is a commonly performed operation in neurosurgicaldepartments.We decided to model the ventricular catheterisation procedure using VRML(Virtual Reality Modelling Language). The intention is to provide a readilyavailable tool that can be used by neurosurgery trainees to practice the procedure before having to perform it in the operating theatre. Such a trainingtool does not currently exist.This paper describes the use and implementation of the world’s first VRMLsimulator for ventricular catheterisation. This work will be used as the basisfor the implementation of many more training tools for surgical procedures. Keywords VRML, Simulators, Neurosurgery, Training, Ventricles, Catheterisation. 1 Introduction Surgical training is largely a matter of closesupervision on the apprenticeship model. There is agrowing requirement in training to practisetechniques and operations in a way that does not put patients at any risk and one way this can bedone is using virtual reality modelling of the procedure. Neurosurgery is especially suitable for this approach as it is a high risk speciality withcomplex procedures which rely heavily on three-dimensional radiological images.Our aim is to provide cost-effective simulators for neurosurgical (and general surgical) procedures,that can be run on any workstation platformincluding a standard PC of average capacity. TheWorld Wide Web provides the obviousinfrastructure to achieve this, and we areconcentrating on the use of VRML 97, and JAVA.The use of these technologies for medicalvisualisation has already been demonstrated [1].Our first surgical simulator – for ventricular catheterisation – has been implemented entirely inVRML 97.  Eurographics UK, Cambridge, April 1999This paper provides an overview of the ventricular catheterisation procedure and the use and implementation of a VRML-based simulator. Wediscuss the effectiveness and usefulness of thesimulator and indicate future work that will becarried out in this area. 2 Ventricular Catheterisation Within the brain, the flow of cerebrospinal fluid can be obstructed in the ventricles by many pathological processes leading to a dangerouscondition known as hydrocephalus. The pressurewithin the ventricles can rise leading to loss of consciousness. The ventricular system can becannulated in the operating theatre, fluid drained and the potentially lethal rise in pressure relieved.A cannula of diameter 2mm is inserted into theventricles after a burr hole is made in the skull. Thecannula is a hollow silastic tube that is inserted afew centimetres into the ventricle. It is thenconnected to a longer tube of similar diameter,which drains the fluid into a reservoir bag thatholds up to 500 ml. Raising or lowering thereservoir, which is a closed system, controls the pressure within the ventricles. The reservoir isusually set 10cm above the head thus the pressurein the ventricle must be greater than 10cm of cerebrospinal fluid to drain. The trajectory mostlyemployed is through the right (non-dominant)frontal lobe, aiming at the anterior horn of theventricles.The procedure is a commonly performed operationin neurosurgical departments and a skill thatneurosurgery trainees need to quickly acquire. Oneof the main problems for the trainee is their lack of three-dimensional awareness of where theventricles lie within the head and brain. It isimportant however that the cannula hits its targetwithin the first few attempts. The more passes of the cannula through the brain the more potential for significant brain damage.Being able to practice this procedure on a suitablesimulator is therefore going to be highly desirable. 3 The Simulator in Action Figure 1 shows the initial layout of the simulator.The head and cannula models are visible. Inaddition to the VRML browser’s control panel, twoslider widgets and a set of three radio buttons have been provided for user interaction. Figure 1 The Ventricular CatheterisationSimulator The simulator closely mimics the actions of aneurosurgeon performing a real ventricular catheterisation. The user (a neurosurgeon trainee)will first use the standard VRML browser interfacecontrols to position the head in the “On Table” position – see Figure 2. This position has also beenincluded as a viewpoint. Figure 2 The On Table Position The cannula must then be positioned at the desired entry point on the head surface: The red radio button makes the head “active”. Amouse click on the head will then translate thecannula so that its tip is at that point. If the mousebutton is kept depressed, then the cannula will follow the cursor around the head surface. This isthe current state in Figure 2.  Next the cannula is put into the correct orientationfor insertion into the head: The amber radio button stops the cannula frombeing translated, and activates a rotation mode.  Eurographics UK, Cambridge, April 1999 With the cannula selected, cursor movements aremapped to rotations around the tip of the cannula. The ventriculostomy can now be performed: The green radio button switches the cannula into atranslation mode once again. This time, however,when the cannula is selected cursor movements aremapped to translating the cannula along its current trajectory i.e. the direction in which it is pointing.Figure 3 depicts the result of such an action. Figure 3 After Insertion of the Cannula The two slider controls can be used to vary thetransparency setting of the head and skull modelsrespectively. It is useful to be able to see theventricles when the user first begins to practice this procedure and he can set the transparency controlsaccordingly. For example, in Figure 3, the skull iscompletely transparent. As the three-dimensionalawareness of the trainee improves, however, hewill use the simulator with both of these modelsopaque. The head and skull can then be madetransparent at the end so that the user can see if hesuccessfully punctured the ventricles. 3 Implementation The simulator has been implemented completely inVRML 97 using javascript as and whenappropriate. 3.1 Generating the Models A variety of methods were employed to create themodels for the simulator. 3.1.1 The CannulaFigure 4 The Cannula Model This is just a very simple object comprising of acylinder, with a cone used at its tip – Figure 4. Areal cannula does not have such a tip, but it wasfound to be a useful in the simulator to be able toidentify each end of the cannula. The centre of transformation for the composite object is set to beat the end of the tip. 3.1.2 The Head and Skull To obtain the head model we used a laser scanner – the Rapid 3D Digitizer Model 15, fromCyberWare[2]. The scan is of the head one of theauthors of this paper!The laser scan is rapid and easy to perform, and theCyberWare software supports VRML output. Itdoes, however, produce extremely large three-dimensional data sets – 8 Mbytes for the head.Polygon reduction techniques were thereforeapplied to the data set to minimise the polygoncount and produce a far more manageable data setsize of 440 Kbytes. This was essential to improvedownload and response times over the World WideWeb. Figure 5a: TheOriginal Head Model The method used was an implementationdeveloped at Manchester Visualization Centre based on Klein et al's technique [3]. This is acomplex but highly accurate polygon reductionalgorithm. The method iteratively removes verticesfrom the mesh that incur the smallest error. When avertex is removed the resulting hole isretriangulated. The quality of the reduction iscontrolled by the error measurement, in this casethe one-sided Hausdorff distance between thesimplified and srcinal surface is calculated for each vertex removed allowing the maximum error to be known at all times. This high degree of accuracy makes it a good choice for medicalapplications.  Eurographics UK, Cambridge, April 1999 Figure 5b: 90% Polygon Reduction The srcinal surface consisted of approximately200,000 triangles (Figure 5a), and reductions of 80% and 90% (Figure 5b) were made. The latter simplification with 20,000 triangles was found thegive the best balance of performance and quality.Both VRML and the polygon reduction method usea vertex indexing connectivity list and soconverting the data to and from VRML wasstraightforward.We also took a laser scan of a human skull borrowed from the University of Manchester Medical School. In the end, however, we used asimpler VRML model of a skull that was found inthe public domain. Figure 6: Registered Head and Skull The CosmoWorlds VRML authoring environmentwas used by a neuroscientist to manually co-register the skull and head models. See Figure 6. 3.1.3 The Ventricles The ventricle model was extracted from a magneticresonance imaging (MRI) scan. A Siemens scanner was used to produce an “mprage” data set throughthe patient’s head of 128 slices, each slice being256 pixels square. An mprage scan is optimised tohighlight the cerebrospinal fluid (CSF) in theventricles.The next step was to segment the boundary of theventricles on each slice, producing a stack of contours that defines its volume. The IDL image processing software [4] was utilised to do this byapplying a simple thresholding technique – quiteeffective on a MRI scan where the CSF has beenhighlighted. Manual editing of each contour wascarried out where necessary. The contour stack canthen be polygonized to create a geometric modeleasily converted into VRML. IDL also providessupport for this type of operation. The result is ananatomically accurate model of the ventricles – seeFigure 7. Again CosmoWorlds was used to scalethe ventricles model to the correct size, and  position it at its correct location and orientation inthe head. Figure 7: The Ventricles Model 3.2 The User Interface Controls In addition to the browser user interface controls,two types of VRML widgets have been used – sliders and radio buttons. These were obtained from the VRML clip art supplied withCosmoWorlds. The widgets have been placed within a Head Up Display so that they follow theviewer as it moves around the scene [5].The sliders are used to change the transparencyvalue of the materials used for the head and skullmodels. This is just a simple matter of a  ROUTE  from the eventOut   result of the slider control scriptto the set_transparency  field of the object’smaterial node. 3.3 Moving the Cannula The manipulation of the cannula model by the user is the most important part of the simulator. It must be straightforward, efficient, and effective inallowing a 2D pointing device (the mouse) to movethe cannula in 3D space.  Eurographics UK, Cambridge, April 1999As described in Section 2, the cannulamanipulation has been implemented in three phases, controlled by radio buttons. 3.3.1 Moving the Cannula to the Entry Point onthe Head A TouchSensor   is grouped as a sibling to the head model, and the red radio button makes this sensor active. The hitPoint_changed eventOut   of the TouchSensor   is sent to a script that generates a translation_changed eventIn  for the cannula. The translation_changed   value is continually updated while the mouse button remains pressed and thecursor is moved around the head surface. Thecannula is thus translated so that is tip is in contactwith the head’s surface at the point selected by theuser. 3.3.2 Orientation of the Cannula Selecting the amber radio button disables thehead’s TouchSensor  , and enables a SphereSensor  that has been grouped as a sibling of the cannula.Remember that when the cannula model wasgenerated the centre of its Transform groupingnode was set to be at the end of the cannula’s tip. A  ROUTE   from the SphereSensor’s   rotation_changed eventOut   to the cannula’s  set_rotation eventIn ,results in the cannula being rotated about its tip.The user can therefore easily get the cannula intothe desired orientation. 3.3.3 Insertion of the Cannula Selecting the green radio button disables thecannula’s SphereSensor  , and enables a PlaneSensor   that has also been grouped as a siblingof the cannula. This sensor has been set up so that translation_changed   events are clamped to the x-direction. The value of this eventOut   is used todetermine whether the cannula is inserted into (xvalue from the sensor is increasing) or withdrawnfrom the head (x value from the sensor isdecreasing).The cannula is moved in fixed steps of 0.1 units.This allows for accurate positioning of the cannula – a real cannula is graduated in centimetre units. Toimplement the motion, the set_translation  fields of the cannula’s constituent components – thecylinder and the cone – are updated in their localcoordinate system. This gives the effect of thecomposite cannula object being moved along itscurrent trajectory, which is what is required here. 3.3.4 Further Interaction The user is allowed to reposition the cannula on thehead’s surface and then repeat the procedure. Whenthis occurs the cannula’s geometry components arereset to their srcinal value in their local coordinatespace. If this is not done, subsequenttransformation operations on the cannula will notwork correctly.The user in not allowed to rotate the cannula onceit has been inserted into the head. 4 Results and Feedback The simulator has been accessible on the World Wide Web since September 1998 at two sites:http://www.man.ac.uk/MVC/research/visual/medical/VirtualSurgeryhttp://neuronott.nottingham.ac.uk/WEBpages/VirtualSurgery/It has been tested on PC, Macintosh, and SGIworkstations using the CosmoPlayer VRML browser.Feedback received to date has been very promisingand has suggested several improvements that arecurrently being worked on. For example: •   When the cannula successfully enters theventricles, the neurosurgeon will experience a“popping” sensation. An appropriate audioevent could be triggered in the VRMLsimulator to provide this feedback. •   In a real catheterisation procedure movementof the cannula in any direction other than alongits long axis while it is in the brain cannot beallowed. In the simulator, however, it has beensuggested that this would be helpful for educational purposes. Appropriate warningsshould be given to the user to inform him thathe should not do this in practice.Clinicians from across Europe and the USA are providing feedback and comments. We will bereporting on these in detail in a future paper. 5 Conclusions The technique of introducing a cannula through the body tissues to reach a target and then performingsome action at that target is a common concept inmedicine. Our modelling of ventricular catheterisation can easily be extrapolated intotraining a variety of procedures.
Search
Tags
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks