Simulating tumour removal in neurosurgery

https://doi.org/10.1016/S1386-5056(01)00197-6Get rights and content

Abstract

In this article the software system ROBO-SIM is described. ROBO-SIM is a planning and simulation tool for minimally invasive neurosurgery. Different to the most other simulation tools, ROBO-SIM is able to use actual patient's datasets for simulation. Same as in real neurosurgery a planning step, which provides more functionality as up-to-date planning systems on the market, is performed before undergoing the simulated operation. The planning steps include the definition of the trepanation point for entry into the skull and the target point within the depth of the brain, checking the surgical track and doing virtual trepanations (virtual craniotomy). For use with an intra-operative active manipulator, which is guided by the surgeon during real surgery (robotic surgery), go- and non-go-areas can be defined. During operation, the robot restricts the surgeon from leaving these go-areas. After planning, an additional simulation system, which is understood as an extension to the planning step, is used to simulate whole surgical interventions directly on the patient's anatomy basing on the planning data and by using the same instruments as for the real intervention. First tests with ROBO-SIM are performed on a phantom developed for this purpose and on actual patient's datasets with ventricular tumours.

Introduction

It is taken for granted, that a pilot has to pass many flights successfully in a simulator before flying a real airliner. This is necessary because of the increasing complexity of airliners and the new risks resulting from stressing the pilots with the supervision of numerous flight control instruments. However, in the field of surgery, training is still performed in more traditional ways. One way for surgical training is the use of actual patients, which increases the risk of complications unnecessarily.

Surgical training can also be performed with animals, such as pigs or rabbits. However, for small and middle-sized medical centres, it is often impossible to raise funds for animal training, due to the costs of general anaesthesia, the maintenance of surgical units and instruments and the animals themselves [1]. Moreover, in many countries, the use of animals for experimental surgery is controversial or even prohibited.

In some medical centres, surgical training is performed partly on cadavers, but in addition to high costs and poor availability, surgery on post-mortem tissue is not very realistic [2].

Because of these problems with traditional training methods, high efforts have been taken in the last years to develop surgical simulators for computer-assisted training. However, since very high computational demands are required to simulate the physiology, most of these simulators are limited to the use of simplified models of the human's anatomy [3]. Thus, these simulators are inappropriate for the training of surgeons. Especially in neurosurgery, were image-guided planning has become an increasingly accepted procedure under the term of ‘neuronavigation’, complex preoperative planning and simulation are mandatory and represent an important part of the total duration of an operative procedure [4].

However, simulations, such as the real-time visualization of movements during manipulations, or the transfer of tactile sensations to the surgeon, or the visualization of the effect of robotic activities, provide a formidable challenge for high-end graphical computing and other disciplines. To reduce the required amount of graphical power and nevertheless enable planning and simulation of neurosurgical procedures on actual patient's datasets, the use of a combination of volume- and surface-rendered data for visualization and simulation seems most practical [5]. Three-dimensional Magnet Resonance Imaging (MRI) datasets of the brains from actual patients are used to preplan surgical approaches and to simulate views of the outer surface of the head as well as of inner surfaces, such as the ventricular system or cystic brain lesions, by aid of virtual endoscopy. Surgical manipulations are trained by use of segmented surface models of the ventricular system.

The integrated planning and simulation station called ROBO-SIM is designed for manipulator-assisted virtual procedures through a trepanation in the skull of 1–2 cm diameter and a miniaturized approach of a few millimetres diameter to target areas in the depth of the brain and its ventricular system.

Section snippets

Neurobot

ROBO-SIM is part of the operating system ROBOSCOPE (project of the EU-Telematics program), including the robot arm NEUROBOT (Fig. 1). For real surgical interventions, NEUROBOT is used by the surgeon as an active manipulator with inbuilt robotic capabilities such as active constraints, within permitted regions, precise pattern control and the ability to automatically track structures as they move and deform. For ROBO-SIM the robot is used as an input device. Thus, the surgeon who plans and

ROBO-SIM

The computer platform for the development and use of the system is a SGI Onyx2 Infinite Reality with two MIPS10000 CPUs. The rendering engine bases on OpenGL [6] for surface rendering and OpenGL Volumizer [7] for virtual endoscopy by using direct volume rendering. To speed up the rendering, a field of view rendering facility is used, which allows only the data visible through the field of view of an endoscope to be rendered. This enables an update frequency of about 10 frames per second for

Conclusions

The development of surgical simulators, comparable to flight simulators, has been initiated in a number of institutions (see, for example, [15], [16], [17], [18], [19], [20]). Most of these simulators are working with simplified models of the human anatomy instead of using the anatomy of an actual patient. ROBO-SIM is able to use directly the digital imaging datasets of the actual patient's neuro-anatomy. A simulation of minimally invasive neurosurgical procedures is considerably more complex

Acknowledgements

This project was partly supported by the Austrian Ministery of Research and by the European Commission, programme telematics (no. 4.018). The segmentations of the ventricles are done by algorithms provided by INRIA, Sophia Antipolis.

References (20)

  • A.C. Jambon et al.
  • T. Clerici, J. Lange, A. Zerz, S. Beller, G. Szinicz, U.O. Losert, H. Siegl, Educational Opportunities in Minimally...
  • A. Radetzky et al.

    Elastodynamic shape modeler: a tool for defining the deformation behavior of virtual tissues

    RadioGraphics

    (2000)
  • L.M. Auer et al.

    Virtual endoscopy for planning and simulation of minimally invasive neurosurgery

    Neurosurgery

    (1998)
  • L.M. Auer, A. Radetzky, C. Wimmer, G. Kleinszig, F. Schröcker, D.P. Auer, H. Delingette, B. Davies, D.P. Pretschner,...
  • R.S. Wright et al.

    OpenGL SuperBible

    (2000)
  • R. Grzeszczuk, C. Henn, R. Yagel, Advanced Geometric Techniques for Ray Casting Volumes, Proc. of ACM SIGGRAPH 98...
  • A. Radetzky, F. Schröcker, L.M. Auer, in: Westwood, J.D. (Ed.), Improvement of Surgical Simulation Using Dynamic Volume...
  • B. Lichtenbelt et al.

    Introduction to Volume Graphics

    (1998)
  • X. Pennec, P. Cachier, N. Ayache, Understanding the ‘Demon's Algorithm’: 3D Non-rigid Registration by Gradient Descent,...
There are more references available in the full text version of this article.

Cited by (0)

View full text