InitRech 2015/2016, sujet 6
Summary
This article speaks about the training simulators for the medical environment, and more exactly for the Intracranial Aneurysm Surgery (IAS). This kind of simulator has to get closer as much as possible to the reality. It means that it can offer an advanced visual and physical reality as well as a precise haptic feedback. The authors describe in particular the standard IAS procedure to show what the simulator has to supply. The purpose of the document is to demonstrate the feasibility of a realistic training for vascular neurosurgery procedure using virtual reality systems.
They begin by giving some examples of works already make on this subject. They speak at first about the brain tissue deformation models: it is mechanical simulations of brain tissues. The main drawback is that the brain model should be selected according to the application. Then they speak about the haptic rendering of soft tissues: they say that the link between the quality of haptic feedback and the precision of the interactive simulation is not direct because the real-time constraints of haptic feedback are bigger.
This simulator integrates a 3D modeling of virtual surgical scene. This modelling was obtained thanks to 3D imaging data of an anonymous patient carrying a right unruptured middle cerebral artery aneurysm. A 3D modelling software was used to design the surgical tools. Surgical complications are taken into account: the bleeding fills the operative field when accidental collision with a surgical device occurs. For that, a clipping plane obstructs the field. The user can reduce the blood level by using a suction tool and he can stop it by applying a temporary clip at the source of the bleeding. In a real situation, the final clip should be definitely placed five minutes after the bleeding has started : there is a counter on screen of the simulator to indicate the remaining time to place the final clip. As regards the hardware setup, the simulator integrates a double haptic force feedback virtual reality application. The surgical scene is represented by a non-stereoscopic 3D environment and on a 2D flat screen. The exchange of tools is possible thanks to some scripts which were developed in Python and the forces feedback was rendered thanks to two haptic arms.
Then, the authors describe how they reproduced a realistic behavior of the tissues. To model this dynamic behavior, they use the Newton’s second law and some mathematics tools to solve the equations. As regards the brain deformation, they use a FEM model to balance real-time performance and precision. Two separate meshes were used for the two lobes of the brain that are separated during the sylvian fissure dissection.
The vessel model combines a FEM model of the surface of the vessel and a pressure constraint to model the pressure induced by the blood. Obviously, it is very complex to simulate all the existing connections between the brain and the vessels, so the authors used a simplified model.
To model the connective tissues, they use springs. The spring links two points that belongs to the brain lobe models and to the vessel model. The user can cut the strands by using a surgical instrument and to simulate the tissues resistance, there is a time lapse between when the blade touches the strand and the time it is effectively cut.
In another section, the authors speak about the models and algorithms that are employed for the simulation of the interactions of the instruments with the tissues and the haptic rendering. They do a projection of the mechanical models in the space and they consider a mapping function which provides the violation of the constraint. Thanks to some mathematics tools, in particular a Gauss-Seidel iterative solver, contact and friction constraints are solved together in a block. The problem is that building the system takes more time than solving it. That is why they use a multi-threading approach: the system is shared between the simulation and a haptic loop. First, the system is build and solved in the simulation. Then this system is shared with the haptic feedback and re-solved at high rates.
The authors then give the results of their work. They propose four different pedagogic scenarios. In fact, these scenarios represent the different steps of the IAS procedure. The simulation time is stable even with a high number of constraints. But there are limitations: when the tools are in contact with the vessels, the haptic realism is reduced. Furthermore, the simulation stops after the release of the clip.
Some operators practiced on the prototype and gave a good opinion of this tool. These operators recommend to implement a 3D stereoscopic visualization, to redesign the surgical instruments aspects, to improve the haptic feedback realism and to develop different types of aneurysm: it is the future work to do to improve the prototype.
