gms | German Medical Science

Jahrestagung der Gesellschaft für Medizinische Ausbildung (GMA)

16.11. - 18.11.2007, Hannover

"CranioTrain" - a novel concept for training craniotomy localizations

Vortrag/Lecture

  • author Thomas Rodt - Medizinische Hochschule Hannover, Institut für Radiologie, Hannover, Deutschland
  • corresponding author Ute von Jan - Medizinische Hochschule Hannover, Institut für Medizinische Informatik, Hannover, Deutschland
  • G. Köppen - Affiliated Hospital Univ. of Münster, Dept. of Neurosurgery Gilead, Bielefeld-Bethel, Bielefeld, Deutschland
  • author Joachim Krauss - Medizinische Hochschule Hannover, Neurochirurgische Klinik, Hannover, Deutschland
  • author Herbert Matthies - Medizinische Hochschule Hannover, Institut für Medizinische Informatik, Hannover, Deutschland

Jahrestagung der Gesellschaft für Medizinische Ausbildung - GMA. Hannover, 16.-18.11.2007. Düsseldorf: German Medical Science GMS Publishing House; 2007. Doc07gma150

Die elektronische Version dieses Artikels ist vollständig und ist verfügbar unter: http://www.egms.de/de/meetings/gma2007/07gma150.shtml

Veröffentlicht: 14. November 2007

© 2007 Rodt et al.
Dieser Artikel ist ein Open Access-Artikel und steht unter den Creative Commons Lizenzbedingungen (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.de). Er darf vervielfältigt, verbreitet und öffentlich zugänglich gemacht werden, vorausgesetzt dass Autor und Quelle genannt werden.


Gliederung

Text

Purpose: Determining the correct placement of craniotomies for neurosurgical procedures performed for certain pathologies such as tumors, vascular pathologies or hemorrhage has great influence on the outcome, since complications may arise by choosing an inappropriate localization. Intra-operative manipulation of deeper structures may be considerably harder with an adverse angle for accessing the pathology, whereas pathologies near the brain's surface may require an enlargement of the craniotomy which is later detrimental to healing and stability.

For the inexperienced resident, determining its correct placement is a difficult process due to the skull's complex anatomy as well as the different angulations of the provided image data such as CT and MRI. In clinical routine, one generally uses distances from certain anatomical structures to localize the ideal place for the craniotomy, which is aggravated by having to do this on the surface of the convex skull. This method also requires considerable practice to be able to confidently find the correct localization.

Although neuronavigation can provide access to spatial information of the image data while performing neurosurgical procedures, which can simplify the process of finding the craniotomy's ideal placement, it is at present not commonly employed for standard neurosurgical procedures or in emergency cases.

In this study, we present a novel, software based approach for training how to determine the correct craniotomy localization.

Methods: A grid was engraved on a standard anatomical skull phantom using a distance of 1cm between the horizontal lines. The vertical lines were placed in 10° steps. The intersection points of the grid were labeled to provide a reference to be used in the training software (see figure 1 [Fig. 1]).

For reconstructing a virtual model of the skull's surface, the phantom was scanned using a helical multi-slice CT (120 kV, 80 mA, pitch 3, collimation 1.25 mm). Image reconstruction was done with180 LI with a FOV of 25 cm and 1 mm reconstruction interval. A high-resolution virtual 3-D model of the skull was then generated from the CT volume data by employing the surface reconstruction methods.

To determine the position data for the grid intersection points, an infrared-optical navigation system was used. The skull phantom was attached to a reference frame using a Mayfield clamp. The 3-D coordinates of the grid's intersection points as well as those of certain anatomical landmarks – which could also be easily determined on the skull’s virtual model – were acquired with a point by point exploration. Registration between the CT volume based virtual model and the grid data acquired using the navigation system was then performed based on the position data of these anatomical landmarks. The 3-D surface model of the skull was then aligned with the coordinates of the grid based on the coordinate transformation determined in the registration process.

All data were then integrated into a platform independent training software called CranioTrain, which was developed using the Visualization Toolkit (VTK 5.0.2, http://www.vtk.org/) and the Fast Light Toolkit (FLTK 1.1.7 , http://www.fltk.org/). In this program, the user can first choose from a number of provided training cases representing different pathologies. For each case provided in the software, a case report is available. In the next step, it is possible to view the pathology, which is represented by a geometric object, overlaid into a 2-D slice view of the original skull CT data or into an MRI dataset (see figure 2 [Fig. 2]). Depending on the chosen case, the image data can be viewed either in axial orientation alone or in coronal and sagittal orientation as well.

The user must then locate the – in his opinion – ideal craniotomy location, e.g. by measuring distances between the pathology and reference anatomical structures. In the next step, the user determines the center of the craniotomy with help of the skull phantom. The label of the corresponding intersection point of the reference grid is then entered into the software. After entering the craniotomy position, the program then shows the chosen as well as the ideal center point of the craniotomy in the 3-D view and also provides the distance of both points on the skull's surface. It is also possible make the skull transparent to give the user the possibility to view the position of the pathology within the 3-D model (see figure 3 [Fig. 3]).

Results: Measuring the distances between the pathology and reference anatomical structures yielded good localizations of the craniotomy spots if varying angulations of the cross-sectional data sets and the surface bend of the calvaria were taken into account. The presented concept for learning the correct craniotomy localization proved to be easy to handle and can be beneficial in neurosurgical training.

Conclusion: The evaluation of the teaching benefit suggested that localizing a craniotomy was more precise if training cases had been performed within CranioTrain beforehand.