gms | German Medical Science

56. Jahrestagung der Deutschen Gesellschaft für Neurochirurgie e. V. (DGNC)
3èmes journées françaises de Neurochirurgie (SFNC)

Deutsche Gesellschaft für Neurochirurgie e. V.
Société Française de Neurochirurgie

07. bis 11.05.2005, Strasbourg

Multimodal image fusion of low and high field MRI

Bilddatenfusion intraoperativer 0,15 Tesla und präoperativer 1,5 Tesla Daten für die multimodale navigierte Hirnchirurgie

Meeting Abstract

  • corresponding author R. Krishnan - Department of Neurosurgery, Johann-Wolfgang-Goethe-University, Frankfurt/Main
  • E. Firle - Fraunhofer Institut für Computer Graphics, Cognitive Computing & Medical Imaging, Darmstadt
  • E. Herrmann - Department of Neurosurgery, Johann-Wolfgang-Goethe-University, Frankfurt/Main
  • G. Marquardt - Department of Neurosurgery, Johann-Wolfgang-Goethe-University, Frankfurt/Main
  • A. Raabe - Department of Neurosurgery, Johann-Wolfgang-Goethe-University, Frankfurt/Main
  • V. Seifert - Department of Neurosurgery, Johann-Wolfgang-Goethe-University, Frankfurt/Main

Deutsche Gesellschaft für Neurochirurgie. Société Française de Neurochirurgie. 56. Jahrestagung der Deutschen Gesellschaft für Neurochirurgie e.V. (DGNC), 3èmes journées françaises de Neurochirurgie (SFNC). Strasbourg, 07.-11.05.2005. Düsseldorf, Köln: German Medical Science; 2005. DocP047

Die elektronische Version dieses Artikels ist vollständig und ist verfügbar unter: http://www.egms.de/de/meetings/dgnc2005/05dgnc0315.shtml

Veröffentlicht: 4. Mai 2005

© 2005 Krishnan 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

Objective

Multimodal imaging provides valuable and complementary information to the neurosurgeon for diagnosis and therapy planning. Recently a new intraoperative MRI system (PoleStar N20) was developed for guiding and controlling the surgical treatment of brain tumors. The main advantages – clear delineation of the lesion, “under the surface” vision, real time feedback on the extent of resection and the position of residual tumor tissue – must be weighed against the disadvantage of lower image quality of the intraoperative scans. Fusion of high quality preoperative images is merely desirable as reported in this study.

Methods

A phantom data set and clinical data sets of two patients acquired both in a PoleStar 0,15 T and a Siemens Magnetom 1,5 T scanner were used for image matching. The applied automatic alignment procedure is based on the information theoretic approach of maximizing mutual information of two arbitrary images, a voxel-based similarity measure of the statistical dependency between two datasets. It evaluates the amount of information that one variable contains about the other. Distortions in the low-filed images were evaluated. The software used was developed by the Fraunhofer Institute and runs on a standard PC, taking the limitations of computational power in the OR into account by developing strategies for accelerating the registration process.

Results

Although the field of view of the low field scanner represents just a part of the high field scanner field of view and intraoperative images showed anatomical displacement fusion of the data sets was accurate in all cases. Time for fusion varied between 0.5 and 2 minutes. In the phantom study, image distortion in the outer FOV in the low field images was quantified, highlighting the necessity to center the 0,15 T FOV on the surgical target. The intraoperative progression of tissue displacement was documented in the fused data sets.

Conclusions

Multimodal image fusion of low and high field MRI is a versatile and powerful tool to augment image guidance during brain surgery aiming for multi-informational online navigation. Image co-registration applied to compare anatomic information of the changing surgical site is helpful for visualizing changes of anatomical structures. Fully automatic registration algorithms with high levels of connectivity and compatibility are needed to be successfully applicable in clinical routine applications.