Topic 1: Cooperative Fusion of Anatomical, Physiological and Positional Data for Planning and Performance of Neurosurgical Procedures

Introduction

Data fusion in medical imaging is not restricted to image fusion. Data fusion includes extraction of relevant multimodal information from different modalities and its integration to help the decision making process.

Data fusion problems are resulting from a double issue:

• There is no single modality providing both anatomical and physiological information.
• Information provided by different modalities may be in agreement or of a complementary nature.

Our research concerns multimodal image guided neurosurgery in its preplanning, planning and surgery stages. For each stage, we describe main objectives, methods, applications and validation processes.

Pre-Planning

The pre-planning step includes mono or multimodal image analysis (e.g. segmentation, detection) and registration. Our objectives consist in the characterization of the target volume and its functional environment, in the understanding and the validation of new imaging modalities (i.e. functional imaging). For that purpose, methods include development and validation of image registration algorithms and the extraction (i.e. segmentation or detection) of relevant and significant information from these different imaging modalities. In this context we develop segmentation, registration, visualization and interaction tools. Applications include the study of the added value of multimodal information for the understanding of the epileptogenic network and for the surgery of cerebral lesions located in central areas, as well as the comprehension and validation of non invasive functional imaging (e.g. perfusion MRI in epilepsy). Concerning the validation of these image processing tools, we develop realistic models allowing the simulation of SPECT images from MR images, therefore providing directly a ground truth. This approach also allows to simulate in a realistic way pathologies from clinical data sets for robustness studies.

Some publications:

• Jannin P, Grova C, and Gibaud B. Fusion de données en imagerie médicale : une revue méthodologique basée sur le contexte clinique. ITBM-RBM Innovation et Technologie, (22):196-215, 2001
• Jannin P, Grova C, and Gibaud B. Applications of NDT data fusion, chapter Medical applications of NDT data fusion, pages 227-267. Kluwer academic, gros x.e edition, 2001
• B Aubert-Broche, C Grova, P Jannin, I Buvat, H Benali and B Gibaud, "Detection of inter- hemispheric asymmetries of brain perfusion in SPECT", Phys. Med. Biol. 48 No 11 (7 June 2003) 1505-1517
• Godey B, Schwartz D, de Graaf JB, Chauvel P, and Liegeois-Chauvel C. Neuromagnetic source localization of auditory evoked fields and intracerebral evoked potentials : a comparison of data in the same patients. Clinical Neurophysiology, 112(10):1850-1859, 2001.
• Grova C., Biraben A., Scarabin J.M., Jannin P., Buvat I., Benali H. and Gibaud B. A methodology to validate mri / spect registration methods using realistic simulated spect data. In MICCAI 2001, 2001
• Le Rumeur E., Allard M., Poiseau E., and Jannin P. Role of the fmri sensory stimulation type for the understanding of normal and lesional cortex connections for presurgical brain mapping. Journal of Neurosurgery, 93(3):427-431, 2000
• Schwartz D., Badier J.M., Bihouée P., and Bouliou A. Evaluation of a new meg-eeg spatio-temporal approach using realistic sources. Brain Topography, 11(4):279-289, 1999
• Biraben A., Taussig D., Bernard A.M., Vignal J.P., Scarabin J.M., Chauvel P., and Duncan R. Video-eeg and ictal spect in three patients with both epileptic and non- epileptic seizures. Epileptic Disorders, 1(1):51-55, 1999

Planning

The main objective of this planning step is to facilitate the definition of the surgical script. We consider the planning step as the selection, by the neurosurgeon, of relevant information (called entities) extracted from preoperative multimodal images of the patient. Then, for each selected entity, we define its location(s) and role(s) in the surgical script. For that purpose, we define models of surgical procedures, as well as visualisation and interaction tools to select and integrate the relevant multimodal entities. The applications concern the definition of a multimodal planning step, the creation of a simulation step for the surgical training based on the images of the patient, and the management of the multimodal data along the surgical process. This approach improves the tracability of the various processes involved in image guided surgical procedures. Models of surgical procedures are validated both in a retrospective and a prospective way.

Jannin P., Raimbault M., Morandi X., and Gibaud B. , "Modeling Surgical Procedures for Multimodal Image-Guided Neurosurgery". Journal of Computer Aided Surgery, 2003 (in press)
Jannin P., Raimbault M., Morandi X., Seigneuret E., and Gibaud B. Design of a neurosurgical gestures model to improve neuronavigation procedures. In H.U. Lemke, M.W. Vannier, K. Inamura, A.G. Farman, and K. Doi, editors, Computer Assisted Radiology and Surgery 2001, pages 102-107. Elsevier, 2001

Image guided surgery

The main objective of this step is the matching between pre-operative virtual images of the patient and the real views of the patient in the operating room. For that purpose, we address problems related to neuronavigation, 3D stereoscopic reconstruction with strong calibration, augmented reality. We also introduce the concept of enhanced virtuality. The applications concern the development of multimodal neuronavigation and the study of the intraoperative modification of the anatomy due to surgery. Validation of enhanced virtuality and 3D stereoscopic reconstruction use both simulations and clinical tests. We study the overall uncertainty of the enhanced virtuality capabilities.

Some publications:

• Jannin P, Morandi X, Fleig O.J, Le Rumeur E, Toulouse P, Gibaud B. and Scarabin J.M. Integration of sulcal and functional information for multimodal neuronavigation, Journal of Neurosurgery, 96:713-723, 2002
• Fleig O., Jannin P., Scarabin J.M., and Devernay F. Stereo reconstruction of the surgical field in neurosurgery. In H.U. Lemke, M.W. Vannier, K. Inamura, A.G. Farman, and K. Doi, editors, Computer Assisted Radiology and Surgery 2001, pages 259-264. Elsevier, 2001
• Jannin P., Fleig O.J., Seigneuret E., Grova C., Morandi X., and Scarabin J.M. A data fusion environment for multimodal and multi-informational neuro-navigation. Journal of Computer Aided Surgery, 5(1):1-10, 2000
• Jannin P., Seigneuret E., Morandi X., Fleig O.J., Riffaud L., Le Goualher G., Brassier G., and Scarabin J.M. Repérage sulcal et neuro-navigation dans la chirurgie des cavernomes supratentoriels. Neurochirurgie, 46(6):534-540, 2000

Main collaborations : INRIA Rennes (Vista) et Sophia Antipolis (ChIR), U494 Paris, CHRU Rennes, CAC Rennes