Titre : Biomechanical graph matching for hepatic intra-operative image registration

Abstract:

This thesis presents the development of an automatic elastic registration methodbased on matching of vascular graphs extracted from both pre-operative and intra-operative images. The method is able to fuse accurate pre-operative information onto an organ undergoing small to large deformations during surgery, to compensatefor the limited details provided by intra-operative imaging modalities and improvethe visualization of tumor(s), vasculature and other important internal structures.Although methods dedicated to non-rigid graph matching exist, they are not efficient when noise, topology changes and large intra-operative deformations arepresent. The first contribution presented is a biomechanical graph matching method(BGM) that builds on the work ofSerradell et al.(2015). BGM combines the GPRmatching with a biomechanical model of the organ, as a mean to discard matchinghypotheses which would lead to non-plausible deformations (Garcia Guevara et al.,2018). However, BGM is not robust to noise, only matches limited size graphs andhas high computation time.The second contribution is the Adaptive ComplianceGraph Matching (ACGM) method (Garcia Guevara et al.,2019), which allows toefficiently find the best graph matches with a novel compliance-based search andan adaptive rigid to soft approach. This reduces the computation time by predict-ing first the most plausible matching hypotheses. It also reduces the sensitivity onthe search space parameters and improves the registration quality.The proposedregistration methods are evaluated with realistic synthetic and real porcine dataset, showing that ACGM is compatible with intra-operative constraints.

Rapporteurs:     Patrick Clarisse- CNRS (section 7) CREATIS (Lyon)

Stephanie Speidel- National Center for Tumor Diseases (NCT)

Examinateurs:  Jean-Francois Mari- Université de Lorraine

Ingerid Reinertsen- SINTEF Digital, Health Researc

Encadrants:    Marie-Odile Berger- INRIA Equipe MAGRIT

Stéphane Cotin- INRIA Equipe MIMESIS