FEM-based confidence assessment of non-rigid registration

Paul Baksic (1, 2) , Hadrien Courtecuisse (1, 3) , Matthieu Chabanas (2) , Bernard Bayle (1)

1 ICube - Laboratoire des sciences de l'ingénieur, de l'informatique et de l'imagerie
2 TIMC-IMAG-GMCAO - Gestes Medico-chirurgicaux Assistés par Ordinateur
3 MIMESIS - Computational Anatomy and Simulation for Medicine

Non-rigid registration is often used for 3D representations during surgical procedures. It needs to provide good precision in order to guide the surgeon properly. We propose here a method that allows the computation of a local upper bound of the registration confidence over the whole organ volume. Using a bio-mechanical model, we apply tearing forces over the whole organ to compute the upper bound of the degrees of freedom left by the registrations constraints. Confrontation of our method with experimental data shows promising results to estimate the registration confidence. Indeed, the computed maximum error appears to be a real upper bound.

Domaines: Modélisation et simulation; Imagerie médicale; Vision par ordinateur et reconnaissance de formes [cs.CV]

Paul Baksic, Hadrien Courtecuisse, Matthieu Chabanas, Bernard Bayle. FEM-based confidence assessment of non-rigid registration. Surgetica 2019, 2019, Rennes, France. ⟨hal-02238974⟩