TIBM: Registration

Registration is a crucial step in medical image processing. Registration may be mono- or multimodal, rigid (intra-patient) or deformable (inter-patient) and may involve two or more images. Thus, we have proposed a deformable registration method for jointly mapping of a set of images 2-NHHA12, which is a necessary prerequisite for conducting population studies. A registration algorithm dedicated to retinal images has also been developed 2-FLP11. Contributions were made for the non-rigid registration of binary images 2-GNKF12 as well as for the warping of binary images under topological constraints 2-FPNC11. The problem of warping 4th order tensor fields was also tackled 4-GRNH11, the 4th order tensor being a mathematical model used in diffusion MRI to represent fiber crossings. Estimating non-rigid registration between two exams of a given subject is an interesting tool for the quantification of cerebral atrophy over time. In this context, a method was proposed for estimating the uncertainty in atrophy quantification using a Bayesian framework 2-SRHR13.

PhD thesis

• S. Sharma, Estimation de l'atrophie cérébrale en IRM : application à la sclérose en plaques, september 2011
• A. Charnoz, Recalage d'organes intra-patient à partir de l'étude de leur réseau vasculaire : application au foie, january 2007
• V. Noblet, Recalage non rigide d'images cérébrales 3D avec contrainte de conservation de la topologie, march 2006
• O. Musse, Contribution à la mise en correspondance non rigide d’images médicales : une approche paramétrique hiérarchique sous contraintes topologiques. Application au recalage déformable du cerveau en imagerie IRM, december 2000
• C. Nikou, Contribution au recalage d’images médicales multimodales : approches par fonctions de similarité robustes et modèles déformables physiques sous contraintes statistiques, may 1999.

→ Back to Biomedical image processing page