“…In order to do so, an common strategy [1,7,9,10,11,15,16,18,21,24,25,30] is to reconstruct one initial image I 0 and a trajectory of deformations t → ϕ t such that a each time t the image φ t · I 0 (deformation of I 0 by φ t ) matches the observed data. Then the framework has two intertwined components, estimation of I 0 and estimation of t → ϕ t , that can be alternatively performed in an iterative optimization scheme.…”