Low-rank tensor based methods for accelerating deformable 3D medical image registration

Justin Buhendwa Nyenyezi (Speaker)

Activity: Talk or presentation typesOral presentation


The registration process aims to determine the suitable spatial transformation that deforms an image, such that it becomes as similar as possible to a fixed reference image. Although this problem seems relatively easy to address, its numerical resolution faces significant technical and practical issues. Algorithms with linear complexity O(N) based on fast Discrete Cosine Transforms or additive operator splitting are yet available.
However, this linear complexity is far too large for large 3D medical images.
Our aim is to replace this linear complexity by logarithmic complexity O(logN). we propose to use a compressed representation of data with a given accuracy using preconditioning techniques in tensor train format .
Period20 Apr 201721 Apr 2017
Held atComex: Belgian mathematical optimization workshop, Belgium
Degree of RecognitionNational


  • Low-rank, preconditioner, tensor train, 3D medical images, large linear systems