Joint Estimation and Smoothing of Clinical DT-MRI with a Log-Euclidean Metric

Pierre Fillard (1) , Vincent Arsigny (1) , Xavier Pennec (1) , Nicholas Ayache (1)

1 EPIDAURE - Medical imaging and robotics

Diffusion tensor MRI is an imaging modality that is gaining importance in clinical applications. However, in a clinical environment, data has to be acquired rapidly at the detriment of the image quality. We propose a new variational framework that specifically targets low quality DT-MRI. The hypothesis of an additive Gaussian noise on the images leads us to estimate the tensor field directly on the image intensities. To further reduce the influence of the noise, we optimally exploit the spatial correlation by adding to the estimation an anisotropic regularization term. This criterion is easily optimized thanks to the use of the recently introduced Log-Euclidean metrics. Results on real clinical data show promising improvements of fiber tracking in the brain and we present the first successful attempt, up to our knowledge, to reconstruct the spinal cord.

Mots clés: TENSORS; DT-MRI; DTI; ESTIMATION; REGULARIZATION; FIBER TRACKING; LOG-EUCLIDEAN; RIEMANNIAN GEOMETRY

Domaines: Autre [cs.OH]

Pierre Fillard, Vincent Arsigny, Xavier Pennec, Nicholas Ayache. Joint Estimation and Smoothing of Clinical DT-MRI with a Log-Euclidean Metric. [Research Report] RR-5607, INRIA. 2005, pp.18. ⟨inria-00070400⟩