High-resolution dynamic cardiac MRI on small animals using reconstruction based on Split Bregman methodology

Montesinos, P.; Abascal, Juan Felipe Pérez Juste; Chamorro Servent, Judit; Chavarrías, Cristina; Benito, M.; Vaquero, Juan J.; Desco, Manuel

doi:http://dx.doi.org/10.1109/NSSMIC.2011.6152633

High-resolution dynamic cardiac MRI on small animals using reconstruction based on Split Bregman methodology

Citation

Montesinos P, Abascal J, Chamorro J, Chavarrias C, Benito M, Vaquero JJ, Desco M. High-resolution dynamic cardiac MRI on small animals using reconstruction based on Split Bregman methodology. In: Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2011 IEEE; 2011 Oct 23-29; València, Spain. p. 3462-4. DOI: 10.1109/NSSMIC.2011.6152633

Abstract

Dynamic cardiac magnetic resonance imaging in small animals is an important tool in the study of cardiovascular diseases. The reduction of the long acquisition times required for cardiovascular applications is crucial to achieve good spatiotemporal resolution and signal-to-noise ratio. Nowadays there are many acceleration techniques which can reduce acquisition time, including compressed sensing technique. Compressed sensing allows image reconstruction from undersampled data, by means of a non linear reconstruction which minimizes the total variation of the image. The recently appeared Split Bregman methodology has proved to be more computationally efficient to solve this problem than classic optimization methods. In the case of dynamic magnetic resonance imaging, compressed sensing can exploit time sparsity by the minimization of total variation across both space and time. In this work, we propose and validate the Split Bregman method to minimize spatial and time total variation, and apply this method to accelerate cardiac cine acquisitions in rats. We found that applying a quasi-random variable density pattern along the phaseencoding direction, accelerations up to a factor 5 are possible with low error. In the future, we expect to obtain higher accelerations using spatiotemporal undersampling.