7.4.1 Revues Internationales avec comit´e de lecture
[Frindel 09b] C. Frindel, M. Robini, P. Croisille and Y. M. Zhu, Comparison of reg- ularization methods for human cardiac diffusion tensor MRI, MedIA, volume 13, (2009), pp. 405-418.
[Frindel 09d] C. Frindel, M. Robini, J. Schaerer, P. Croisille and Y. M. Zhu, A graph- based approach for cardiac tractography, submitted to MRM in September 2009.
7.4.2 Congr`es internationaux avec actes
[Frindel 07] C. Frindel, M. Robini, S. Rapacchi, E. Stephant, Y. M. Zhu and P. Croisille, Toward In-Vivo Diffusion Tensor MRI on Human Heart using Edge-Preserving Regularization, Proc. 29th Int. Conf. IEEE EMBS, Lyon (France), August 2007, pp. 6007- 6010.
[Frindel 08] C. Frindel, J. Schaerer, P. Gueth, P. Clarysse, Y. M. Zhu and M. Robini, A global approach to cardiac tractography, Proc. 5th Int. Conf. IEEE ISBI, Paris (France), May 2008, pp. 883-886.
[Frindel 09c] C. Frindel, M. Robini, J. Schaerer, P. Croisille and Y. M. Zhu, Cardiac Fibre Trace Clustering for the interpretation of the Human Heart Architecture, Proc. 4th Int. Conf. FIMH, Nice (France), LNCS 5528, June 2009, pp. 39-48.
7.4. PUBLICATIONS DE L’AUTEUR 131
[Frindel 09e] C. Frindel, M. Robini, J. Schaerer, P. Croisille and Y. M. Zhu, Improved Global Cardiac Tractography with Simulated Annealing, Proc. Int. Conf. IEEE ICIP, Cairo (Egypt), November 2009.
7.4.3 Congr`es internationaux avec r´esum´es courts
[Frindel 09a] C. Frindel, S.Rapacchi, M. Robini, H. Wen, M. Viallon, L. Fanton and P. Croisille, In vivo cardiac NMR Diffusion Weighted Imaging (DWI) for the human heart : improved quantification of FA and MD by edge-preserving regularization, ISMRM, Hon- olulu (USA), April 2009, pp. 3771.
[Stephant 09] E. Stephant, C. Frindel, M. Robini, L. Fanton, M. Viallon and P. Croisille, Analytic description of MR diffusion indices in ex-vivo human hypertrophic car- diomyopathy, ISMRM, Honolulu (USA), April 2009, pp. 3588.
7.4.4 Congr`es nationaux
[Rapacchi 07] S. Rapacchi, C. Frindel, E. Stephant, Y.M. Zhu and P. Croisille, Vers un mod`ele statistique des fibres du myocarde humain en IRM du tenseur de diffusion, Journ´ees de Recherche en Imagerie M´edicale (JRIM), 2007.
Annexe
A
Towards In Vivo Diffusion Tensor MRI on
Human Heart using Edge-Preserving
Regularization
1
Sommaire
A.1 INTRODUCTION . . . 133 A.2 MATERIALS AND METHODS . . . 134 A.2.1 Data Acquisition Protocols . . . 134 A.2.2 Edge-Preserving Regularization . . . 135 A.2.3 Coherence Index . . . 136 A.3 RESULTS . . . 136 A.3.1 Protocol comparison . . . 136 A.3.2 Edge-preserving restoration of DW data . . . 137 A.4 CONCLUSIONS . . . 139
Abstract – We investigate the noise sensitivity in various Diffusion Tensor MRI acqui- sition protocols in sixteen human ex vivo hearts. In particular, we compare the accuracy of protocols with various numbers of excitations and diffusion sensitizing directions for estimating the principal diffusion directions in the myocardium. It is observed that noise sensitivity decreases as the number of excitations and the number of sensitizing directions increase (and hence as the acquisition time increases). To reduce the effects of noise and to improve the results obtained with a smaller number of excitations and/or a smaller num- ber of sensitizing directions, we introduce a 3-D edge-preserving regularization method operating on diffusion weighted images. It allows to maintain the quality of the principal diffusion direction field while minimizing the acquisition time, which is a necessary step for in vivo diffusion tensor MR imaging of the human heart.
A.1
INTRODUCTION
Diffusion Tensor Magnetic Resonance Imaging (DT-MRI) – which measures the diffu- sion of water molecules along various directions in tissues – provides unique biologically and clinically relevant information that cannot be obtained with other imaging modalities.
1Proc. 29th Int. Conf. IEEE EMBS, Lyon (France), pp. 6007-6010, Aug. 2007
This information includes parameters that help to characterize tissue composition as well as the physical properties of tissue constituents, tissue microstructure and its architectural organization. The construction of the diffusion tensor distribution requires the acquisition of a set of diffusion-weighted (DW) images associated with diffusion sensitization along Nd = 6 or more non-collinear gradient directions. More specifically, the diffusion tensor
D is related to the DW measurements Si (i = 1, . . . , Nd) according to the Stejskal-Tanner
diffusion equations : Si = S0exp(−bgiTDgi), where gi is the encoding gradient directionas-
sociated with Si, S0 is an MR measurement without diffusion-sensitizing gradients, and
the constant b is the diffusion factor.
DT-MRI is very sensitive to noise and artifacts, and thus the estimation of the diffusion tensor field, the diffusion direction field and the parametric maps (i.e. fiber coherence) is subject to errors. Error propagation in the DT-MRI processing chain has been studied extensively (e.g., [Basser 96], [Pierpaoli 96], [Conturo 96], [Martin 99b]), but the perfor- mances of different DTI acquisition schemes have been rarely compared (see [Skare 00b] for the brain). In fact, because of the lack of standard validation methods, it is very difficult (if not impossible) to define a noise sensitivity measure to compare different DTI schemes quantitatively.
There are two main possibilities to improve the signal-to-noise ratio at the time of data acquisition :
(i) to increase the number Nd of diffusion sensitizing directions ;
(ii) to increase the number Ne of excitations used for signal averaging.
However, it must be kept in mind that both methods increase the acquisition time. Our main goal is to come out with efficient acquisition protocols by comparing the results we obtain with different set of values for Nd and Ne and thus for different acquisition
times — by efficiency, we mean the best compromise between the acquisition time and the quality of the direction field obtained via edge-preserving restoration of the DW images. More specifically, we consider the imaging of sixteen human ex vivo hearts (twelve healthy hearts and four explanted hearts) using nine DTI acquisitions protocols. We will compare the principal diffusion direction fields (defined by the diffusion tensors computed from the Stejskal-Tanner equations) qualitatively and quantitatively using a fiber coherence index. We propose to denoise the sets of DW images by edge-preserving regularization. As our experiments will show this approach has two advantages. Firstly, it allows reduction of the acquisition time while preserving the quality of the measurements and, secondly, it has a real meaning in the case of the heart which is a very well organized and structured organ. Our regularization approach is a 3-D extension of the method introduced in [Charbon- nier 97] for tomography. Given a reference result (typically, the protocol with the longest acquisition time), we show that similar information content quality can be obtained with the other protocols.