- Oral presentation
- Open Access
Improving the accuracy of multi breath-hold diffusion tensor MRI tractography of the heart using dynamic motioncorrection
Journal of Cardiovascular Magnetic Resonance volume 15, Article number: O81 (2013)
Diffusion Tensor MRI (DTI) tractography of the human heart can be performed in vivo, but requires multiple breatholds per slice to achieve adequate SNR [1, 2]. The physiological noise inherent to multiple breatholds results in a diffusion-encoded volume in which the data vary as a function of space and time. This may hinder the assessment of diffusion based-indices in the heart and accurate tractography of myofiber architecture. In this study, we investigate the effect of a novel automated motion correction method on in vivo DTI of the heart.
DTI of 10 normal volunteers was performed on a 3T clinical scanner (Skyra, Siemens) with the following parameters: 6 diffusion-encoding directions, b=350 s/mm2, fat saturation, TR/TE=1100/23 ms, BW=2442 Hz/pixel, spatial resolution=2.7x2.7x8 mm3, 8 averages. This required 24 separate breatholds for a 3-slice diffusion-encoded volume. On a chosen reference frame, n radial scanlines (Figure 1A) starting at the LV-RV junction and crossing at the center of the LV cavity were defined as the registration axes. Motion was estimated by matching the intensity gradient profiles of corresponding radial scanlines between frames and at all 3 slice levels. A rigid registration described by a set of translations and rotations was then determined between pairs of images, iteratively (Figure 1B). Convergence was achieved upon reaching a global minimum of the quadratic error at each level. The signal-noise-ratio (SNR) was calculated at each pixel r with a given number of averages (repetitions) t, such that SNRt(r) = MEANt(r) / SDt(r) . SNR, with and without motion correction, was measured in each voxel at end-diastole and end-systole. Fiber tracking was performed with a 4th order Runge-Kutta approach .
At end-diastole, an increase of 24% was observed in the SNR of the T2-weighted images, and an average SNR increase of 21% was seen in the diffusion-weighted images (Figure 1C). At end-systole, a relatively uniform SNR increase of 15% was seen with motion correction for diffusion-free and diffusion-weighted images. Figure 2 shows tractograms in the lateral wall of the left ventricle at end-diastole and end-systole. Motion correction frequently increased the fiber track lengths, consistent with the observed increase in SNR, and in some cases also rectified their orientation (see endocardial fibers at end-systole, Figure 2B vs. 2D).
DTI tractography can be performed in vivo over multiple breatholds, but is affected by physiological noise. Here, we show that motion correction over space and time can reduce the noise produced by multiple breatholds while increasing the quality of the resulting tractograms. This approach paves the way for the use of free breathing navigator-based DTI for high-resolution tractography of the heart.
Reese TG, et al: . Magn Reson Med. 1995
Nielles-Vallespin S, et al: . Magn Reson Med. 2012
Reeder SB, et al: . Magn Reson Med. 2005
Mekkaoui , et al: . J Cardiovasc Magn Reson. 2012
About this article
Cite this article
Mekkaoui, C., Nielles-Vallespin, S., Jackowski, M.P. et al. Improving the accuracy of multi breath-hold diffusion tensor MRI tractography of the heart using dynamic motioncorrection. J Cardiovasc Magn Reson 15, O81 (2013). https://doi.org/10.1186/1532-429X-15-S1-O81
- Motion Correction
- Fiber Track
- Rigid Registration
- Physiological Noise
- Motion Correction Method