Robust free-breathing SASHA T1 mapping using high-contrast image-based registration
Journal of Cardiovascular Magnetic Resonance volume 18, Article number: P28 (2016)
Myocardial T1 correlates with fibrosis, but detection of sub-clinical disease requires accurate and precise T1 measurements, which are limited in breath-hold (BH) techniques. Navigator-gating (NAV) enables longer free-breathing (FB) acquisitions, but residual cardiac motion causes blurring without image registration. Variable flip angle (VFA) SASHA's  high accuracy and independent image acquisitions are well suited to this approach, however poor blood-tissue contrast makes accurate image registration difficult. We present a novel technique to generate high-contrast (HC) images to improve registration, enabling robust FB T1 mapping.
SASHA-VFA images have poor contrast due to the acquisition of the k-space center early in the bSSFP readouts. By increasing the flip angle to 120° after the k-space center, accumulated T2/T1 weighting generates blood-tissue contrast without affecting the accuracy of T1 maps calculated using the primary images. Additional low-frequency k-space lines collected following the primary images can be used to generate "key-hole" images with high contrast (Fig. 1), which can be used to motion correct the primary images. Sequence parameters include: 1.2/2.8 ms TE/TR, 340 × 255 mm FOV, 256 × 150 matrix, GRAPPA R=2, 65 phase encodes for the primary image, and 15 phase encodes at R=3 for the HC image.
14 healthy subjects were imaged on a 1.5T Siemens Avanto scanner. BH data was acquired with one non-saturated image and 10 TS images and FB data with 7-10 non-sat images separated by >5 seconds and 27-30 TS images for a total acquisition time of ~90 seconds. HC T1 maps were calculated using 50% of images selected automatically using an image based algorithm and registered using ANTs  with both the difference (Fig. 1) and primary images. NAV T1 maps were also calculated using images from these sets within a ± 3 mm window of a respiratory navigator. All T1 maps were calculated using a 2-parameter model, and the mean and coefficients of variation (COV) of myocardial T1 was determined. NAV and HC T1 maps were ranked by a blinded observer for myocardial border sharpness. Means and COVs were compared between the 3 techniques using repeated measures ANOVA with Tukey correction.
T1 maps using BH, NAV, and HC are shown in Fig. 2. HC maps were ranked as sharper than NAV maps in 13 of 14 cases. Mean myocardial T1 with BH, NAV, and HC were 1152 ± 29 ms, 1159 ± 28 ms, and 1147 ± 28 ms respectively (p = NS). T1 COV was significantly lower for HC (4.2 ± 0.8%) compared with both NAV (4.9 ± 1.1%) or BH (5.9 ± 1.1%) (p < 0.05). T1 maps with BH, NAV, and HC used 11, 19 ± 6, and 22 ± 1 images respectively.
A HC "key-hole" image can be acquired with <50 ms of additional data and used to improve image-registration. High-contrast SASHA is a robust approach to free-breathing T1 mapping, with HC T1 maps scoring sharper than navigator maps in 93% of cases and having 29% lower variability.
Chow K, et al: JCMR. 2014, 16 (Suppl 1): P29-
Avants BB, et al: Neuroimage. 2011, 54: 2033-
About this article
Cite this article
Chow, K., Yang, Y. & Salerno, M. Robust free-breathing SASHA T1 mapping using high-contrast image-based registration. J Cardiovasc Magn Reson 18 (Suppl 1), P28 (2016). https://doi.org/10.1186/1532-429X-18-S1-P28