STEAM-SASHA: a novel approach for blood- and fat-suppressed native T1 measurement in the right ventricular myocardium

Figures

• 

  Fig. 1. Each repeat of the sequence provides 1 anchor image (without the saturation pulse) and five TS=300 ms images a acquired across 2 breath holds. The first of these 2 breath holds is used to acquire the preparation scans and is followed by a T1 recovery period before the anchor image is acquired. The STEAM EPI sequence requires 2 heartbeats per image b. The dimension of the region contributing signal to the image in the phase encode direction is reduced by making the first two RF pulses slice selective in the phase encode direction (as denoted by “(y)” above the pulses), therefore allowing a shorter EPI readout. The third RF pulse is slice selective in the through plane direction (“(z)”). Non-stimulated echoes (i.e. gradient echoes and stimulated anti-echoes) are removed using spoiler gradients (Gspoil) that run at the same delay (Tspoil) from the ECG R-wave in both heartbeats. This spoiler timing avoids bulk cardiac motion related signal loss due to movement between the gradients, but spoils signal from the moving blood. T1 weighting is provided by a saturation pulse with TS=300 ms. A schematic plot of the evolution of the magnitude of the longitudinal (Mz) available to generate…

• 

  Fig. 2. An example of the approximate locations of the 5 regions of interest used in the analysis. (1) left ventricle (LV), (2) septum, (3) inferior RV, (4) anterosuperior RV, (5) anterolateral RV overlaid on a STEAM-SASHA anchor image

• 

  Fig. 3. Example averaged anchor (a) and TS = 300 ms (b) magnitude images from STEAM-SASHA. T1 maps are shown for STEAM-SASHA (c) and MOLLI (d). STEAM-SASHA data were acquired with 2 sequence repeats/4 breath holds MOLLI data with 2 sequence repeats/2 breath holds

• 

  Fig. 4. A plot of the error in T1 obtained with MOLLI and STEAM-SASHA in the T1MES phantom. Reference values from [14] acquired using a slow inversion recovery sequence were taken as ground truth. A positive value represents an over-estimation of T1 using the method shown

• 

  Fig. 5. Example STEAM-SASHA images acquired in 3 example subjects with poor, typical and high image quality. Both anchor data (8 averages) and the TS = 300 ms data (40 averages) are shown. The TS = 300 ms data are scaled × 5 in brightness relative to the anchor image data and signal outside the heart has been nulled. Figures S2–S4 show STEAM-SASHA images without averaging

• 

  Fig. 6. Example in vivo T1 maps calculated from both MOLLI and STEAM-SASHA data (a). To show the range in quality of STEAM-SASHA T1 maps, poor, typical and high-quality examples (based on the STEAM-SASHA maps) are shown. As blood is suppressed in STEAM-SASHA, pixels outside the myocardium are nulled in the STEAM-SASHA T1 map. Native T1 obtained with MOLLI and STEAM-SASHA for both scans of each subject are plotted for each of the five regions (b). Box plots in b show the range and quartiles and reference 3T SASHA values from the literature are shown as green bars [16]. Note the difference in colour scale between the T1 maps shown here and those shown in Fig. 3 for the phantom. Figure S6 shows equivalent Bland–Altman plots for the data in B

• 

  Fig. 7. The reproducibility (inter-study, intra-observer, inter-observer, measured as coefficient of variation) for each of the five myocardial regions assessed using both MOLLI and STEAM-SASHA

• 

  Fig. 8. The signal level in a septal region of interest for the noise-only, TS = 300 ms and anchor images in all 20 healthy volunteer scans. While for the TS = 300 ms and anchor images, the mean signal within the region of interest was used, the intensity at the maximum of a Rayleigh distribution fitted to a histogram of the signal intensities within the region of interest was used for the noise measurement

• 

  Fig. 9. The signal level in the RV blood pool from all 20 healthy volunteer scans from the anchor, TS = 300 ms, and noise-only images. The signal level was calculated by fitting a Rayleigh distribution to the histogram of the signal intensities within the region of interest values and then using the signal level at the maximum of this fitted distribution