Reducing Noise Induced by Cardiac Pulsatility in Brain Maps of R ₂ * and Magnetic Susceptibility Using Tailored k‐space Sampling

Figures

• 

  FIGURE 1. Cardiac noise and the proposed k‐space trajectory. (A) Distribution of the amplitude of the modelled cardiac‐induced noise Scardiac along the two phase‐encoding directions (ky,kz) of a 3D GRE sequence [27]. (B) Effective noise level (Nk,n) and averaging efficiency (Ek,n) as a function of the number of averages n. The optimal number of averages N opti at a given k‐space location is reached when the averaging efficiency (Ek,n) equals that of a point at the k‐space periphery dominated by thermal noise. (C) Target k‐space distribution of N opti . (D) Graphical representation of the CASPR trajectory in 3D k‐space. For each spiral arm, the dark/light data points are acquired on the Cartesian grid in the outward/inward direction. (E) First three spiral arms of the optimal CASPR trajectory. The white/black data points are acquired on the Cartesian grid in the outward/inward direction. (F) Distribution of N opti achieved with the optimal CASPR trajectory. The k‐space corners (in black) are acquired linearly at the end of the CASPR acquisition.

• 

  FIGURE 2. Proposed cardiac‐triggered sampling. (A) Global signal instability induced by cardiac pulsation. (B) Phase of the cardiac cycle at the time of acquisition of the data with the proposed cardiac‐triggered linear trajectory. The fast (kz)/slow (ky) phase‐encoding directions of the 3D GRE sequences are displayed horizontally/vertically. K‐space sampling was started from the top‐left corner. The red cardiac triggering lines enforce acquisition of the k‐space centre data during the most stable part of the cardiac cycle (0<φc<π2). (C) Pulse sequence diagram of the cardiac triggered sampling. At kz=k1 or kz=k2, analogue‐to‐digital conversion (ADC) is disabled until the detection of the next peak in the pulse oximeter signal. (E) Cardiac‐induced noise level as a function of the position of the two cardiac triggering lines k1 and k2.

• 

  FIGURE 3. (A) Example maps of the mean and standard deviation (SD) of R 2* across repetitions, computed from data acquired with the standard linear trajectory, CASPR trajectory and cardiac triggered sampling. To improve visibility, the presented maps were masked using a whole‐brain mask including grey and white matter voxels (see Section 2.3) and overlaid on the MPRAGE image. (B) Regional estimates of the SD of R 2* across repetitions in the brainstem, cerebellum and whole brain, across all 10 healthy volunteers.

• 

  FIGURE 4. (A) Example maps of the mean and standard deviation (SD) of RMSE across repetitions, computed from data acquired with the standard linear trajectory, CASPR trajectory and cardiac triggered sampling. To improve visibility, the presented maps were masked using a whole‐brain mask including grey and white matter voxels (see Section 2.4.2) and overlaid on the MPRAGE image. (B) Regional estimates of the mean RMSE in the brainstem, cerebellum and whole brain, across all 10 healthy volunteers. (C) Regional estimates of the SD of RMSE in the brainstem, cerebellum and whole brain, across all 10 healthy volunteers.

• 

  FIGURE 5. (A) Example maps of the mean and standard deviation (SD) of χ across repetitions, computed from data acquired with the standard linear trajectory, CASPR trajectory and cardiac triggered sampling. To improve visibility, the presented maps were masked using a whole‐brain mask including grey and white matter voxels (see Section 2.4.2) and overlaid on the MPRAGE image. (B) Regional estimates of the SD of χ across repetitions in the brainstem, cerebellum and whole brain, across all 10 healthy volunteers.

• 

  FIGURE 6. Maps of R 2* and χ computed from higher resolution data acquired with the standard linear and CASPR trajectories. (A) R 2* map from one repetition and standard deviation (SD) of R 2* across repetitions. (B) χ map from one repetition and SD of χ across repetitions. Lower SDs can be observed throughout the brain for the CASPR trajectory. The blue and green arrows highlight coherent aliasing along the anterior–posterior direction, observable in the data from the standard linear trajectory but absent in the data from the CASPR trajectory. To improve visibility, the presented maps were masked using a whole‐brain mask including grey and white matter voxels (see Section 2.4.2) and overlaid on the MPRAGE image.