Flow-suppressed 2D spin-echo imaging with high tolerance to B1 inhomogeneity using hyperbolic secant pulses

Abstract

Purpose: To demonstrate flow-suppressed two-dimensional (2D) spin-echo and spin-echo diffusion echo-planar imaging (EPI) sequences using hyperbolic secant (HS) pulses for both π/2 excitation and π refocusing.

Theory and methods: A theoretical framework to derive phase dispersion of moving spins under π/2 excitation and π refocusing using HS pulses was described. Numerical simulations were performed to verify the validity of the theoretical analysis. All experiments were performed on a 3T clinical scanner. Phantom and human-brain imaging was performed using 2D spin-echo sequence, and liver imaging was performed using 2D spin-echo diffusion EPI. The signal-to-noise ratio and residual blood flow signal of the proposed sequences were compared with those of conventional spin-echo sequences using sinc pulses.

Results: Results from human brain and liver images demonstrated that the proposed method substantially reduced blood flow artifacts. In the brain, venous blood flow was suppressed more effectively with the proposed method than with conventional spin-echo sequence using presaturation. In the liver, as compared with spin-echo sequence using sinc pulses, the proposed method showed noticeable attenuation of bright blood signals at low b-values, whereas the overall tissue signal in peripheral regions was higher. The signal-to-noise ratio was enhanced by 10% to 30%, indicating improved B₁ tolerance due to the adiabatic π refocusing HS pulse.

Conclusion: Flow suppression and partial B₁ insensitivity were achieved by replacing sinc pulses with HS pulses in conventional 2D spin-echo imaging and spin-echo diffusion EPI sequences. This approach may be particularly useful in various applications requiring reduced vascular signal contamination, such as liver and brain imaging.

Keywords: B1 insensitivity; DWI; flow suppression; hyperbolic secant pulse; spin‐echo MRI.

FIGURE 1

The proposed two‐dimensional spin‐echo sequence diagram using hyperbolic secant (HS) pulses for both π/2 excitation and π refocusing, satisfying Condition II to compensate for nonlinear phases across the slice (i.e., β ₁ = β ₂, T _p,1 = T _p,2, BW ₁ = 2BW ₂, G ₁ = 2G ₂). Only slice‐selection gradients are illustrated, along with the amplitude modulation (AM) and frequency modulation (FM) functions of the HS pulses, to highlight the applied condition (Condition II). The slice‐selection gradient for the π refocusing HS pulse is combined with crusher gradients. This sequence can provide high B₁ insensitivity due to the adiabatic property of the π refocusing HS pulse. BW, bandwidth; TE, echo time.

FIGURE 2

Phase distributions generated by π/2 excitation (red) and π refocusing (blue) hyperbolic secant (HS) pulses, as well as the total phase distributions at echo time (black), were calculated by numerical simulation and results were compared with the theoretical description for both static and moving spins. To aid understanding, Eqs. (9) and (10) was vertically shifted to match the phase range of Eqs. (23) and (24). For all cases, a slice thickness of 5 mm and echo time of 22.36 ms was used. HS pulses with β ₁ = β ₂ = 5.3, T _p,1 = T _p,2 = 5.12 ms, BW ₁/2π = 2BW ₂/2π = 4.14 kHz were applied for π/2 excitation and π refocusing, respectively. (A) For static spins, quadratic‐like phase distributions with opposite polarities were produced by the π/2 excitation and π refocusing HS pulses, and these phase profiles were fully compensated at echo. (B) For moving spins, a nearly linear phase distribution was formed at echo (black dotted line) due to the shifted quadratic‐like phase distributions with opposite polarities, which validates the theoretical description of flow suppression (black solid line). BW, bandwidth.

FIGURE 3

The results of the numerical simulations were illustrated. (A) Signal intensity as a function of flow velocity at echo time when sinc and hyperbolic secant (HS) pulses are used for both π/2 excitation and π refocusing. HS pulses with β ₁ = β ₂ = 5.3, T _p,1 = T _p,2 = 5.12 ms, and BW ₁/2π = 2BW ₂/2π = 4.14 kHz were used for π/2 excitation and π refocusing HS pulses, respectively. Scan parameters were as follows: echo time = 22.36 ms and slice thickness = 5 mm. For sinc pulses, signal intensity decreased linearly with increasing velocity. For HS pulses, signal intensity decreased more rapidly than that obtained using sinc pulses, attributed to phase dispersion caused by shifted quadratic‐like phase distributions with opposite polarities (red dotted line). The flow velocity that minimizes the signal intensity (i.e., 19.77 cm/s) was close to the result of the theoretical description (i.e., 22.36 cm/s). (B) Signal intensity at a flow velocity of 10 cm/s as a function of π refocusing HS pulse bandwidth and duration, when HS pulses are used for both π/2 excitation and π refocusing. Signal intensity decreased with increasing π refocusing HS pulse duration and bandwidth.

FIGURE 4

The proposed two‐dimensional spin‐echo echo‐planar imaging (EPI) sequence diagram using hyperbolic secant (HS) pulses for both π/2 excitation and π refocusing, satisfying Condition II to compensate for nonlinear phases across the slice (i.e., β ₁ = β ₂, T _p,1 = T _p,2, BW ₁ = 2BW ₂, G ₁ = 2G ₂). A bipolar diffusion‐weighted gradient, representing a basic form of motion‐compensated diffusion‐weighted gradients, was used for simplicity (cyan). Only slice‐selection‐direction diffusion‐encoding gradients are illustrated to highlight the applied condition (Condition II). The sequence was robust to B₀ inhomogeneity effects due to relatively high radiofrequency pulse bandwidth and exhibits high B₁ insensitivity due to the adiabatic property of π refocusing HS pulse. AM, amplitude modulation; BW, bandwidth; FM, frequency modulation; TE, echo time.

FIGURE 5

Two‐dimensional spin‐echo phantom images acquired using a head coil at 3 T. (A) The image obtained using 5‐lobe sinc pulses with T _p  = 3.07 ms. (B) The image obtained using hyperbolic secant (HS) pulses. HS pulses with β ₁ = β ₂ = 5.3, T _p,1 = T _p,2 = 5.12 ms, BW ₁/2π = 2BW ₂/2π = 4.14 kHz were used for both π/2 excitation and π refocusing. To fully leverage B₁ insensitivity, the π refocusing HS pulse was applied with nearly 3‐dB higher power than the adiabatic threshold. For both images, the scan parameters were as follows: repetition time = 500 ms, echo time = 27 ms, field of view = 216 mm × 216 mm, matrix size = 192 × 192, and slice thickness = 5 mm. (C) One‐dimensional profiles along the white line in (A) and (B). Signal‐to‐noise ratio (SNR) decreases more slowly near the periphery of the phantom image acquired with HS pulses, whereas the SNR at the center remains nearly the same in both cases, reflecting the adiabatic property of the π refocusing HS pulse. (D) Inversion profiles (M _z /M ₀ ) obtained through Bloch simulation, shown as a function of offset frequency and peak radiofrequency amplitude. The same parameters as those used for π refocusing HS pulse in the phantom imaging were applied in the simulation. Spins within the pulse bandwidth experience uniform π rotation when the adiabatic condition is met.

FIGURE 6

Two‐dimensional spin‐echo human brain images acquired using a head coil at 3 T. (A) Axial images obtained using 5‐lobe sinc pulses without/with presaturation and hyperbolic secant (HS) pulses without presaturation. Eight axial slices were acquired with a 20% slice gap (= 1 mm), and the third and fourth slice covering superior sagittal sinus (SSS) was chosen to compare the performance of the sequence. (B) Coronal images with the same parameters. Eight coronal slices were acquired with a 20% slice gap (= 1 mm), and the fourth and fifth slice covering internal cerebral veins (ICVs) was chosen to compare the performance of the sequence. HS pulses with β ₁ = β ₂ = 5.3, T _p,1 = T _p,2 = 5.12 ms, and BW ₁/2π = 2BW ₂/2π = 4.14 kHz were used for π/2 excitation and π refocusing. The numbers shown in the upper row of (A) and (B) represent the signal‐to‐noise ratio (SNR) in the white matter and gray matter (white boxes in the brain). SNRs were also measured within the SSS and ICVs to compare the efficiency of venous blood flow suppression. Background noise was calculated from the white box positioned in the upper left corner of the image. The image acquired with HS pulses demonstrates higher SNR, attributed to the adiabatic property of the π refocusing HS pulse. In the image acquired with sinc pulses, flow artifacts were prominent, particularly in the superior sagittal sinus, whereas signals from moving blood were effectively suppressed in the image acquired with HS pulses (yellow arrow).

FIGURE 7

Two‐dimensional spin‐echo diffusion echo‐planar imaging (EPI) human liver images acquired using a body coil at 3 T. A bipolar motion‐compensated diffusion‐weighted gradient was used for diffusion encoding. (A) b = 50, 400, and 800 s/mm² axial diffusion‐weighted (DW) images and apparent diffusion coefficient (ADC) maps obtained using 5‐lobe sinc pulses and hyperbolic secant (HS) pulses. (B) Axial DW images and ADC map of different slice. HS pulses with β ₁ = β ₂ = 5.3, T _p,1 = T _p,2 = 5.12 ms, and BW ₁/2π = 2BW ₂/2π = 6.13 kHz were used for π/2 excitation and π refocusing. The numbers shown in (A) and (B) represent the signal‐to‐noise ratio (SNR) in the right and left lobes of the liver (white boxes in the liver). Background noise was calculated from the white box positioned in the lower left or right corner of the image. Owing to the adiabatic π refocusing HS pulse, the image acquired with HS pulses shows high SNR in the peripheral area of the liver, with better definition of abdominal structures at the periphery. Additionally, chemical shift artifacts, which were prominent along the phase‐encoding direction in (A) (magenta arrow), were suppressed well with HS pulses, attributed to the fat suppression effect of Condition II. In the image acquired with sinc pulses, bright blood signals were prominent, particularly in the abdominal aorta and veins, including the branches of the right portal vein and the right hepatic vein. In contrast, these signals were effectively suppressed in the image acquired with HS pulses (yellow arrows).