Reduction of Scan Time and Echo Time in 3 D Time-of-Flight ( TOF ) MR Angiography , Using Reverse Elliptical Centric Dual-Echo Acquisition

In the conventional single-echo three dimension (3D) time-of-flight (TOF) magnetic resonance angiography (MRA), it always takes long scan time to obtain the good images. In addition, high-resolution MRA with flow compensation requires long echo time (TE). Here, a reverse elliptical centric dual-echo 3D TOF MRA p ulse sequence was developed to reduce the scan time and minimize the effective TE. Dual-echo and reverse elliptical centric acquisition in the phase encoding and slice phase encoding direction were used to reduce the scan time by 61%. Furthermore, the effective echo time (TE) was also minimized from 4.8ms to 4 .1ms. T he developed pulse sequence was tested on clinical MRI systems and the contrast to noise ratio (CNR) was measured to evaluate the results. The results show that the CNR only decrease by 2.5% with the scan time reduced to 39% when compared with the conventional single -echo 3D TOF. Compared with the conventional dual-echo 3D TOF acquisition, the developed pulse sequence brings advantages in three aspects: (1) reduce scan time by 22%; (2) increase CNR by 10%; (3) reduce the artifacts induced by dual-echo effects. In addition, while keeping the similar scan time as the conventional single-echo 3D TOF acquisition, the results show that reverse elliptical centric dual-echo 3D TOF acquisition increase CNR by 54.3%.


Introduction
In magnetic resonance imaging (MRI), the image degradation due to the signal loss remains a significant problem.And TOF M RA is sensitive to signal loss in two basic mechanisms: (1) saturation of flo wing spins (2) intra-vo xel spin phase dispersion.Here we mainly discuss about the second mechanis m.In TOF M RA, it can be caused by two factors: (1) the inho mogeneity of the main magnetic field, the influence of eddy current and magnetic susceptibility; (2) the comp lex flow of blood, such as acceleration flo w, pulsation 1 .Previous studies have indicated that the signal loss caused by the intra-voxel spin phase dispersion is greatly imp roved in short echo t ime (TE).Thus, on the previous basis using partial RF pulse 2 , the maximu m amp litude of the flo w co mpensation gradients, as well as partial echo acquisition technology 34 and so on, the novel design for 3D TOF in this art icle reduce the TE further to imp rove the blood vessel imaging.
In the conventional 3D TOF single -echo imaging, the scan time often requires more than 10 minutes due to the phase encoding in two direct ions and the use of mu ltip le overlapped thin slab (mult i-slab) 5 .Therefore, it's difficult for the patients to tolerate it, wh ich will absolutely affect the final imag ing.To solve this, someone proposed the conventional dual-echo 3D TOF 6 to reduce the scan time by 50% when co mpared with the conventional single-echo 3D TOF.In this paper, a reverse elliptical centric dual-echo 3D TOF pulse sequence is developed, and the scan time can be decreased to 39% o f the conventional single-echo 3D TOF.

First order flow-compensation
For the 3D TOF pulse sequence, the first order flow co mpensation is applied on the readout and slice phase encoding directions to null the first-order mo ment at the peak of the echo.
Figure 1 shows a first order flow-co mpensation diagram.‫ܩ‬ ௫ and ܶ are the maximu m grad ient amp litude and the corresponding shortest ramp time, respectively.The widths of the flowcompensation gradient pairs along ݇ ௭ direction are derived as: 78 where A are the net encoding areas along the phase and slice directions, respectively, and they must be incremented with constant steps with the differences being between two consecutive steps proportional to Δk .β is the time between the end of the second flow co mpensation pulse with W ଶ .W , T ୰ , and G are the widths of the plateau, the ramp t ime, and the amp litude of the plateau, respectively.M and M ଵ are the zeroth and first mo ments of the slab-selection gradient.

Revers elliptical centric dual-ec ho view order
To realize this, each of the ൫N ୷ × N ൯ views in the k ୷ − k plane is init ially sorted according to their distance from the center of k-space.The views that are outside of the ellipse:  2) By collecting the views close to the center of k-space centrally, the art ifacts can be reduced.Because 3D TOF acquisitions are commonly phase encoded in two orthogonal directions, there is considerable flexib ility in the k ୷ − k view order.Using the reverse elliptical centric v iew order, the conspicuous ghosts can be reduced by evenly spreading the ghosts in two dimensions (phase and slice) 10 .(3) Co mb ining the dual echo with reverse elliptical centric view order techniques , the scan time can be reduced to about 39% of the conventional single-echo 3D TOF.In addition, owing to the decrease of the slice phase encoding area in the first echo, from the calculat ion equations ( 1) and ( 2) in the flow co mpensation part above, the effective TE can be reduced.The degree of reduction is determined by the maximu m gradient and corresponding ramp t ime used in the design and the number of phase encodings in the slice direction.

MRI
To demonstrate the potential advantages of reverse elliptical dual-echo 3D TOF acquisition, an init ial comparison was made with the conventional single-echo 3D TOF and the conventional dual-echo 3D TOF acquisition, respectively.Figure 3 shows the reverse elliptical centric dual-echo 3D TOF sequence.The acquisition of a second echo with the ra phase encoding and slice encoding is added to a conventional mu ltiple overlapped thin slab acquisition of a 3D TOF pulse sequence.Both the view orders of the t wo echoes are rearranged to cover the elliptical k ୷ − k space.
Five healthy volunteers were scanned on a 1.5T MR scanner (Un ited Imaging Healthcare, Shanghai, China) using a 16-channel phased array head coil, the maximal gradient field adopted here was the 15.5mT/ m, and the corresponding ramp time was 270us.For the three pulse sequences, an asymmetric RF pulse and 71% of part ial echoes acquisition were used to reduce TE.The effective TE of conventional single-echo 3D TOF sequence and conventional dual-echo 3D TOF were both limited to 4.8ms, but was reduced to 4.1ms in the reverse elliptical centric dual-echo 3D TOF.In addit ion, the second TE in the conventional dual-echo 3D TOF and reverse elliptical centric dual-echo 3D TOF acquisition were 11.8ms and 11.1ms, respectively.The other parameters for these scans were same, i.e., 22cm field-of-view (FOV), 256* 256 matrix, 1mm slice thickness, 25 ° flip angle, 26ms repetition time (TR), ±30kHZ bandwidth (BW).To reduce the inflow saturation 11 , four slabs were acquired with 24 slices per slab and a 25% overlap between two slabs.In order to avoid the signal wrap -around along the slice-selection d irect ion, the imaged slab width was set as 28mm corresponding to 24mm wide for the excited slab.
Images reconstruction and post-processing were performed by using MATLA B. In order to reduce the partial volu me effect 12 , zero-filling interpolation 13 was applied to k-space data to reconstruct 3D complex images with a total 256*256*156 matrix.Then CNR was measured on the final M RA images, including different anatomic locations.To quantify the CNR, the method developed by Du et al and refined by Chapman et al 14 , was used.

Results
Figure 4 shows the projections of the MRA images acquired by the conventional single-echo 3D TOF acquisition in the left co lu mn and the reverse elliptical centric dual-echo 3D TOF acquisition in the right colu mn, respectively.The three rows correspond to the MIP (maximu m-intensity projection) of MRA in the transverse, sagittal and coronal plane.The CNR of the four vessels indicated by the arrows in the Fig. 4 was measured to evaluate the performance.To distinguish them, the four blood  Fro m analysis of the results, it can be found that for the big vessels like vessel one, two, the CNR of reverse elliptical centric dual-echo 3D TOF M RA is a little higher than that in the conventional single-echo 3D TOF MRA.And the CNR for small vessels, such as vessel three, four, decrease a little because of the long echo t ime of the acquisitions for h igh frequency parts .Furthermore, the blurring and ghosting due to off-resonance effects that can be predicted theoretically by point spread function (PSF) analysis 15 may also degrade the CNR.
In conclusion, compared with the conventional single-echo 3D TOF, only 39 percentage of the scan time is taken, but the MRA with similar CNRs is obtained in the reverse elliptical centric dualecho 3D TOF pulse sequence.Fig. 6 shows the MRA data acquired by the reverse elliptical centric dual -echo 3D TOF (the left column) and the conventional dual-echo 3D TOF (the right colu mn) acquisition, respectively.The top row shows the MIP of MRA, and the bottom row shows the specific slice of the same patient.In the bottom row, the art ifacts indicated by the arrow in Fig. 6 (d ) were reduced much in the reverse elliptical centric dual-echo 3D TOF.The artifacts due to the off-resonance effects were predicted theoretically by PSF analysis 15 .Beyond that, the local vess els CNR of the four vessels were measured to compare in Fig. 7 (a) and (b).Hence, co mpared with the conventional dual-echo 3D TOF MRA, the current method not only reduce the scan time by 22%, but also decrease the artifacts and improve the CNR of images.
To further co mpare the reverse elliptical centric dual-echo 3D TOF and the conventional singleecho 3D TOF pulse sequence, And the CNR measurements were made in the Fig. 9.For the big vessels like "1" and "2", the CNRs of reverse elliptical centric dual-echo 3D TOF MRA with averaging are much higher.In addition, the CNRs of s mall vessels such as "3" and "4", are also higher.

Discussion
The applications of MRA always confront two problems.(1) Due to the long scan time, it may bring the motion art ifacts and also occupy the time required for other scans, affecting the effective diagnosis of the patient's condition.(2) In vascular M RI, the phase dispersion caused by complex flow might lead to the signal loss, which can be avoided by reducing TE.The developed pulse sequence can resolve the two problems by reducing the scan time to about 39% of the conventional method and decreasing the effective TE fro m 4.8ms to 4.1ms.I shows the mean and peak CNR of the MRA of the vessels averaged fro m the five volunteers acquired by different acquisition methods.Co mpared with the conventional single-echo 3D TOF, the mean and peak CNR o f reverse elliptical centric dual-echo 3D TOF M RA are decreased by 2.5% and 2%, respectively.Co mpared with the conventional dual-echo 3D TOF, the mean and peak CNR of reverse elliptical centric dual-echo 3D TOF MRA are increased by 10.9% and 8.9%, respectively.In addition, the comparison of CNR between the convent ional single-echo TOF MRA and the reverse elliptical centric dual-echo TOF MRA with t wo-average shows that the mean CNR and peak CNR are increased by 54.3% and 56.4%, respectively.In conclusion, the results reveal the CNR advantage of reverse ellipt ical centric dual-echo 3D TOF pulse sequence.

Figure 1 .
Figure 1.Gradient diagram of the flow-compensations along the slice phase-encoding directions for 3D TOF imaging.It starts from the center of RF pulse and ends in the echo peak.
standard k ୷ − k rectangle are skipped.As shown in the Fig.2, the open dots in the four corners are the omitted views, wh ich are not acquired and substituted with zeros during reconstruction.And in the previous work, it has been shown that the elliptical v iews can reduce the scan time of 3D acquisition by 21% with a slight penalty on the image quality9 .Next, the dots located in the ellipse are further div ided into two equal parts according to the d istance fro m the o rig in of k-space.The red dots are the views close to the center of k-space and are acquired in the first echo with a relatively short echo time (TE), determining the signal-noise ratio (SNR) and contrast-noise ratio (CNR) o f the images.The black dots are the views far away fro m the origin of k-space and are acquired in the second echo with a long TE.For dual-echo, two echoes are acquired within one repetition time (TR).The SNR and CNR of images are mainly determined b y the first echo, which is also called the effective TE.Starting fro m their respective edge, both red dots and black dots with the views of the decreasing distance from the orig in of k-space are acquired subsequently.Such design has brought some advantages.(1) Starting fro m the periphery of k-space views, it can reduce the use of du mmy scans (i.e., scan with the data acquisition disabled), wh ich are used to reach a steady state8 .(

Figure 2 .
Figure 2. The views order in ݇ ௬ − ݇ ௭ space for reverse elliptical centric dual-echo 3D TOF acquisition.The data for the views in the corners are not acquired and set to zeros during image reconstruction.The remaining views (the solid black dots and red dots) are further divided into two equal parts according to the distance from the origin of k-space.The solid red dots and the black dots are acquired in the first and second echo, respectively.

Figure 3 .
Figure 3. Diagram of the reverse elliptical centric dual-echo 3D TOF pulse sequence.To achieve the short TE, the asymmetric RF pulse, partial-echo acquisitions and flow-compensated gradients with minimum duration are applied in the sequence.In addition, by reducing the slice phase-encoding areas of the first echo, the effective TE ‫ܧܶ(‬ ଵ ) is shortened.A fly-back gradient is located in the middle of two echoes to restore flow compensation in the second echo.And the gradients in phase-encoding and slice phase-encoding directions are rearranged to accomplish the reverse elliptical centric dual-echo view order.The red gradient means that the gradients will vary according to the specific view order.

Figure 4 .Figure 5 .
Figure 4.The M RA images acquired by reverse elliptical centric dual-echo 3D TOF (column a) and conventional single-echo 3D TOF (column b) acquisition.From top to bottom, the M IP of M RA in the transverse, sagittal, coronal plane were shown to observe the difference.After reducing t he scan time by 61% and the effective TE about 0.7ms, the M RA image of reverse elliptical centric dual-echo 3D TOF method maintains the similar vascular visualization with the conventional single-echo 3D TOF method.

Fig. 5 (
Fig.5 (a) shows the representative local vessel CNRs of the big vessel "1".The solid points designate the mean local vessel CNR and the asterisks designate the peak local ves sel CNR.The histogram in Fig.5 (b) shows the mean CNR and peak CNR of the four vessel segments averaged from five volunteers.Fro m analysis of the results, it can be found that for the big vessels like vessel one, two, the CNR of reverse elliptical centric dual-echo 3D TOF M RA is a little higher than that in the conventional single-echo 3D TOF MRA.And the CNR for small vessels, such as vessel three, four, decrease a little because of the long echo t ime of the acquisitions for h igh frequency parts .

Figure 6 .Figure 7 .
The M RA images acquired by reverse elliptical centric dual-echo 3D TOF (a) and conventional dualecho 3D TOF (b) acquisitions (c) and (d) are one of the images acquired by reverse elliptical centric dual-echo 3D TOF pulse sequence and conventional dual-echo 3D TOF pulse sequence, respectively.The mean CNR and peak CNR of the points along the vessel labeled "1" in conventional dual-echo and reverse elliptical centric dual-echo 3D TOF acquisition are shown in (a).A histogram shown in (b) is used to graphically summarize and display the mean CNR and peak CNR of four vessels averaged from five volunteers.

Figure 9 .
Fig 8.(a) and Fig 8.(b) show the MRA data acquired by reverse elliptical centric dual-echo 3D TOF pulse sequence after averaging and conventional single -echo 3D TOF pulse sequence, respectively.(a) (b) Figure 8.The M RA images acquired by reverse elliptical centric dual-echo 3D TOF after averaging (a) and conventional single-echo 3D TOF (b) acquisitions.The mean CNR and peak CNR of the points along the big vessel labeled "2" in conventional singleecho and reverse elliptical centric dual-echo 3D TOF with two-average acquisition are shown in (a).A histogram shown in (b) is used to graphically summarize and display the mean CNR and peak CNR of four vessels averaged from five volunteers.

Table 1 .
M ean and peak CNR of the vessels