Multitimepoint velocity-selective reconciled spatially -selective current arterial spin labeling MRI mapping cerebral blood flow and arterial transit time

Abstract

Multi-timepoint velocity-selective reconciled with spatially-selective (MULTIVERSE) is used for extending current arterial spin labeling MRI for mapping cerebral blood flow (CBF) and arterial transit time (ATT) at a longer ATT. MULTIVERSE arterial spin labeling (ASL) utilizes multi-time point pseudo-continuous (PC) ASL and multi-time point velocity-selective (VS) ASL with spatially defined bolus and applies combined fitting to measure CBF and ATT. The methods of the present invention extend the capability of existing multi-timepoint arterial ASL methods in estimating cerebral blood flow CBF and ATT with a wider range of ATT.

Description

MULTI-TIMEPOINT VELOCITY-SELECTIVE RECONCILED WITH SPATIALLY-SELECTIVE (MULTIVERSE): EXTENDING CURRENT ARTERIAL SPIN LABELING MRI FOR MAPPING CEREBRAL BLOOD FLOW AND ARTERIAL TRANSIT TIME (ATT) AT LONGER ATT GOVERNMENT SUPPORT

[0001] The present invention was made with government support under grant numbers EB031771, and HL144751 awarded by the National Institutes of Health. The government has certain rights in the present invention. CROSS REFERENCE TO RELATED APPLICATIONS

[0002] This application claims the benefit of U.S. Provisional Patent Application No. 63 / 642,343 filed on May 3, 2024, which is incorporated by reference, herein, in its entirety. FIELD OF THE INVENTION

[0003] The present invention relates generally to medical imaging. More particularly, the present invention relates to systems and methods for multi-timepoint velocity-selective reconciled with spatially-selective (MULTIVERSE) for extending current arterial spin labeling MRI for mapping cerebral blood flow and arterial transit time (ATT) at a longer ATT. BACKGROUND OF THE INVENTION

[0004] Cerebral blood flow (CBF) is a fundamental hemodynamic parameter that characterizes brain perfusion as a marker for both neural function through neurovascular coupling and various cerebrovascular diseases. Spatially selective arterial spin labeling (ASL), such as pulsed ASL (PASL) and pseudo-continuous ASL (PCASL), inverts themagnetizations of incoming blood water through internal carotid arteries (ICA) and vertebral arteries (VA) and applies a post labeling delay (PLD) to allow tagged blood to arrive at brain tissue where image is acquired. Although PCASL with a single PLD of 1800-2000 ms was recommended a decade ago for its ease of use and adequate SNR for clinical perfusion mapping, it has long been recognized that CBF would be underestimated when the prescribed PLD is shorter than the arterial transit time (ATT) from the PCASL labeling plane to the imaging voxel and hence the labeled bolus is not fully captured. Furthermore, if ATT surpasses the sum of the labeling duration (LD) and PLD, PCASL would not produce any ASL signal. In this scenario, the imaging voxel would not yet have received any labeled bolus.

[0005] Velocity-selective ASL (VSASL) was proposed to mitigate the susceptibility to slow flow by labeling the upstream blood in the vascular tree flowing above a small cutoff velocity (Vcut) corresponding to the small arterioles, such that ATT from the leading edge of VSASL labeling to capillary bed is close to zero. A single PLD of 1400 ms was recommended in VSASL. In contrast to PCASL, VSASL is prone to fast flow and would underestimate CBF when acquired at a PLD longer than the bolus duration, defined as the ATT from the trailing edge of VSASL labeling, which is typically affected by both the spatial coverage of the transmit RF coil and the velocity of the blood flow.

[0006] An alternative approach is to acquire multi-PLD / multi-timepoint PCASL and fit with kinetic models to estimate CBF and ATT simultaneously, as not only would CBF be more accurately quantified, but the ATT itself offers valuable hemodynamic information for characterizing vascular conditions. However, most sequential, time-encoded, or hybrid multi- timepoint PCASL protocols were optimized for ATT less than 1800 ms, which are typical for young healthy participants. Designing multi-timepoint ASL with extended PLDs to encompass the ATT range to 2000-2500 ms or beyond for elderly subjects or patients wouldlead to increased errors in CBF and ATT quantification due to reduced SNR caused by T1relaxation effect.

[0007] Combining velocity and spatially selective ASL suggested a promising avenue for various clinical populations. A recent technique termed (VESPA) was proposed by adding PCASL labeling during the PLD right after the VS labeling. Through an encoding / decoding process, VESPA extracted a VSASL (PLD = 1800 ms) as well as a PCASL (LD = 1800 ms, PLD = 0 ms) to fit CBF and ATT concurrently. Unfortunately, as elucidated above, this zero- PLD PCASL would not generate any ASL signal when ATT exceeds its LD, thereby imposing an ATT ceiling at 1800 ms for this method.

[0008] It would therefore be advantageous to develop an ASL method that provides a more accurate CBF and ATT estimation under a wide range of ATT. SUMMARY OF THE INVENTION

[0009] The foregoing needs are met by the present invention which provides a method for magnetic resonance imaging including a scan. The scan includes using a multi-time point pseudo-continuous arterial spin labeling (PCASL) and multi-time point velocity-selective (VSASL). The scan also includes applying combined fitting to measure cerebral blood flow (CBF). Additionally, the method includes applying combined fitting to measure arterial transit time (ATT).

[0010] In accordance with an aspect of the present invention, the multi-time-point VS includes a spatially defined bolus. The method includes using several inferior saturation pulses (ISP) following the VS module to crush the inflow of blood below the labeling plane of PCASL. The fitted parameter of ATT is the same between PCASL and VSASL. VSASL with ISP and PCASL are acquired separately, using the same multi-time point scheme. Themethod can include applying a pair of matched post labeling delays (PLDs) for PCASL and VSASL. The method can also include applying post labeling delays (PLDs) optimized for the scan. The method can include applying mismatched post labeling delays (PLDs). The method includes applying multiple mismatched PLDs. The method includes measuring CBF and ATT simultaneously.

[0011] In accordance with another aspect of the present invention, a system for magnetic resonance imaging of a subject includes a magnetic resonance imaging (MRI) machine. THE MRI machine is configured to execute a scan. The scan includes using a multi-time point pseudo-continuous arterial spin labeling (PCASL) and multi-time point velocity-selective (VSASL). The scan also includes applying combined fitting to measure cerebral blood flow (CBF). Additionally, the method includes applying combined fitting to measure arterial transit time (ATT).

[0012] In accordance with an aspect of the present invention, the multi-time-point VS includes a spatially defined bolus. The system includes using several inferior saturation pulses (ISP) following the VS module to crush the inflow of blood below the labeling plane of PCASL. The fitted parameter of ATT is the same between PCASL and VSASL. VSASL with ISP and PCASL are acquired separately, using the same multi-time point scheme. The system can include applying a pair of matched post labeling delays (PLDs) for PCASL and VSASL. The system can also include applying post labeling delays (PLDs) optimized for the scan. The system can include applying mismatched post labeling delays (PLDs). The system includes applying multiple mismatched PLDs. The method includes measuring CBF and ATT simultaneously.BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIGS.1A, 1B, and 1C illustrate schematic views of MULTIVERSE sequence diagrams include a pseudo-continuous arterial spin labeling (PCASL) sequence. FIG.1A illustrates a velocity selective arterial spin labeling (VSASL) sequence. FIGS.1B and 1C illustrate spatial selectivity of RF pulses. FIG.1D illustrates a schematic diagram of PLD: post-labeling delay. ISP: inferior saturation pulses (red). BS: background suppression (brown).

[0014] FIGS.1D, 1E, 1F, and 1G illustrate graphical views of pulse shapes of rest_2 pulses for different scenarios, according to an embodiment of the present invention.

[0015] FIGS.2A -2I illustrate graphical views of fitted CBF and ATT on the simulated kinetic curves of perfusion-weighted signal (PWS) using true ATT values as 600 ms (left), 1800 ms (middle), and 3000 ms (right).

[0016] FIGS.3A-3F illustrate graphical views of Monte Carlo simulations of accuracy according to an embodiment of the present invention.

[0017] FIGS.4A-4F illustrate histogram views of a normalized confidence interval (nCI) of fitted CBF and ATT from the Monte Carlo simulations for multi-PLD PCASL (dark grey), multi-PLD VSASL (light grey), and MULTIVERSE ASL (medium grey), all with PLD = [500, 1000, 1500, 2000, 2500] ms.

[0018] FIG.5 illustrate image views of PCASL, VSASL, and MULTIVERSE results from a 29-year-old healthy female.

[0019] FIG.6 illustrate image views of PCASL, VSASL, and MULTIVERSE results from a 63-year-old healthy female.

[0020] FIGS.7A-7D illustrate histogram views of normalized 95% nCI of fitted CBF and ATT from the two subjects shown in FIGS.5 and 6.

[0021] FIG.8 illustrates orthogonal views of individual 3D CBF and ATT mats from 9 subjects obtained by MULTIVERSE ASL.

[0022] FIGS.9A and 9B illustrate graphical views of averaged CBF and ATT for grey and white matter over 9 subjects.

[0023] FIGS.10A-10F illustrate graphical views of normalized root-mean-square-error (nRMSE) for estimating CBF, as illustrated in FIGS.10A-10C and ATT, as illustrated in FIGS.10D-10F at different noise levels.

[0024] FIG.11A and 11B illustrate image views of a comparison between VSASL PWS images with and without IS pulses on a 27-year-old female, as illustrated in FIG.11A, and a 32-year-old male, as illustrated in FIG.11B.

[0025] FIG.12A and 12B illustrate image views of a normalized 95% confidence interval (nCI) maps of CBF, as illustrated in FIG.12A and ATT, as illustrated in FIG.12B, from PCASL, VSASL, and MULTIVERSE ASL in the same subject as FIG.5.

[0026] FIGS.13A-13F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using a single matched PLD with using multiple matched PLDs.

[0027] FIGS.14A-14F illustrate Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square-error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using single mis-matched PLD with using multiple matched PLDs.

[0028] FIGS.15A-15F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using single mis-matched PLD with using multiple matched PLDs.

[0029] FIGS.16A-16F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using multiple mis-matched PLDs with using multiple matched PLDs.

[0030] FIGS.17A-17F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using multiple mis-matched PLDs with using multiple matched PLDs.

[0031] FIGS.18A-18D illustrate diagrammatic and graphical views of the arterial input function for the VSASL-labeled bolus under two conditions, according to embodiments of the present invention. DETAILED DESCRIPTION

[0032] The presently disclosed subject matter now will be described more fully hereinafter with reference to the accompanying Drawings, in which some, but not all embodiments of the inventions are shown. Like numbers refer to like elements throughout. The presently disclosed subject matter may be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will satisfy applicable legal requirements. Indeed, many modifications and other embodiments of the presently disclosed subject matter set forth herein will come to mind to one skilled in the art to which the presently disclosed subject matter pertains, having the benefit of the teachings presented in the foregoing descriptions and the associated Drawings. Therefore, it is to be understood that the presently disclosed subject matter is not to be limited to the specific embodiments disclosed and that modifications and other embodiments are intended to be included within the scope of the appended claims.

[0033] Multi-timepoint velocity-selective arterial spin labeling (ASL) reconciled with a spatially-selective bolus (MULTIVERSE) is used for extending current arterial spin labeling MRI for mapping cerebral blood flow and arterial transit time (ATT) at a longer ATT. MULTIVERSE ASL utilizes multi-time point pseudo-continuous (PC) ASL and multi-time point velocity-selective (VS) ASL with spatially defined bolus and applies combined fitting to measure CBF and ATT. The methods of the present invention extend the capability ofexisting multi-timepoint arterial spin labeling (ASL) methods in estimating cerebral blood flow (CBF) and arterial transit time (ATT) with a wider range of ATT.

[0034] MULTIVERSE ASL utilizes multi-time point pseudo-continuous (PC) ASL and multi-time point velocity-selective (VS) ASL with spatially defined bolus and applies combined fitting to measure CBF and ATT. The accuracy and precision of CBF and ATT estimation were compared among MULTIVERSE ASL, multi-time point PCASL, and VSASL in simulation. The 95% confidence intervals of the fit and the normalized standard error of the fit were compared for CBF and ATT among three ASL methods in healthy volunteers. Clinical application of MULTIVERSE was also attempted in patients with Moyamoya disease, sickle cell disease, and brain metastatic cancer.

[0035] With the same scan time, MULTIVERSE ASL improved the accuracy and precision and reduced uncertainty in CBF and ATT quantification across an extended range of ATT (500-4000ms) from the simulation. MULTIVERSE also demonstrated more reliable fitting results in CBF and ATT than multi-delay PCASL and VSASL respectively in healthy participants. Clinical data demonstrated the feasibility of MULTIVERSE in Moyamoya disease with pathological long ATT as well as in sickle cell disease with pathological short ATT. The preliminary comparison between MULTIVERSE and DSC MRI in brain metastatic cancer patients suggested a fair agreement between the two methods.

[0036] This novel approach improves perfusion measurement over the extended range of arterial transit time which was not possible with existing ASL methods. It highlights the clinical potential of ASL-based perfusion mapping in various altered physiological and pathological conditions.

[0037] To improve the measurement of CBF and ATT for a wide range of ATTs (500- 4000 ms) and to compensate for the respective ATT sensitivity for PCASL and VSASL, a more straightforward approach is proposed through separate acquisition of PCASL and VSASL with combined fitting, which is dubbed MULti-TImepoint VElocity-selective Reconciled with Spatially-sElective or “MULTIVERSE” ASL. First, numerical simulations were conducted to evaluate the accuracy and precision of single-delay and multidelay PCASL and VSASL, as well as the proposed MULTIVERSE ASL, in obtaining CBF and AT quantification across an extended range of ATTs. The CBF and ATT were compared fitting between multi-delay PCASL, VSASL, and MULTIVERSE ASL in healthy volunteers.

[0038] In MULTIVERSE ASL, PCASL and VSASL are acquired separately, with only minor modifications of standard VSASL by inserting several inferior saturation (IS) pulses following the VS module to crush the inflow of blood below the labeling plane of PCASL, as illustrated in FIGS.1A and 1C. Similar to the VESPA ASL method, MULTIVERSE ASL seeks to establish the ATT from the trailing edge of the VSASL bolus (bolus duration) equal to the ATT from the labeling plane of the PCASL bolus. The motivation for this design is to make the fitted ATT equal between PCASL and VSASL, so that there will be only two unknow parameters (CBF and ATT) to estimate with two sets of equations (Eqs. [1-3] for PCASL and Eq. [4-5] for VSASL) following the general kinetic model of ASL 20 for a single-compartment model:where SPCASLand SVSASLare the difference signals of PCASL and VSASL; is the labeling efficiency of PCASL and VSASL respectively and BS is the labeling efficiency factor due to the background suppression (BS) pulses; SIPD is the signal intensity (SI) of brain tissue at equilibrium from the proton density (PD) weighted images, is the brain- blood partition coefficient (0.9 mL / g) and the factor of 6000 converts the units for CBF from mL / g / s to mL / 100g / min; T1,eff is considered as blood T1as was standardized previously.

[0039] Here, MULTIVERSE used multiple (500-2500 ms) matched PLDs for both PCASLand VSASLin a straightforward design. The performance of this approach was evaluated through simulations as well as in vivo experiments that compared MULTIVERSE with both PCASLand VSASL.

[0040] Simulations were performed in MATLAB (Mathworks, Natick, MA) with PCASLand VSASLperfusion weighted signal (PWS, S / SIPD) generated by their kinetic functions (Eq. [1-3] for PCASL; Eq. [4-5] for VSASL) with five matched PLDs of [500, 1000, 1500, 2000, 2500] ms. PCASL: LD = 1800 ms, = 0.85; VSASLwith Fourier-transform (FT) based VS inversion (VSI): = 0.564;BS= 1.00; CBF = 60 mL / 100g / min; ATT was ranged from 0 msto 4000 ms; T1,eff= 1.85 s. Five methods were compared: single-PLD PCASL(PLD = 2000 ms), single-PLD VSASL(PLD = 1500 ms), five-PLD PCASL, five-PLD VSASL, and five-PLD MULTIVERSE ASL. For the 1000 repeats of each method, Gaussian white noise with a standard deviation (SD) of a specified percentage of SIPDwas added to the PWS at different PLDs. To simulate the data acquired with the same scan duration (product of the number of PLDs and the number of repeats) for each method, the SDs of the noise for single-PLD PCASLand VSASL were scaled by 5 in the five-PLD PCASL and VSASL, and by 10 in five-PLD MULTIVERSE ASL, respectively. The noise SD for the single-PLD ASL was set as 0.15% of SIPDby default and was also modified to evaluate the results at lower or higher noise levels.

[0041] Data generation, fitting and analysis were performed in MATLAB. CBF and ATT were fit using the general kinetic models (Eqs. [1-5]) with the nonlinear least square algorithm (lsqcurvefit). Note that derivation of Eqs.[1-5] assumes that, after saturation, the arterial blood would be completely replenished by the inflow of unsaturated fresh blood before the labeling module. In the condition of ATT greater than the saturation delay, a mixed bolus of saturated blood and fresh blood will be labeled by the VSASL labeling module. A complete model was derived to characterize the kinetic functions more accurately by incorporating the ATT effect during the saturation delay. In line with the recommended practice, CBF was calculated using Eq. [2] for single-PLD PCASL, assuming the PLD exceeds the ATT, and Eq. [4] for single-PLD VSASL, assuming the PLD is shorter than the ATT. For multi-PLD PCASL, multi-PLD VSASL, and MULTIVERSE ASL, CBF and ATT were fit using the general kinetic models (Eqs. [1-5]) with the nonlinear least square algorithm (lsqcurvefit), with fixed initial values of CBF = 50 mL / 100g / min, ATT = 1500 ms, and bounded by 0 < CBF < 100 mL / 100g / min and 0 < ATT < 6000 ms.

[0042] It is worth noting that the derivation above assumed that, the arterial blood after saturation would be completely replenished by the inflow of unsaturated fresh blood before the VS labeling module, which is only valid when the ATT is shorter than the saturation delay. However, if the saturation delay is shorter than ATT, the leading edge of the bolus at the time of VS labeling would not have been refreshed and its magnetization would still be undergoing saturation recovery. Thus, the simulation results, both generated and fitted using the Eqs. [4-5], would be more favorable for VSASL and MULTIVERSE ASL when the ATT is longer than the saturation delay (= 2.0 s). The model below can be fitted to characterize the kinetic functions moreby incorporating the ATT effect during the saturation delay. However, as shown by the Monte Carlo simulations, its added complexity did not improve combined accuracy and precision (RMSE) when applied to ASL data with very low SNR. The results generated using this more realistic model and fitted using both the simpler model and the same realistic model, indeed revealed increasing bias and RMSE for VSASL and MULTIVERSE ASL as ATT becomes longer. However, they still outperformed multi- PLD PCASL and exhibited minimal bias and RMSE across a wide range of ATT when the saturation delay was increased from 2.0 s to 3.0 s and 4.0 s respectively, as arterial blood after saturation slowly recovered back to equilibrium magnetization.

[0043] For VSASL, the arterial input function (AIF) for the labeled bolus is commonly described as a simple exponential decay with blood T1relaxation: =0 < t < ATTas the trailing edge of the labeled bolus, or the bolus duration. With the impulse response function (IRF) described as an exponential decay with tissue T1relaxation:

[0046] = , t > 0of this AIF and IRF, using a single effective T1, as described by Eqs. [4,5] and simplified as Eq. [7].

[0048] blood would be completely replenished by the inflow of unsaturated fresh blood before the labeling module. This AIF is only valid when the ATT is shorter than the saturation delay (Tsat), ATT < Tsat.

[0050] However, if Tsatis shorter than ATT, the leading edge of the bolus at the time of labeling would not have been refreshed and its magnetization would still be undergoing saturation recovery. Hence when ATT Tsat, the magnetization of the bolus as well as the corrected AIF would be a concatenation of the saturated blood during ATT-Tsat(light grey) followed by the fresh blood during Tsat (dark grey), as illustrated in FIGS.18C and 18D.

[0051] FIGS.18A-18D illustrate diagrammatic and graphical views of the arterial input function for the VSASL-labeled bolus under two conditions: when only unsaturated blood contributes, as illustrated in FIGS.18A and 18B, meaning fresh blood fully replenishes the cerebral blood supply during the saturation delay (Tsat) with TsatATT; and when both saturated and fresh blood contribute, as illustrated in FIGS.18C and 18D.

[0052] Below is the derivation of the corrected AIF and ASL signal under the condition of ATT Tsat:

[0054] AIF and the same IRF as for Eq. [7]:

[0057] on its relationship to ATT. The first two cases, are derived under the condition ATT < Tsat, while the last three cases are derived under the condition ATT Tsat.

[0062] Fitting performance metrics included accuracy and precision. Accuracy was assessed using the difference between the mean of the estimates and the ground truth (Bias) and Bias was normalized by the ground truth (nBias). Precision was assessed using the coefficients of variation (CoV, which corresponds to the SD of estimates divided by the mean of estimates). The root-mean-squared error (RMSE) between estimates and the ground truth was used to assess the combined accuracy and precision, i.e. RMSE = Bias2+ SD2. RMSE was further normalized by the ground truth to yield the normalized RMSE (nRMSE). The fitting performance at different noise levels was also compared across each ASL method. The fitting uncertainty of CBF and ATT was assessed by 95% confidence intervals (CI = (higher bond - lower bond) / 2) using the Matlab function nlparci and the normalized CI (nCI = CI / fitted value), respectively.

[0063] Evaluation of this method with human volunteers was performed on a 3T Philips Ingenia scanner (Philips Healthcare, Best, The Netherlands) using the body coil for transmission (maximum amplitude 13.5 T) and the 32-channel head coil for signal reception. The gradient coil has a maximum strength of 40mT / m and a maximum slew rate of 200 mT / m / ms, respectively. Nine healthy volunteers (47±17 years old, 4 males and 5 females) were included who provided written informed consent approved by the local institutional review board.

[0064] PCASLand VSASLboth started with a tailored saturation pulse train 25 with four 4.2 ms slab-selective Shinnar-Le Roux (SLR) pulses (89°, 98°, 82°, 157°), followed by the 1800 ms PCASLlabel / control module and a 2000 ms post-saturation delay before the FT-VSI labeling module, as illustrated in FIGS.1A-1C. The PCASL labeling plane was placed 90 mm inferior to the midline of the imaging volume which was centered at the anterior commissure-posterior commissure (AC-PC) line. BS pulses (hyperbolic-secant inversion)consisted of slab-selective pulses during PCASL labeling and non-selective pulses post- labeling for each PLD. The 64 ms FT-VSI labeling module and 20 ms vascular crushing module (VCM) both applied Vcut= 2.0 cm / s along foot-to-head (FH) direction. Five IS pulses were added to VSASL with equal spacing during each PLD period. Each IS pulse (2.4 ms 90° SLR) was implemented with a spatial width of 150 mm and immediately below the PCASL labeling plane.

[0065] FIG.1A-1C illustrate schematic diagrams of MULTIVERSE sequence. FIG.1A illustrates a pseudo-continuous arterial spin labeling (PCASL) sequence. FIG.1B illustrates a velocity selective arterial spin labeling (VSASL) sequence. FIG.1C illustrates spatial selectivity of RF pulses. PLD is post-labeling delay. BS is background suppression. Inferior saturation (IS) is applied spatially below the PCASL labeling plane and temporally after the VSASL labeling module; VCM is the vascular crushing module ACQ is acquisition.

[0066] ASL scans were acquired with 3D gradient- and spin-echo (GRASE) with the following parameters: FOV = 220 (anterior-to-posterior, AP) × 220 (left-to-right, LR) × 120 (FH) mm3, acquired resolution = 3.4× 3.6 × 5 mm3, reconstructed resolution = 3.4 × 3.4 × 5 mm3, 24 slices, slice oversampling factor = 1.3, EPI factor = 15, SENSE factor = 2 along the LR direction, turbo spin echo (TSE) factor = 16 along the FH direction, echo spacing = 15 ms, echo train duration = 240 ms, 4 shots for each repeat. PCASLand VSASLwith different PLDs were each repeated four times with 26 min total scan duration. Five-PLD VSASLwithout IS pulses was performed on a subset of two subjects. The SIPD scan was performed with a TR of 10 s and a duration of 1.2 min using the same 3D GRASE acquisition. CBF and ATT maps were obtained through voxel-wise fitting using Eqs. [1-5] as conducted in Simulations and were further interpolated to 1.72 × 1.72 mm2in-plane resolution. To accountfor individual differences in blood T1, the venous blood T1at the dominant side of the internal jugular vein (IJV) was measured using a fast protocol described previously.

[0067] Each participant underwent a high-resolution T1-weighted MPRAGE structural scan. Global gray matter (GM) and white matter (WM) masks were obtained by applying a threshold of 0.9 on tissue probability maps from the segmented and registered MPRAGE image using SPM. The WM mask was further eroded by 1 voxel. Quantitative analysis in the following steps was performed on voxels with nCI (defined in the simulation section) values less than 200% across all three methods for both CBF and ATT to remove outlier voxels. CBF and ATT and their nCI were averaged in the remaining voxels of GM and WM respectively.

[0068] The test-retest reliability of each method was assessed by the scatter plot and within-subject CoV. Within-subject CoVs of CBF, ATT, and their nCI were computed between the two repetitions of each ASL method for each subject across the GM and WM respectively. Student t-tests were applied to test the difference in GM CBF, ATT, nCICBF, nCIATT, and their CoV between every two ASL methods. A P-value of 0.002 (=0.05 / 24, Bonferroni correction for 24 comparisons) or less was considered significant.

[0069] Tables 1A and 1B show the timing parameters of PCASL, in Table 1A, and VSASL, for Table 1B, for each of the 5 post-labeling delay (PLD) used. Table 1A shows timing of background suppression (BS) pulses after the pre-saturation (pre-sat.) and after the PCASL labeling module are listed respectively. The minimal TRs for PLDs of 500 and 1000 ms were limited by the SAR constraint for PCASL and were only slightly shorter than that for PLD = 1500 ms. Hence the actual TRs were set to 4000 ms for these three PLDs; Table 1B shows timing of BS pulses and inferior saturation (IS) pulses after the VSASL labelingare listed respectively. For both PCASL and VSASL, with 4 shots for each k-space data, the duration (dur.) of 1, 2, and 4 repeats (rep.) of paired label and control at each PLD as well as all 5 PLDs are listed below. Note that when the duration of the preparation phase (set by the vendor) for PCASL and VSASL (0.40 and 0.32 min) at each PLD with 4 repeats were included, the total scan durations for PCASL and VSASL with 5 PLDs were 13.46 min and 12.12 min, respectively, resulting in a combined total of approximately 26 min for the entire protocol scanned. If all these scans are combined, only a single preparation phase would be required, potentially reducing the total scan duration. Table 1A PCASL: PLD (ms) 500 1000 1500 2000 2500 after 393, )Table 1B VSASL: PLD (ms) 500 1000 1500 2000 2500 BS timing 33 203 553 613 773 )

[0070] FIGS.1D, 1E, 1F, and 1G illustrate graphical views of pulse shapes of rest_2 pulses for different scenarios, according to an embodiment of the present invention. The pulse shapes of Philips-provided “rest_2” pulses for both the 4.2 ms pre-saturation with flip angles of [89°, 98°, 82°, 157°], as illustrated in FIG.1D and the 2.4 ms 90° inferior saturation (IS), as illustrated in FIG.1E, as well as corresponding spatially-selective normalized profiles through numerical simulation of Bloch equations, as illustrated in FIGS.1F and 1G. The solid vertical bars are the bandwidth defined by the vendor for pre-saturation of a 120 mm slab, as illustrated in FIG.1F and inferior saturation of a 150 mm slab, as illustrated in FIG. 1G, respectively. The brown vertical bands indicate the transition bands of 18 mm for pre- saturation, as illustrated in FIG.1F, and 23 mm for inferior saturation, as illustrated in FIG. 1G, respectively, determined by changes from 95% to 5% of the normalized magnetization. The spatially selective profiles within the transition bands only differ slightly between pulses with different flip angles.

[0071] FIGS.2A-2I provide examples of simulated kinetic curves and corresponding fit values for five-PLD PCASL, five-PLD VSASL, and MULTIVERSE ASL, with noise SD set as 0.34%, 0.34%, and 0.47% of SIPD(corresponding to 0.15% of SIPDfor single-PLD), under conditions of the true CBF = 60 mL / 100g / min and the true ATT = 600 ms (left column), 1800 ms (middle column), and 3000 ms (right column), respectively, and over 1000 random runs for each condition. When ATT fell within the normal range (e.g.1800 ms), both PCASL and VSASL exhibited robust signal strength and provided reliable estimates for CBF and ATT, as illustrated in FIG.2B versus FIG.2E; However, in cases of much shorter ATT (e.g. 600 ms), PCASLdemonstrated higher PWS compared to VSASL, resulting in more accurate CBF estimation and less ATT overestimation, as illustrated in FIG.2A versus FIG.2E.Conversely, with a longer ATT (e.g.3000 ms), VSASLshowed higher PWS than PCASL, leading to more accurate CBF estimation but greater ATT underestimation, as illustrated in FIG.2C versus FIG.2F. In contrast, the MULTIVERSE method yielded more accurate estimates for both CBF and ATT across the ATT range, as illustrated in FIGS.2G-2I compared to either PCASLor VSASL.

[0072] FIGS.2A-2I illustrate graphical views of fitted CBF and ATT on the simulated kinetic curves of perfusion-weighted signal (PWS) using true ATT values as 600 ms (left), 1800 ms (middle), and 3000 ms (right), multi-PLD PCASL, as illustrated in FIGS.2A-2C, multi-PLD VSASL, as illustrated in FIGS.2D-2F, and MULTIVERSE ASL, as illustrated in FIGS.2G-2I. The ground truth of CBF used 60 mL / 100g / min. Multi-PLD PCASL underestimates both CBF and ATT at ATT = 3000 ms, while multi-PLD VSASL underestimates CBF and overestimates ATT at ATT = 600 ms, and underestimates ATT at ATT = 3000 ms. MULTIVERSE ASL offers more accurate estimates of both CBF and ATT for all three cases.

[0073] FIGS.3A-3F illustrate a comparison of accuracy and precision through numerical simulations for single-PLD (2000 ms) PCASL and single-PLD (1500 ms) VSASL for CBF estimation only (noise SD = 0.15% of SIPD), as well as five-PLD PCASL, five-PLD VSASL, and five-PLD MULTIVERSE ASL (noise SD = 0.34%, 0.34%, and 0.47% of SIPD) for both CBF and ATT fitting. The absolute bias, SD, and RMSE of fitted results are shown in the left two columns, as illustrated in FIGS.3A, 3C, and 3E and the corresponding normalized ones are presented in the right two columns, as illustrated in FIGS.3B, 3D, and 3F. Across a broad spectrum of ATTs ranging from 500 ms to 4000 ms, MULTIVERSE ASL showed considerably lower nBias and CoV in CBF fitting compared to both single-PLD or multi-PLD PCASL and VSASL, as illustrated in FIGS.3B and 3D. Additionally, formoderate and long ATT, MULTIVERSE ASL demonstrated lower nBias and lower CoV in ATT estimation compared to multi-PLD PCASL and VSASL. Overall, MULTIVERSE exhibited lower nRMSE in both CBF and ATT fitting across the ATT range compared to multi-PLD PCASL and VSASL, as illustrated in FIG.3F. When the noise level changes, this relation still holds as shown in the nRMSE as a function of noise SD, as illustrated in FIGS. 10A-10F.

[0074] FIGS.10A-10F illustrate graphical views of normalized root-mean-square-error (nRMSE) for estimating CBF, as illustrated in FIGS.10A-10C and ATT, as illustrated in FIGS.10D-10F at different noise levels. The x-axis represents the noise SD in the percentage of the signal intensity in the proton density image (SIPD) for the single-PLD ASL simulation, which was scaled up for five-PLD PCASL (× 5), VSASL (× 5), and MULTIVERSE (× 10) simulations, respectively. nRMSE was obtained from the Monte Carlo simulations of single-PLD PCASL and VSASL, five-PLD PCASL, VSASL, and MULTIVERSE. The ground truth of CBF was 60 mL / 100g / min. The ground truth of ATT used 600 ms (left), 1800 ms (middle), and 3000 ms (right), respectively.

[0075] FIGS.3A-3F illustrate graphical views of Monte Carlo simulations of accuracy. FIG.3A illustrates bias, FIG.3B illustrates normalized Bias (nBias) and precision, FIG.3C illustrates standard deviation (SD), FIG.3D illustrates the coefficient of variation (CoV), and the combination of accuracy and precision, FIG.3E illustrates root-mean-square-error (RMSE), and FIG.3F illustrates normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing different ASL methods. The ground truth of CBF was 60 mL / 100g / min. The PLDs were chosen as follows:

[2000] ms for single-PLD PCASL (dark grey dash);

[1500] ms for single-PLD VSASL (light grey dash); [500, 1000, 1500, 2000, 2500] ms for multi-PLD PCASL (dark grey solid), multi-PLD VSASL (light grey solid), andMULTIVERSE ASL (medium grey solid). The black solid horizontal line indicates 0% in FIGS.3A and 3B.

[0076] FIGS.4A-4H illustrate the histogram of nCI counts of fitted CBF and ATT with ATTs ranging from 0 to 1000 ms, 1000 to 2000 ms, 2000 to 3000 ms, and 3000 to 4000 ms separately. When true ATTs are shorter than 2000 ms, multi-PLD VSASL exhibits more counts with larger nCI (indicating increased uncertainty, light grey) in derived CBF, as illustrated in FIGS.4A and 4C and ATT, as illustrated in FIGS.4B and 4D values compared to multi-PLD PCASL (dark grey). As true ATTs increase beyond 2000 ms, multi-PLD VSASL shows more counts with smaller nCI (lower uncertainty) in CBF values than multi- PLD PCASL, as illustrated in FIGS.4E and 4G. For true ATTs longer than 3000 ms, multi- PLD VSASL and multi-PLD PCASL achieve similar nCI in fitted ATT values, as illustrated in FIG.4H. Across a broad ATT range, CBF values derived from MULTIVERSE ASL (medium grey) yield more counts with smaller nCI (lower uncertainty) compared to either PCASL or VSASL, as illustrated in FIGS.4A, 4C, 4E, and 4G. When true ATTs are shorter than 1000 ms, ATT values estimated from MULTIVERSE are comparable to those from multi-PLD PCASL, as illustrated in FIG.4B. As true ATT values increase, MULTIVERSE ASL leads to more counts with smaller nCI (lower uncertainty) for ATT estimations compared to PCASL and VSASL, as illustrated in FIGS.4D, 4F, and 4H.

[0077] FIG.4A-4H illustrates graphical views of histograms of the normalized 95% confidence interval (nCI) of fitted CBF and ATT from the Monte Carlo simulations for multi- PLD PCASL (dark grey), multi-PLD VSASL (light grey), and MULTIVERSE ASL (medium grey), all with PLD = [500, 1000, 1500, 2000, 2500] ms. FIGS.4A, 4, C, 4E, and 4G illustrate fitted CBF and FIGS.4B, 4D, 4F, and 4H illustrate ATT from the Monte Carlo simulations. The true CBF is 60 mL / 100g / min and true ATT values range from 0 to 1000 ms,as illustrated in FIGS.4A and 4B, 1000 to 2000 ms, as illustrated in FIGS.4C and 4D, 2000 to 3000 ms, as illustrated in FIGS.4E and 4F, 3000 to 4000 ms, as illustrated in FIGS.4G and 4H.

[0078] The evaluation of this method showed that the difference in VSASL between those with and without IS pulses is more obvious in the fast flow territories such as the anterior cerebral circulation, especially for participants with the faster flow rate, as illustrated in FIGS.11A and 11B. FIG.5 and FIG.6 illustrate the ASL results obtained from a 29-year-old and a 63-year-old healthy female, respectively. Multi-delay PCASL (1st row) and VSASL (2nd row) at each PLD averaged from four repetitions exhibit higher PWS in VSASL than in PCASL from the posterior part of the brain, as expected for the territories with longer ATT. The CBF and ATT maps generated from 5-PLD PCASL (3rd row, two repetitions) and VSASL (4th row, two repetitions), as well as MULTIVERSE ASL with combined fitting (5th row, one repetition), all in 6.5 min, maintain reasonable consistency with those derived from the entire 26-minute dataset of MULTIVERSE ASL (6th row, four repetitions) while exhibiting higher noise levels. Compared to multi- PLD PCASL, CI and nCI of CBF and ATT maps are substantially larger in VSASL, but smaller in the 6.5 min MULTIVERSE ASL and smallest in the 26 min MULTIVERSE ASL results for both subjects. In addition to the higher CBF and shorter ATT, the GM also shows lower CI and nCI than the WM. As the 29- year-old female had notably higher CBF and shorter ATT than the 63-year-old female, the CI and nCI values were also correspondingly smaller, as illustrated in FIG.5 vs FIG.6.

[0079] FIG.5 illustrates image views of PCASL, VSASL, and MULTIVERSE results from a 29-year-old healthy female. Perfusion-weighted signal (PWS) images were acquired by five-PLD PCASL, and five-PLD VSASL with inferior saturation (IS) pulses, both with 4 repetitions (rep.). CBF, ATT, and their 95% confidence interval (CI) and normalized CI (nCI)maps were estimated by fitting the data from 6.5-minute scans of PCASL (2 repetitions), VSASL (2 repetitions), MULTIVERSE ASL (1 repetition), and a 26-minute MULTIVERSE ASL scan (4 repetitions).

[0080] FIG.6 illustrate image views of PCASL, VSASL, and MULTIVERSE results from a 63-year-old healthy female. Perfusion-weighted signal (PWS) images were acquired by five-PLD PCASL, and five-PLD VSASL with inferior saturation (IS) pulses, both with 4 repetitions (rep.). CBF, ATT, and their 95% confidence interval (CI) and normalized CI (nCI) maps were estimated by fitting the data from 6.5-minute scans of PCASL (2 repetitions), VSASL (2 repetitions), MULTIVERSE ASL (1 repetition), and a 26-minute MULTIVERSE ASL scan (4 repetitions).

[0081] FIG.11A and 11B illustrate image views of a comparison between VSASL PWS images with and without IS pulses on a 27-year-old female, as illustrated in FIG.11A, and a 32-year-old male, as illustrated in FIG.11B. VSASL with IS pulses shows lower PWS than VSASL without IS pulses in faster flow territories from the anterior circulation at earlier PLDs, as the IS pulses shorten the bolus from below the labeling plane reaching the imaging volume during the post-labeling delay (PLD) period.

[0082] The histograms of nCI in the same two subjects, as illustrated in FIGS.7A-7D also suggest MULTIVERSE ASL produced a higher percentage of voxels in smaller nCI in both CBF and ATT than multi-PLD PCASL alone. Additionally, multi-PLD VSASL generated clearly more counts of large nCIs than both multi-PLD PCASL and MULTIVERSE ASL. Note that the distributions of nCI were comparable between simulations with ATT ranges from 1000 to 2000 ms, as illustrated in FIGS.4C and 4D. and these two in vivo data sets, as illustrated in FIGS.7A-7D. FIGS.12A and 12B illustrate the nCI maps of three ASLmethods in one subject. Voxels with nCI higher than 200% were mostly from multi-PLD VSASL and were primarily located in WM, near the circle of Willis, and in regions with pulsatile CSF.

[0083] FIG.12A and 12B illustrate image views of a normalized 95% confidence interval (nCI) maps of CBF, as illustrated in FIG.12A and ATT, as illustrated in FIG.12B, from PCASL, VSASL, and MULTIVERSE ASL in the same subject as FIG.5. Note that voxels with nCI higher than 200% were mostly from VSASL, primarily located in white matte, near the circle of Willis, and regions with pulsatile CSF.

[0084] FIGS.7A-7D illustrate graphical views of histograms of the normalized 95% confidence interval (nCI) of fitted CBF and ATT from the two subjects shown in FIG.5 and FIG.6 for multi-PLD PCASL (dark grey), multi-PLD VSASL (light grey), and MULTIVERSE ASL (medium grey), all with PLD = [500, 1000, 1500, 2000, 2500] ms. FIGS.7A and 7C, illustrate fitted CBF and FIGS.7B and 7D illustrate ATT from the Monte Carlo simulations. These in vivo results are similar to the simulation results of ATTs ranging from 1000 to 2000 ms, as illustrated in FIGS.4C and 4D.

[0085] FIG.8 illustrates the CBF and ATT maps in three orthogonal orientations generated by fitting MULTVERSE ASL with two of the four repetitions (6.5 minutes each) from all 9 participants. The first and third scans demonstrate good test-retest repeatability across regions and subjects, with a clear contrast between GM and WM in CBF maps (P<0.001) and a trend in ATT maps (P=0.065). ATT maps also highlight the regional differences between the anterior and posterior circulations of the brain. Additionally, females show a trend of higher CBF and shorter ATTs compared to males, and younger subjects exhibit a trend of higher CBF and shorter ATT compared to older individuals.

[0086] FIG.8 illustrates orthogonal views of individual 3D CBF and ATT maps from 9 subjects obtained by MULTIVERSE ASL with first and third repetitions (6.5 minutes each) showing good test-retest repeatability. The sex (F: female; M: male) and age (in years) of each subject are listed on the left, with their CBF and ATT values averaged from global grey matter and white matter masks (GM / WM) shown below respective images.

[0087] The averaged CBF and ATT values of global GM and WM from these individuals show high correlations between repetitions of multi-PLD PCASL, multi-PLD VSASL, and MULTIVERSE ASL, respectively, as illustrated in FIGS.9A and 9B. Table 2 lists quantitative fitting results by the three ASL methods with test-retest evaluations averaged across 9 subjects (36±14%of voxels in GM and 9±6% of voxels in WMwere included after the 200% nCI thresholding). The mean GM values of between two ASL methods were compared: Multi-PLD VSASL demonstrated about 57% higher CBF than multi-PLD PCASL (54.5 vs.34.8 mL / 100g / min) (P<0.001), while MULTIVERSE ASL yielded intermediate CBF values (42.4 mL / 100g / min); Multi-PLD VSASL showed comparable ATT to multi-PLD PCASL (1287 vs.1415 ms, P=0.16), while MULTIVERSE ASL obtained the longest ATT (1600 ms, P<0.001). MULTIVERSE ASL demonstrated 36% and 29% narrower nCI in fitting CBF and ATT than multi-PLD PCASL (P=0.008 for nCICBF, P<0.001 for nCIATT), and 53% and 68% narrower nCI than multi-PLD VSASL (P<0.001 for both). The CoV between two repetitions of MULTIVERSE ASL (8.4% for CBF, 7.6% for ATT) was not significantly different from that of multi-PLD PCASL (7.9% for CBF, 7.2%for ATT, P=0.3 for CBF, P=0.5 for ATT), and was 27% and 47% lower respectively, compared to multi-PLD VSASL (11.5% for CBF, 14.3% for ATT, p<0.001 for both). WM presented similar trends as GM but were with wider nCI and higher CoV for all the metrics, as expected with lower SNR.

[0088] FIGS.9A and 9B illustrate graphical views of scatter plots for test-retest evaluations of the averaged CBF, as illustrated in FIG.9A and ATT, as illustrated in FIG.9B, of global gray matter (GM, dark grey) and white matter (WM, medium grey) over 9 subjects obtained from multi-PLD PCASL (averages from 1st and 2nd repetitions (rep.1, 2) vs. averages from 3rd and 4th repetitions (rep.3, 4), multi-PLD VSASL (rep.1, 2 vs. rep.3, 4), and MULTIVERSE ASL (rep.1 vs. rep.3), all with PLD = [500, 1000, 1500, 2000, 2500] ms. The solid black line indicates the line of equality. Their correlation coefficients are 0.99 for all the plots.

[0089] Table 2 lists quantitative fitting results by the three ASL methods with test-retest evaluations averaged across 9 subjects (36±14% of voxels in GM and 9±6% of voxels in WM were included after the 200% nCI thresholding). The mean GM values of between two ASL methods were compared: Multi-PLD VSASL demonstrated about 57% higher CBF than multi-PLD PCASL (54.5 vs.34.8 mL / 100g / min) (P<0.001), while MULTIVERSE ASL yielded intermediate CBF values (42.4 mL / 100g / min); Multi-PLD VSASL showed comparable ATT to multi-PLD PCASL (1287 vs.1415 ms, P=0.16), while MULTIVERSE ASL obtained the longest ATT (1600 ms, P<0.001). MULTIVERSE ASL demonstrated 36% and 29% narrower nCI in fitting CBF and ATT than multi-PLD PCASL (P=0.008 for nCICBF, P<0.001 for nCIATT), and 53% and 68% narrower nCI than multi-PLD VSASL (P<0.001 for both). The CoV between two repetitions of MULTIVERSE ASL (8.4% for CBF, 7.6% for ATT) was not significantly different from that of multi-PLD PCASL (7.9% for CBF, 7.2% for ATT, P=0.3 for CBF, P=0.5 for ATT), and was 27% and 47% lower respectively, compared to multi-PLD VSASL (11.5% for CBF, 14.3% for ATT, p<0.001 for both). WM presented similar trends as GM, but were with wider nCI and higher CoV for all the metrics, as expected with lower SNR.

[0090] Table 2: The mean and standard deviation (SD) of CBF and ATT values from global gray matter (GM) and white matter (WM) masks across 9 subjects along with their normalized 95% confidence interval (nCI) of the fitting. PCASL, VSASL, and MULTIVERSE ASL all used five matched PLDs, [500, 1000, 1500, 2000, 2500] ms. Multi- PLD PCASL and VSASL each had 4 repetitions (rep.), with rep.1,2 or rep.3,4 indicating that the averaged values of repetitions 1 and 2 or 3 and 4 are reported. MULTIVERSE ASL combined PCASL and VSASL, with rep.1 and rep.3 indicating that the results of the repetition 1 and 3 are reported respectively. Thus, all data below were obtained from the same scan durations of 6.5 min. The within-subject coefficient of variance (CoV) for each metric is also reported for evaluating test-retest repeatability. CBF ATT (ms) nCICBF(%) nCIATT(%) (mL / 100g / min) .1 9 6

[0091] The present invention sets forth a novel method called MULTIVERSE ASL. It applies combined fitting of separately acquired PCASL and VSASL with multiple PLDs for concurrent CBF and ATT measurements. Its utility is illustrated numerically in terms of accuracy and precision over an extended ATT range of up to 4000 ms, which is impossiblewith existing ASL methods. The brain scans from healthy young participants demonstrate the feasibility and reliability of the technique, with its fitting certainty and repeatability compared with multi-PLD PCASL and VSASL. Alternative sampling and imaging strategies as well as potential technical improvements should be examined in future studies.

[0092] PLD timings are an important consideration to allow fitting CBF and ATT within a wide range of ATTs. In addition to multiple matched PLDs in PCASL and VSASL as implemented in the current invention, choosing single matched or mismatched, as well as multiple mis-matched PLDs, were also explored numerically for MULTIVERSE ASL: single-matched PLDs (

[1000] ms,

[1500] ms,

[2000] ms) for both PCASL and VSASL; single mis-matched PLDs (

[1500] /

[1000] ms,

[2000] /

[1000] ms,

[2000] /

[1500] ms) for PCASL / VSASL or VSASL / PCASL; multiple mis-matched PLDs ([1000, 1500] / [1500, 2000] ms, [500, 1000, 1500] / [1000, 1500, 2000] ms, [500, 1000, 1500, 2000] / [1000,1500, 2000, 2500] ms) for PCASL / VSASL or VSASL / PCASL. The preliminary simulation results showed that compared to the fitting performance by the MULTIVERSE ASL with five matched PLDs as the default in this invention, the single matched PLDs resulted in up to (~8.6%) lower accuracy in CBF / ATT fitting when ATTs were close the chosen PLDs, as illustrated in FIGS.13A-13F, while the single mis-matched PLDs produced considerably lower accuracy and precision, as illustrated in FIGS.14A-14F and FIGS.15A-15F. The multiple mismatched PLDs generated similar performance compared to the five matched PLDs, and choosing PCASL PLDs shorter than VSASL PLDs slightly underperformed than selecting the opposite configurations, as illustrated in FIGS.16A-16F and FIGS.17A-17F. The in vivo data derived from mis-matched PLDs ([1000, 2000] / [1500, 2000] ms, [500, 1000, 1500] / [1000, 1500, 2000] ms, [500, 1000, 1500, 2000] / [1000,1500, 2000, 2500] ms) for PCASL / VSASL with 2 repetitions were also close to the results from the default matchedPLDs (data not shown). For better accuracy and precision of CBF and ATT estimation by MULTIVERSE ASL, the PLD sampling strategy in terms of the number of PLDs and their timings for both PCASL and VSASL as well as the PCASL LDs can be further investigated using an established optimization framework. Furthermore, analysis techniques other than full nonlinear model fitting, e.g. Bayesian inference or a signal-weighted delay algorithm, may be employed for greater robustness.

[0093] FIGS.13A-13F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using a single matched PLD with using multiple matched PLDs.

[0094] FIGS.14A-14F illustrate Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square-error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using single mis-matched PLD with using multiple matched PLDs.

[0095] FIGS.15A-15F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using single mis-matched PLD with using multiple matched PLDs.

[0096] FIGS.16A-16F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using multiple mis-matched PLDs with using multiple matched PLDs.

[0097] FIGS.17A-17F illustrate graphical views of Monte Carlo simulations of accuracy, (a) Bias, (b) normalized Bias (nBias), precision, (c) standard deviation (SD), (d) coefficient of variation (CoV), and the combination of accuracy and precision, (e) root-mean-square- error (RMSE), (f) normalized RMSE (nRMSE), of fitted CBF and ATT as a function of true ATT comparing MULTIVERSE ASL using multiple mis-matched PLDs with using multiple matched PLDs. The ground truth of CBF was 60 mL / 100g / min.

[0098] Compared to multi-PLD PCASL, MULTIVERSE ASL improved the fitting certainty of CBF and ATT by lowering the nCI, as shown by both simulations, as illustrated in FIGS.4A-4H and in vivo results of normal cerebral vasculature with ATT values between 1000 ms and the lower 2000 ms, as illustrated in FIG.5, FIG.6, and Table 2. Conversely, multi-PLD VSASL displayed much lower fitting certainty, which might reflect its lower sensitivity to ATT in the normal range, as illustrated in FIG.4D. The resemblance between FIGS.4C and 4D and FIGS.7A and 7B supports the 95% nCI as a reliable metric to evaluate the model fit. Better fitting certainty (narrower 95% nCI) could be achieved with higher SNR, as illustrated by the contrast between GM and WM as illustrated in FIG.5, FIG.6, and Table 2.

[0099] The in vivo repeatability (test-retest CoV) of CBF and ATT derived from multi- PLD PCASL and MULTIVERSE ASL were comparable, and both were significantly better than multi-PLD VSASL, as shown in Table 2. This might also be due to the normal ATT range in this cohort. When ATTs become longer, the repeatability of multi-PLD PCASL would be worse than that of multi-PLD VSASL and MULTIVERSE ASL, as suggested by the precision (CoV) of the Monte-Carlo simulation results suggested, as illustrated in FIGS. 3A-3F.

[0100] Instead of using the conventional VS saturation pulse train in VESPA, this work applied the FTVSI for labeling to improve the SNR. The usefulness of FT-VS pulse trains has been demonstrated in VSMRA and quantitative mapping of blood flow blood volume, and venous oxygenation. MULTIVERSE ASL added the IS pulses onto VSASL following the VS labeling module. Defining spatial selectivity of the VS labeled bolus and ensuring its trailing edge aligns with the PCASL labeling plane was first proposed by the VESPA technique, which replaced the excitation hard pulses with slab-selective pulses and associated gradients in the conventional VS saturation labeling module. This strategy would inevitably lengthen the FT-VS pulse train and increase its sensitivity to T2. as well as flow acceleration / deceleration in large vessels, both leading to reduced labeling efficiency. Inferior saturation has been implemented in PASL such as QUIPSS II and Q2TIPS to define the distal edge of the labeling bolus, as well as in PCASL to suppress the unwanted inflowing blood. Given the location of the targeted area, the IS pulses may need to be further optimized to attain both a sharp transition band and robustness to B0 / B1 field inhomogeneities.

[0101] Notably, the in vivo results showed that multi-PLD VSASL produced higher CBF and shorter ATT than multi-PLD PCASL, FIG.5, FIG.6, and Table 2, and the MULTIVERSE-derived results were in between those from VSASL and PCASL. TheVESPA study also reported higher CBF measurements than PCASL and alluded to several sources. In the current work, an additional issue is that the applied labeling efficiency of = 0.56 might be underestimated for FT-VSI labeling pulse trains. Earlier FT-VSI-based VSASL studies estimated = 0.57 for PLD of 1.5 s and = 0.61 for PLD of 1.2 s. As recently observed,when PLDs are acquired longer than the bolus duration, VSASL would underestimate CBF. This affects mostly populations and territories with fast flow. Furthermore, the leading edge of bolus acquired at shorter PLDs would be closer to the brain and experience less B0 / B1 inhomogeneities, in contrast to the trailing edge of the bolus labeled at the neck and below with suboptimal shimming conditions, thus contributing various labeling efficiencies.

[0102] Several technical explorations of the current work merit continuous pursuit. MULTIVERSE ASL with 3D rotated spiral RARE readout or stack-of-spirals turbo FLASH readout would achieve single-shot acquisition by using compressed sensing or more advanced reconstruction techniques. This would allow increasing the number of repeats and / or the number of PLDs within the same scan time and further enhance the fitting performance. Additionally, when sampling at greater PLDs, the effect of T1relaxationin different tissues (GM or WM, normal or pathology) could also be taken into account by fitting T1, eff as the third parameter in addition to CBF and ATT. MULTIVERSE ASL could also be employed on the rest of the body, potentially alleviating the vast uncertainty of the transit time variability for different vascular territories and among different populations, in which multi-PLD PASL instead of PCASL might be a more practical choice for the spatially selective labeling method.

[0103] The combined fitting of multi-PLD PCASL and VSASL, MULTIVERSE ASL improved the quantification of CBF and ATT across an extended range of ATTs. Itsfeasibility and reliability were demonstrated among healthy young volunteers, exhibiting greater fitting certainty than, and comparable repeatability to, multi-PLD PCASL. This novel method highlights the clinical potential of ASL in various altered physiological and pathological conditions.

[0104] It should be noted that the pulse sequences, imaging protocols, described herein can be executed with a program(s) fixed on one or more non-transitory computer readable medium. The non-transitory computer readable medium can be loaded onto a computing device, server, imaging device processor, smartphone, tablet, phablet, or any other suitable device known to or conceivable by one of skill in the art.

[0105] It should also be noted that herein the steps of the method described can be carried out using a computer, non-transitory computer readable medium, or alternately a computing device, microprocessor, or other computer type device independent of or incorporated with an imaging or signal collection device. An independent computing device can be networked together with the imaging device either with wires or wirelessly. The computing device for executing the present invention can be a completely unique computer designed especially for the implementation of this method. Indeed, any suitable method of analysis known to or conceivable by one of skill in the art could be used. It should also be noted that while specific equations are detailed herein, variations on these equations can also be derived, and this application includes any such equation known to or conceivable by one of skill in the art.

[0106] A non-transitory computer readable medium is understood to mean any article of manufacture that can be read by a computer. Such non-transitory computer readable media includes, but is not limited to, magnetic media, such as a floppy disk, flexible disk, hard disk, reel-to-reel tape, cartridge tape, cassette tape or cards, optical media such as CD-ROM,writable compact disc, magneto-optical media in disc, tape or card form, and paper media, such as punched cards and paper tape.

[0107] It should be noted that the software associated with the present invention is programmed onto a non-transitory computer readable medium that can be read and executed by any of the computing devices mentioned in this application. The non-transitory computer readable medium can take any suitable form known to one of skill in the art. The non- transitory computer readable medium is understood to be any article of manufacture readable by a computer. Such non-transitory computer readable media includes, but is not limited to, magnetic media, such as floppy disk, flexible disk, hard disk, reel-to-reel tape, cartridge tape, cassette tapes or cards, optical media such as CD-ROM, DVD, Blu-ray, writable compact discs, magneto-optical media in disc, tape, or card form, and paper media such as punch cards or paper tape. Alternately, the program for executing the method and algorithms of the present invention can reside on a remote server or other networked device. Any databases associated with the present invention can be housed on a central computing device, server(s), in cloud storage, or any other suitable means known to or conceivable by one of skill in the art. All of the information associated with the application is transmitted either wired or wirelessly over a network, via the internet, cellular telephone network, RFID, or any other suitable data transmission means known to or conceivable by one of skill in the art.

[0108] The many features and advantages of the invention are apparent from the detailed specification, and thus, it is intended by the appended claims to cover all such features and advantages of the invention which fall within the true spirit and scope of the invention. Further, since numerous modifications and variations will readily occur to those skilled in the art, it is not desired to limit the invention to the exact construction and operation illustratedand described, and accordingly, all suitable modifications and equivalents may be resorted to, falling within the scope of the invention.

Claims

What is claimed is:

1. A method for magnetic resonance imaging of a subject comprising: a scan, wherein the scan comprises: using a multi-time point pseudo-continuous arterial spin labeling (PCASL) and multi-time point velocity-selective spin labeling (VSASL); and applying combined fitting to measure cerebral blood flow (CBF); applying combined fitting to measure arterial transit time (ATT).

2. The method of claim 1 wherein the multi-time-point VS comprises a spatially defined bolus.

3. The method of claim 1 further comprising using several inferior saturation pulses (ISP) following the VS module to crush the inflow of blood below the labeling plane of PCASL.

4. The method of claim 1 wherein the fitted parameter of ATT is the same between PCASL and VSASL.

5. The method of claim 1 wherein VSASL with ISP and PCASL are acquired separately, using the same multi-time point scheme.

6. The method of claim 1 further comprising applying a pair of matched post labeling delays (PLDs) for PCASL and VSASL.

7. The method of claim 1 further comprising applying post labeling delays (PLDs) optimized for the scan.

8. The method of claim 1 further comprising applying mismatched post labeling delays (PLDs).

9. The method of claim 8 further comprising applying multiple mismatched PLDs. The method of claim 1 further comprising measuring CBF and ATT simultaneously.

11. A system for magnetic resonance imaging of a subject comprising: a magnetic resonance imaging (MRI) machine, wherein the MRI machine is configured to execute a scan, wherein the scan comprises: using a multi-time point pseudo-continuous arterial spin labeling (PCASL) and multi-time point velocity-selective spin labeling (VSASL); and applying combined fitting to measure cerebral blood flow (CBF); applying combined fitting to measure arterial transit time (ATT).

12. The system of claim 11 wherein the multi-time-point VS comprises a spatially defined bolus.

13. The system of claim 11 further comprising using several inferior saturation pulses (ISP) following the VS module to crush the inflow of blood below the labeling plane of PCASL.

14. The system of claim 11 wherein the fitted parameter of ATT is the same between PCASL and VSASL.

15. The system of claim 11 wherein VSASL with ISP and PCASL are acquired separately, using the same multi-time point scheme.

16. The system of claim 11 further comprising applying a pair of matched post labeling delays (PLDs) for PCASL and VSASL.

17. The system of claim 11 further comprising applying post labeling delays (PLDs) optimized for the scan.

18. The system of claim 11 further comprising applying mismatched post labeling delays (PLDs).

19. The system of claim 18 further comprising applying multiple mismatched PLDs.

20. The system of claim 11 further comprising measuring CBF and ATT simultaneously.

Images