A Test-Retest Study of Single- and Multi-Delay pCASL for Choroid Plexus Perfusion Imaging in Healthy Subjects Aged 19 to 87 Years

Abstract

There is a growing interest in the choroid plexus (ChP) due to its critical role in cerebrospinal fluid (CSF) production and its involvement in neurodegenerative and cerebrovascular diseases. However, comprehensive studies comparing the accuracy and reliability of single- and multi-PLD (post-labeling delay) arterial spin labeling (ASL) techniques, specifically in relation to the ChP, remain limited. This study systematically evaluated the test-retest reliability and quantification accuracy of cerebral blood flow (CBF) measurements, focusing on the ChP, using single-delay and multi-delay 3D gradient-and-spin echo (GRASE) pseudo-continuous ASL (pCASL) on 28 subjects (aged 19 to 87 years, 14 males/14 females) at 3.0 tesla. Both single-delay (2 sec) and 5-PLD (0.5 – 2.5 sec) pCASL scans were repeated approximately one week apart with a spatial resolution of 2.5×2.5×3 mm³. Voxel-wise and regional CBF and arterial transit time (ATT) measurements were compared to assess test-retest reliability, with a particular focus on ChP perfusion changes with age. In this study, 12.15% of ChP voxels exhibited ATTs longer than 2 sec, potentially leading to a significant underestimation of CBF using single-delay ASL. Multi-delay ASL showed improved accuracy in estimating CBF values for the ChP compared to single-delay ASL when ATT > PLD. Additionally, ChP volume (mean ± std = 1.72± 0.85 ml) increased (p < 0.01) and ChP perfusion (43.07±14.18 mL/100 g/min) decreased (p = 0.04) with age. These findings underscore the robustness of multi-delay ASL with model-fitting quantification in assessing ChP perfusion, making it the preferred method for accurate CBF and ATT estimation, particularly in regions with prolonged transit time such as ChP.

1. INTRODUCTION

The choroid plexus (ChP) plays a major role in maintaining brain homeostasis through various processes, including the production and secretion of cerebrospinal fluid (CSF), regulation of nutrient and metabolite exchange, and the expression of neurotrophic factors and signaling molecules (Lun et al., 2015). The ChP is composed of a monolayer of epithelial cells connected by tight junctions, forming the blood-CSF barrier. These cells are highly vascularized and surrounded by fenestrated capillaries within the stroma, allowing for the filtration of fluids and the transport of essential nutrients and waste products between the blood and CSF (Liddelow, 2015).

ChP is involved in the glymphatic waste clearance system, further demonstrating its significance in neurological health, particularly in conditions such as Alzheimer’s disease (Carna et al., 2023; Hubert et al., 2019; Krzyzanowska & Carro, 2012). In AD, ChP abnormalities, such as increased volume, structural health, and inflammatory transcriptomic, are linked to cognitive decline and disrupted homeostasis (Carna et al., 2023). Recent studies have also suggested that ChP perfusion is sensitive to ischemic stress, microvascular compliance, and aging-related changes, making it a vital area of investigation for neurodegenerative diseases and normal aging processes (Alisch et al., 2021; Johnson et al., 2020). In multiple sclerosis (MS), ChP enlargement is associated with chronic inflammation, lesion expansion, and brain atrophy (Klistorner et al., 2022), while in depression, it correlates with neuroinflammation and reduced blood-brain barrier permeability (Althubaity et al., 2022; Turkheimer et al., 2021). These findings emphasize the ChP as a central regulator of brain health, a biomarker for disease progression, and a promising therapeutic target for a wide range of neurological and psychiatric disorders.

However, despite its critical functions, the ChP has received relatively little attention in neuroimaging studies compared to other brain structures, partly due to its small size and the difficulty of distinguishing it from surrounding tissues using conventional imaging techniques (Johnson et al., 2020; Sakka et al., 2011). The development of high-resolution non-invasive imaging methods, such as Arterial Spin Labeling (ASL), has provided new opportunities to explore ChP perfusion and its role in brain health and disease.

ASL is a non-invasive MR technique for perfusion imaging of the brain and body organs (Buxton et al., 1998; Detre et al., 1992; Taso et al., 2023; Williams et al., 1992). ASL quantifies cerebral blood flow (CBF) by magnetically labeling arterial blood water as it flows into the brain. Among the various ASL techniques, pseudo-continuous ASL (pCASL) is recommended due to its higher labeling efficiency, superior signal-to-noise ratio (SNR), and reliability compared to pulsed ASL (PASL) (Pohmann et al., 2010). One of the key parameters in pCASL is the post-labeling delay (PLD), which is the time between the end of labeling and the signal acquisition. and an optimal PLD is crucial for accurate CBF quantification (Ferre et al., 2013).

Most clinical studies use single-delay ASL due to shorter scan times and easy implementation. However, single-delay ASL assumes that the arterial transit time (ATT) is shorter than the PLD, an assumption that may be violated due to the heterogeneous distribution of ATT across different brain regions, aging, and pathological conditions (Alsop et al., 2015; Haller et al., 2016). To address these limitations, multi-delay ASL can be used, acquiring perfusion signals at multiple PLDs, thus allowing for the simultaneous estimation of CBF and ATT through kinetic modeling (Woods et al., 2024). Given the ChP’s highly vascularized structure and complex interactions within the brain, there is a growing interest in applying advanced ASL techniques to study ChP perfusion. However, whether single-delay or multi-delay ASL offers the potential to better understand ChP perfusion dynamics remains unknown.

The goal of this study is to perform a comprehensive and quantitative comparison between single-delay and multi-delay ASL measurements of perfusion of the ChP as well as the gray and white matter. By employing relatively high spatial resolution, we aim to achieve voxel-wise analysis of both single-delay and multi-delay ASL approaches, providing insights into the reliability and accuracy of CBF measurements in smaller brain structures, such as the ChP. These findings will contribute to the optimization of ASL techniques in clinical studies and future research on ChP function and brain health.

2. Methods

2.1. ASL kinetic model

For pCASL, CBF can be calculated from perfusion signal ΔM according to single compartment perfusion model (Wang et al., 2013):

$$CBF=\frac{\lambda \cdot \mathrm{\Delta }M\cdot R_{1a}}{2\cdot \alpha \cdot M_0\cdot \left[exp\left(\left(min\right(ATT-PLD,0)-ATT)\cdot R_{1a}\right)-exp\left(-(\tau +PLD)\cdot R_{1a}\right)\right]}$$ [1]

where $\lambda$ is the partition coefficient of water in the brain (0.9 g/ml), $\alpha$ is labeling efficiency = 0.735 (considering the efficiency of 2 background suppression pulses), $R_{1a}$ is longitudinal relaxation rate of blood = 0.61 s⁻¹, τ is labeling duration = 1.8 s. PLD is the post-labeling delay, and M0 is the equilibrium magnetization of brain tissue.

2.2. MRI experiments

The imaging parameters for 3D gradient-and-spin echo (GRASE) pCASL were: FOV=240×240 mm², resolution=2.5×2.5×3 mm³, 40 slices (20% oversampling), 4 segments along partition direction, TE = 37.8 ms, TR = 4500 ms, labeling duration = 1800 ms. A background suppression scheme including a pre-saturation pulse and two hyperbolic secant inversion pulses was used and can be flexibly adjusted for different PLDs (Shao et al., 2017). Both single-delay and 5-PLD pCASL scans were performed. PLD = 2 s was used for the single-delay protocol, 9 repetitions were acquired in 5 min 24 s. PLDs of [0.5, 1, 1.5, 2, and 2.5] s with [2, 2, 2, 3, 3] repetitions were used for the multi-delay protocol in 7 min 12 sec, respectively. A separate M₀ scan was acquired in 18 s. 3D magnetization-prepared rapid gradient echo (MPRAGE) with 1mm isotropic resolution was acquired along with the ASL scans.

A total of 28 healthy participants (14/14 females/males, age=38.6 ±19.4 years, range 19 to 87 years) underwent both single-delay and multi-delay ASL scans on a Siemens 3T Prisma scanner (Siemens Healthiness, Erlangen, Germany) using a 32-channel head coil. Test and retest scans were acquired with an average interval of 7.9±2.8 days. All subjects provided written informed consent according to a protocol approved by the Institutional Review Board of the University of Southern California. Subjects were screened for neurologic/psychiatric and major systemic disorders and were restrained from caffeine intake and smoking 3 hours before the MRI.

2.3. Quantification approach for 5-PLD ASL

Two quantification approaches were compared for 5-PLD ASL, including weighted delay (WD) and model fitting (MF). In the WD approach, a weighted sum of perfusion signals across multiple PLDs was calculated, where the weights (w) are the PLD values according to the multi-delay protocol. The weighted delay calculation is shown below, where N = 5 PLDs:

$$WD=\frac{\sum _{i=1}^N({\mathrm{\Delta }M}_{PLDi}·w(i\left)\right)}{\sum _{i=1}^N\left({\mathrm{\Delta }M}_{PLDi}\right)}$$ [2]

The resulting $\text{WD}$ was then used to derive the ATT for each voxel by applying a monotonic function that correlates $\text{WD}$ and ATT. This ATT was then used in Eq. [1] to estimate the CBF, ensuring that it fell within physiologically realistic limits, which were bounded from 0.1 to 200 ml/100g/min in our experiments. The $\text{WD}$ approach underscores computational simplicity and avoids iterative fitting, making it less noise-sensitive and computationally efficient.

The MF approach directly estimates CBF and ATT by fitting the multi-delay perfusion signals to an established kinetic model (Eq. [1]) using non-linear least-square curve fitting (Shao et al., 2018). The MF was performed in MATLAB using a non-linear least-square curve fitting toolbox that fits the perfusion signals. Lower and upper bounds of variables were set to 0 and 200 ml/100g/min for CBF, 500 and 2500 ms for ATT. The default value of CBF was 60 ml/100g/min and 1300 ms for ATT. MF captures the underlying kinetics of labeled blood water, including transit delays and signal decay, providing a robust and accurate estimate of brain hemodynamics. This method is particularly advantageous in capturing complex flow dynamics in regions with variable perfusion. However, compared with WD, MF requires more computational effort and is sensitive to initial parameters and noise.

2.4. Image Processing and Data Analysis

All image processing and data analysis were performed in MATLAB 2023a. Control and label images were corrected for rigid head motion and co-registered to M₀ using SPM12 (Wellcome Trust Centre for Neuroimaging, UCL, London, UK). Control and label images were subtracted to obtain perfusion images, and temporal fluctuations caused by residual motion and physiological noise were minimized using an algorithm based on principal component analysis (PCA) (Shao et al., 2017). To minimize motion artifacts, post-processing motion correction using SPM12 aligns the paired images and reduces the effects of gross patient motion. Image frames with framewise displacement (FD) greater than 1mm would be discarded. Grey matter (GM), white matter (WM), and CSF masks were segmented from MPRAGE images using SPM12, with 95% threshold to minimize the partial volume effect. ChP mask from all subjects was manually segmented from the atria of the lateral ventricles from the T₁-weighted images using ITK-SNAP (Yushkevich et al., 2006).

CBF values for a single-delay acquisition were derived from the single-compartment perfusion model (Alsop et al., 2015; Detre et al., 1992) based on the collected perfusion signal $\Delta M$ and M0, according to Eq. [1], under the assumption that PLD was longer than ATT. Both WD and MF approaches were used to estimate CBF and ATT values from multi-delay ASL signals. CBF and ATT values of the ChP, GM, and WM were extracted for each subject with the segmented masks. The ATT threshold was established at 2500 ms, and the CBF threshold was set below 200 mL/100g/min. Test-retest reliability of ASL measurements was assessed using the Intraclass Correlation Coefficient (ICC). According to the previous study, we designated ICC values < 0.4 as poor, 0.4 to 0.59 as fair, 0.6 to 0.74 as good, and ≥ 0.75 as excellent test and retest reliability (Senn, 2011).

2.5. ASL statistical analyses of choroid plexus

ATT histograms for the ChP, GM, and WM were generated using the MF approach. The proportion of ATT voxels in each region of interest (ROI) with values greater than 2000 ms was calculated. The total blood flow, F, to the choroid plexus (ChP) was calculated in a unit of mL/min using the function (Eisma et al., 2023):

$$F=\mathrm{\rho }\mathrm{v}f$$ [3]

where $\mathrm{\rho }$ represents the density of the choroid plexus in a unit of g/mL, v is the volume of the ChP segmented from the T1w 3D MRI in a unit of mL, and $f$ indicates the average perfusion rate of the ChP of each subject in a unit of mL/100g/min. To ensure consistency with the brain-blood partition coefficient used in perfusion calculations in the previous model, the density of the brain parenchyma (1.08 g/mL) was used as an approximation for the ChP density (Eisma et al., 2023).

The relationships between age and various parameters, including the CBF and ATT of the ChP, GM, and WM; as well as the volume and total blood flow of the ChP, were analyzed. Additionally, fitting curves were generated along with the corresponding equations, slopes, p-values, and R-squared values. Outliers were excluded when the values were higher than two times of the standard deviation (STD).

3. RESULTS

3.1. Test-retest repeatability of perfusion measurements in ChP and GM, WM

Figure 1 presents scatter plots of test and retest ChP CBF values for single-delay ASL, and CBF and ATT values for multi-delay ASL with WD or MF approach in 28 subjects. ICCs are shown in each subplot. All three methods show excellent repeatability for the ChP, with single-delay (ICC = 0.90) performing similarly to the MF and WD approaches (ICC = 0.92 and 0.91, respectively). However, the ICCs for ChP ATT are lower, while the MF approach (ICC = 0.71) performed better than the WD approach (ICC = 0.67).

Figure 1.

(a), (b), and (c) present scatter plots of the test and retest ChP CBF values across 28 subjects obtained from single-delay and 5-PLD ASL. The ICC values for single-delay, MF, and WD in GM are 0.90, 0.92, and 0.91, respectively. (d) and (e) show scatter plots of test and retest ChP ATT values from 5-PLD ASL. The ICC values are 0.71 (MF) and 0.67 (WD). The x-axis represents the data from visit 1, zand the y-axis represents the data from visit 2.

For comparison, Figure 2 shows scatter plots of test and retest CBF values in GM and WM for single-delay ASL, as well as CBF and ATT values for multi-delay ASL using the WD or MF approach, respectively. The ICCs of the corresponding GM and WM measurements are indicated in each subplot. The results show that there is little difference in ICCs between single- and multi-delay ASL approaches for CBF in GM (ICC = 0.84 for single-delay, 0.83 for MF, 0.82 for WD). Meanwhile, the ICC for WD-estimated CBF in WM is relatively lower (ICC = 0.63) compared to that of MF (ICC = 0.75), with single-delay showing similar reliability for CBF in WM (ICC = 0.77). The reliability of ATT estimated using MF was higher compared to WD in both GM (ICC = 0.89/0.81) and WM (ICC = 0.70/0.57).

Figure 2.

(a) (b)(c) present the scatter plot of test and retest CBF values across 28 subjects from single-delay and 5-PLD ASL. GM and WM results were shown as blue and red markers, respectively. ICC of single-delay, WD, and MF in GM are 0.84,0.83, 0.82; while ICC of single-delay, WD, and MF in WM is 0.77,0.75,0.63. Figure 2. (d) and (e) Scatter plot of test and retest ATT values across 28 subjects from 5-PLD ASL MF and WD approach. ICC of GM are 0.89 (MF) and 0.81 (WD); ICC of WM are 0.70 (MF) and 0.57 (WD). The values on the x-axis refer to the data from visit 1, and the values on the y-axis refer to the data from visit 2, respectively.

Table 1 summarizes test-retest CBF and ATT values (mean±STD) and their ICCs in the ChP, GM, and WM, respectively, consistent with the findings in Figures 1 and 2. The CBF of WM shows significantly lower values compared to GM and ChP using both single- and MF methods. Specifically, CBF values from the test/retest using MF are 43.25±15.00/42.88±13.30, 51.58 ± 10.68/53.87±13.15, and 35.88±6.34/39.19±8.22 ml/100g/min for ChP, GM, and WM, respectively. Post-hoc comparisons present significant differences between ChP and GM (p < 0.01), ChP and WM (p < 0.01), and GM and WM (p < 0.01), with Bonferroni correction applied (correction factor = 3). ANOVA results further supported these findings (p < 0.01). The estimated ATT in ChP (1218.02±215.0/1371.69±154.0 ms) was slightly longer than in GM (1172.95 ± 187.9/ 1157.36±213.3 ms). Post-hoc comparisons showed significant differences between ChP and GM (p < 0.01). ANOVA also indicated significant differences in ATT between these regions (p < 0.01). Additionally, the MF approach yielded higher ICCs for ATT values than those of WD in the ChP, GM, and WM. The table also shows that the MF approach yielded ATT values approximately 10% longer than those from the WD approach in WM, with slightly longer values observed in ChP and GM.

Table 1.

Mean CBF values, ICC of CBF, and ATT in ChP, GM, and WM were measured by single-delay ASL, 5-delay ASL (WD approach), and 5-delay ASL (MF approach) from test-retest scans of 28 subjects. (CBF unit: ml/100 g/min, ATT unit: ms)

	Single-PLD		Multiple PLD(MF)				Multiple PLD(WD)
	ICC	CBF Average	CBF ICC	CBF Average	ATT ICC	ATT Average	CBF ICC	CBF Average	ATT ICC	ATT Average
Choroid Plexus
Test	0.90	39.60+14.81	0.92	43.25±15.00	0.71	1218.02±215.0	0.91	41.12±15.80	0.67	1230.19±231.6
Retest		37.46±12.19		42.88±13.30		1371.69±154.0		46.54±12.98		1346.86±144.4
Gray Matter
Test	0.84	48.32±9.50	0.83	51.58±10.68	0.89	1172.95±187.9	0.82	51.07±9.33	0.81	1107.87± 177.7
Retest		50.47±11.18		53.87±13.15		1157.36±213.3		53.73±11.46		1093.51±181.6
White Matter
Test	0.77	34.08±5.20	0.75	35.88±6.34	0.70	1271.12±204.1	0.63	35.03±5.29	0.57	1164.74±195.6
Retest		35.36±7.01		39.19±8.22		1240.04±176.4		39.33±7.81		1142.70± 178.8

Furthermore, a mathematical model developed using 5000 Monte Carlo simulations was employed to evaluate the CBF and ATT values for both single- and multi-PLD approaches. The results demonstrate that the MF approach is robust when the ATT is longer than the PLD. When the ATT exceeds the PLD in single-delay ASL, the labeled blood may not have fully reached the imaging region, leading to an underestimation of CBF. This underestimation becomes more pronounced as the ATT increases beyond the PLD. Figure S2(A) illustrates this effect, showing CBF values obtained from 5000 Monte Carlo simulations where the ATT exceeds 2000 ms in both GM and WM. The simulations demonstrate that the single-delay approach underestimates CBF as ATT increases, aligning with in vivo experimental results. Further details are provided in the Supplementary Materials.

Figure 3 shows one slice of CBF and ATT maps from a representative subject (50 years old, male) from both test and retest scans. Figures 3A and 3C display the test and retest CBF maps, while Figures 3B and 3D present the ATT maps obtained using the MF and WD methods from test and retest scans. The zoomed-in panels show ChP perfusion and ATT overlaid on T1 images. All methods demonstrate relatively robust results in both test and retest scans. CBF and ATT maps of two more subjects are shown in supplementary Figure S4 and S5.

Figure 3.

CBF and ATT estimation maps (left), ChP axial view overlaid on a T1 image (middle), and ChP coronal view overlaid on a T1 image (right) show the test-retest results from a single subject (50 years old, male). (A) CBF maps of the whole brain and ChP from single-delay and 5-PLD ASL. The results of single-delay ASL are at the top; the results of the 5-PLD MF approach are in the middle; the results of the 5-PLD WD approach are at the bottom. The pink square indicates a zoomed-in view of the ChP. (B) ATT maps of the whole brain and ChP from 5-PLD ASL. The results of the 5-PLD MF approach are shown at the top; the results of the 5-PLD WD approach are shown at the bottom. (C) CBF retest estimation map of the whole brain and ChP. (D) ATT retest estimation map of the whole brain and ChP.

3.3. ATT distributions in ChP and GM, WM

Figure 4 displays the ATT histograms for all voxels in ChP, GM, and WM measured using the MF approach. Approximately 12.15% of ChP voxels have ATT values greater than 2000 ms, 5.35% of GM voxels, and 15.50% of WM voxels exhibit ATT values exceeding 2000 ms. Prolonged ATT can lead to a significant underestimation of CBF when the PLD is shorter than the ATT in single-delay ASL, particularly in ChP and WM.

Figure 4.

(a) to (c) present histograms of ATT voxels for the ChP, GM, and WM across 28 subjects using the MF approach. The red square highlights voxels with an ATT longer than 2000 ms. (a) Histogram of ATT voxels for the ChP of all subjects, showing that approximately 12.15% of voxels have an ATT longer than 2000 ms. (b) Histogram of ATT voxels for the GM of all subjects, showing that approximately 5.35% of voxels have an ATT longer than 2000 ms. (c) Histogram of ATT voxels for the WM of all subjects, showing that approximately 15.50% of voxels have an ATT longer than 2000 ms.

3.4. Choroid Plexus Perfusion and Volume Changes with Age

Figure 5 presents the relationship between perfusion and ATT of ChP and age, showing scatter plots and the fitted linear models of CBF using single-delay (a), MF (b), and WD (c) approaches, respectively. In general, the CBF values in the ChP demonstrate a negative relationship with age. The R² values for single-delay, MF, and WD are 0.52, 0.63, and 0.36, respectively (p<0.01 for single-delay, p = 0.04 for MF, and p = 0.06 for WD). In contrast, Figure 5 (d) and (e) show a positive trend of ATT with age, though it doesn’t reach significance. The p-values for MF and WD are 0.2 and 0.5, respectively.

Figure 5.

(a), (b), and (c) show the relationship between the CBF values and age of ChP for the 28 subjects from single-delay and 5-PLD ASL. The red line represents the linear regression fitting curve, and the corresponding equation for each fitting curve is labeled on the graph. Figure 5. (d) and (e) display scatter plots of the linear relationship between ATT and age across the 28 subjects from 5-PLD ASL, with the corresponding equation for each regression fitting curve labeled on the graph.

Figure 6 shows the changes in CBF and ATT in GM and WM with age in comparison to ChP. The fitted models for each relationship are shown in the subplots. Only the results from the MF approach are presented, as MF demonstrates robust performance in both CBF and ATT measurements. Figures 6 (a) and (b) indicate that the CBF of both GM and WM does not exhibit a significant relationship with age (p = 0.17 for GM and p = 0.68 for WM). Figures 6 (c) and (d) show the relationship between ATT in GM and WM with age, consistent with the ChP results, showing a positive linear relationship between ATT and age in both GM (p < 0.01) and WM (p = 0.019). The R² values for MF and WD approaches are 0.45 and 0.28, respectively.

Figure 6.

(a) and (b) show the linear relationships between the CBF values of GM and WM and age for the 28 subjects from the MF 5-PLD ASL. The red line represents the regression fitting curve, and the corresponding equation for each fitting curve is labeled on the graph. Figure 6. (c) and (d) display scatter plots of the linear relationships between the ATT of GM and WM and age across the 28 subjects from the MF 5-PLD ASL, with the corresponding equation for each fitting curve labeled on the graph.

Figure 7 illustrates the relationship between ChP volume, total blood flow, and age. Panel (a) displays a significant positive linear correlation between ChP volume and age (p < 0.01), indicating that the ChP volume increases with age. The mean ChP volume is (mean ± std) 1.72± 0.85 ml. Panel (b) shows a non-significant increase in ChP total blood flow with age (p = 0.41). The mean blood flow of ChP is 0.79±0.52 mL/min. The R² values for ChP volume and total blood flow with age are 0.47 and 0.37, respectively.

Figure 7.

(a) shows the relationship between ChP volume and age for the 28 subjects from the MF 5-PLD ASL. (b) shows the relationship between ChP blood flow and age for the 28 subjects from the MF 5-PLD ASL.

4. DISCUSSION

In this study, we evaluated the test-retest reliability and accuracy of the ChP perfusion measurement using single- and multi-delay ASL methods. We further studied variations of ChP perfusion and volume with age in a healthy cohort aged 19 to 87 years. The test-retest reliability of CBF measurements in the ChP was excellent and comparable to those of GM (all ICC > 0.8) and was higher than those of WM (ICC between 0.6 and 0.8). In general, the ICC values of CBF measurement using single-delay ASL were comparable with those using MF, while MF yielded higher ICCs compared to the WD approach.

Previous studies have demonstrated that ChP volume increases with age, reflecting potential compensatory mechanisms or structural and functional changes associated with aging (Alisch et al., 2021; Eisma et al., 2023) which aligns with the findings of our study. Eisma et al. also reported that total blood flow to the ChP increases with age, a trend that was not observed in our study, possibly due to the smaller sample size. In addition, age-related decreases in ChP CBF were observed in both single- and multiple-delay methods, demonstrating consistency with the findings of Alisch et al. (Alisch et al., 2021). Our results are highly consistent with these studies regarding ChP structural and perfusion changes with age. However, it is important to note that the previous study utilized single-delay ASL with a PLD of 2 seconds to measure ChP perfusion, which, based on our observations, may have led to an underestimation of CBF. Additionally, we observed increasing trends between ATT and age, further supporting the age-related decline in CBF and underscoring the potential for reduced cerebral blood flow with aging. Using multi-delay ASL with MF, we measured the average ChP CBF to be 43.07±14.18 ml/100g/min and the total blood flow of ChP 0.79±0.52 mL/min. These values are consistent with those reported by previous studies (Eisma et al., 2023; Zhao et al., 2020). Although there is no significant relationship between age and ChP blood flow, the trend aligns with findings in the literature. Our study also demonstrates the relationship between ATT and age, which is related to the structural and functional decline of the ChP with aging (Kadel et al., 1990).

This study also highlights the use of multi-delay ASL allowed for more accurate and reliable quantification of CBF and ATT, particularly in smaller brain regions like the ChP, compared to single-delay approaches. The observation shows a high chance that ATT will exceed 2000 ms in specific brain regions such as ChP and WM. Recent finding indicates that single-delay pCASL may underestimate GM CBF due to inadequate PLD, with estimated CBF approximately 10% lower than those obtained by multi-delay sequences (37.2 ± 8.1vs. 47.3 ±8.5 ml/100g/min) (Guo et al., 2018), which is consist with our finding (38.53±13.57 vs. 43.07±14.18 ml/100g/min). Previous studies have also shown that multi-delay protocols provide higher CBF values and more reliable ATT measurements due to their ability to cover a wider range of transit times and reduce estimation errors (van der Thiel et al., 2018; Yu et al., 2023; Zhang et al., 2021). Multi-delay ASL protocol can be an effective strategy to assess both ATT and CBF with reliability that is comparable to traditional single-delay methods. This is particularly valuable for comprehensive cerebrovascular evaluations, especially when simultaneous ATT and CBF quantification is desirable. The 5-PLD ASL protocol provides the appealing capability to concurrently quantify ATT and CBF. The MF approach was found to be preferable to the WD method due to its higher quantification accuracy and reliability in both simulations (see Supplementary Material) and in vivo experiments. When comparing WD and MF, overall, they exhibit similar average CBF values among ChP, GM, and WM, as well as comparable repeatability. However, the ATT in WM estimated by MF was longer than that by WD. Additionally, the ICCs of ATT in all three brain regions were higher with MF than with WD. Therefore, MF of multidelay ASL is recommended as the optimal technique, considering its accuracy and reliability in estimating CBF with longer ATT, such as in elderly subjects or brain regions like ChP and WM.

This study has a few limitations. Firstly, the study only focuses on healthy subjects, and the data size is relatively small, which may limit the generalizability of the results to populations with cerebrovascular pathology. The inherent variability of ATT and the volume change in ChP across different pathologies, such as NPH or hemicrania, (Bonney et al., 2022) could influence the efficacy of both ASL techniques and the ChP volume change before and after surgery (Johnson et al., 2021; Johnson et al., 2020), further research is needed to explore these effects. The behavior of ATT and CBF in the ChP for specific diseases remains largely unknown. Also, this study was only performed on a single 3T scanner, and results need to be further validated by cross-vendor studies in the future. In addition, the resolution of 2.5×2.5×3 mm³ may not be high enough for imaging ChP and may induce partial volume effects and variability in measured perfusion. Moreover, estimating CBF and ATT were performed using a single-compartment model, ignoring the arterial compartment or exchange of labeled blood from capillaries to tissue (Shao et al., 2019; Woods et al., 2024), which could introduce bias to the estimated CBF. Lastly, a comparison between multi-delay ASL with other time-encoded multi-delay acquisitions, such as the Hadamard encoding matrix, which also provides accurate ATT and CBF quantification utilizing more combinations of labeling duration and PLD (Wells et al., 2010), can be studied in the future.

CONCLUSION

In conclusion, the results of this study suggest that the structure and hemodynamic properties of ChP have a strong relationship with age; it also has relatively longer ATT compared with other brain regions. Multi-delay ASL with model fitting quantification provides both ATT and CBF measurements with high accuracy and reliability in the ChP and other brain regions.