Age and gender-related patterns of arterial transit time and cerebral blood flow in healthy adults

Abstract

Normal aging has been associated with increased arterial transit time (ATT) and reduced cerebral blood flow (CBF). However, age-related patterns of ATT and CBF and their relationship remain unclear. This is partly due to the lengthy scan times required for ATT measurements, which caused previous age-related CBF studies to not fully account for transit time. In this work, we aimed to elucidate age-related ATT and ATT-corrected CBF patterns. We examined 131 healthy subjects aged 19 to 82 years old using two pseudo-continuous arterial spin labeling (PCASL) MRI scans: one to measure fast low-resolution ATT maps with five post-labeling delays and the other to measure high-resolution perfusion-weighted maps with a single post-labeling delay. Both ATT and perfusion-weighed maps were applied with vessel suppression. We found that ATT increases with age in the frontal, temporoparietal, and occipital regions, with a more pronounced elongation in males compared to females in the middle temporal gyrus. ATT-corrected CBF decreases with age in several brain regions, including the anterior cingulate, insula, posterior cingulate, angular, precuneus, supramarginal, frontal, parietal, superior and middle temporal, occipital, and cerebellar regions, while remaining stable in the inferior temporal and subcortical regions. In contrast, without ATT correction, we detected artifactual decreases in the inferior temporal and precentral regions. These findings suggest that ATT provides valuable and independent insights into microvascular deficits and should be incorporated into CBF measurements for studies involving aging populations.

1. Introduction

Healthy aging is accompanied by various physiological changes, including modifications in the structure and function of the brain’s blood vessels, which in turn affect blood transit time and perfusion to brain tissue (Claassen et al., 2021; Erdő et al., 2017; Popescu et al., 2009). Research has consistently demonstrated diminished perfusion and elongated blood transit time at both global and local scales as people age (Wang et al., 2011; Zhang et al., 2018; Lorthois et al., 2011). These age-related hemodynamic changes can decrease the supply of oxygen and nutrients, potentially causing neuronal damage, loss of function (Bhatti et al., 2020; Daulatzai, 2017; Kidd, 2005), and cognitive decline (Abdelkarim et al., 2019; Zimmerman et al., 2021). If cognitive decline progresses, it may develop into dementia. It is critical for distinguishing between normal age-related changes and pathological alterations indicative of neurodegenerative disorders such as Alzheimer’s disease and vascular dementia (Bennett et al., 2009). By identifying biomarkers for age-related perfusion decline or blood transit time, researchers and clinicians can develop more accurate diagnostic tools and therapeutic interventions, as these parameters are closely linked to brain oxygen supply and glymphatic clearance. Moreover, such understanding can inform strategies to maintain or improve cerebral blood flow and blood transit time in older adults, potentially mitigating cognitive decline associated with aging and enhancing overall brain health.

Arterial spin labeling (ASL) magnetic resonance imaging (MRI) stands out as a noninvasive technique designed to measure cerebral blood flow (CBF) (van der Kleij and Petersen, 2018). ASL involves labeling of arterial blood (i.e., an endogenous tracer) without the need for contrast agents, making it an appealing option for a study of healthy populations (Grade et al., 2015; Haller et al., 2016; Huang et al., 2023; Lindner et al., 2023). ASL is conventionally acquired with a single post-labeling delay. However, arterial transit time (ATT), the time delay between labeling in the feeding arteries and the arrival of labeled blood in tissue, can have a large effect on the quantification of CBF maps. It has been shown that the ATT effect can be minimized if the post-labeling delay is optimized to allow the labeled blood to reach the tissue at the imaging time but they require an estimate of the ATT range for the study population (Alsop and Detre, 1996; Luh et al., 1999; Wong et al., 1998). However, there are large ATT variations across different brain regions and participants (Dai et al., 2017; Mutsaerts et al., 2015) and with aging, with gray matter ATT increasing with advanced age (Dai et al., 2017; Liu et al., 2012; Campbell and Beaulieu, 2006; Juttukonda et al., 2021). If one chooses an optimal ASL post-labeling delay, 2 s, recommended for the post-labeling delay for elderly population in the consensus review of ASL studies (Alsop et al., 2015), then the sensitivity of the ASL perfusion images will be reduced for young adults. If a post-labeling delay appropriate for young adults is used, the labeled blood might still be in the large arteries of elderly adults, leading to overestimation of CBF in regions with prominent arteries if their signals are not suppressed. Because of this, it is difficult to have an optimal post-labeling delay across the adult lifespan to study CBF changes. Therefore, to accurately measure CBF, ATT must be accounted for. We have shown degraded CBF accuracy in a presurgical elderly population when not correcting for the ATT effect (Dai et al., 2017).

The ATT map can be measured using different strategies, including acquiring images with multiple post-labeling delays, vessel-suppressing gradients, or both (Alsop and Detre, 1996; Wang et al., 2003; Gunther et al., 2001; Gonzalez-At et al., 2000; Petersen et al., 2006; Dai et al., 2012; Dai et al., 2013). However, all of them require lengthy scan time or sacrifice the signal-to-noise ratio of CBF measurement. As we suggested previously, a reduced-resolution ATT map combined with a high-resolution ASL acquisition with a single post-labeling delay can not only reduce the longer scan time but also minimize the SNR time penalty of the CBF measurement (Dai et al., 2012). In addition, if the labeled signals from large vessels (such as the circle of Willis) are not crushed, the measured ATT using ASL signals with multiple post-labeling delays can be underestimated in those areas with large vessels (Dai et al., 2017) because the measured ATT only reflects the arrival time at the large vessel rather than in the microvasculature and tissue. Therefore, accurate ATT measurement requires crushing the signal from large vessels by applying vessel suppression during the measurement sequence. Because ATT can vary strongly with age, correcting CBF measurements with ATT is crucial for understanding pure perfusion changes as a person ages.

Most age-related CBF studies adopted a single post-labeling delay approach (Chen et al., 2011; Zhang et al., 2018; Alisch et al., 2021; Leidhin et al., 2021; Biagi et al., 2007; Parkes et al., 2004). Four studies used relatively short post-labeling delays (≤ 1.6 s) and so their results may be confounded by the altered ATT in aging (Chen et al., 2011; Zhang et al., 2018; Biagi et al., 2007; Parkes et al., 2004). Two studies used relatively long post-labeling delay (1.8 s and 2.0 s) (Alisch et al., 2021; Leidhin et al., 2021), but the CBF changes with age and gender were investigated in the whole brain, gray matter, or large brain cortices. Two pulsed ASL studies acquired ASL images from multiple post-labeling delays and fitted ATT and CBF maps simultaneously, but these studies investigated either optimized ATT in the gray matter for the elderly group (Campbell and Beaulieu, 2006) or the age dependence of these hemodynamic parameters in the gray matter and two regions of interest (Liu et al., 2012). The Human Connectome Project-Aging (HCP-A) used multiple post-labeling delay pseudo-continuous ASL (PCASL) protocol and reported age-related CBF and ATT changes only in two gray matter regions and two white matter regions (Juttukonda et al., 2021) but not voxel-wise age-related CBF and ATT patterns. In addition, these ASL aging studies, including those using single post-labeling delay and multiple post-labeling delays, did not include the vessel suppression, and therefore the intravascular signals will underestimate the ATT and cause systematic error to CBF measurement.

Previous studies have reported age-related changes (Liu et al., 2012; Juttukonda et al., 2021; Chen et al., 2011; Zhang et al., 2018; Alisch et al., 2021; Leidhin et al., 2021; Biagi et al., 2007; Parkes et al., 2004; Mutsaerts et al., 2017) and gender-related differences (Liu et al., 2012; Juttukonda et al., 2021; Chen et al., 2011; Alisch et al., 2021; Leidhin et al., 2021) in global gray matter CBF and ATT. Consequently, gender could be a confounding factor when examining age-related hemodynamic changes. In this study, we aimed to assess age- and gender-related patterns on tissue/microvasculature-level ATT and true CBF free of ATT contamination in healthy adults spanning a broad age range. This approach enables us to distinguish between vascular abnormalities and perfusion deficits (Grade et al., 2015; Huang et al., 2023; Pelizzari et al., 2020). Additionally, to facilitate comparisons with existing literature, we also evaluated CBF without ATT correction. Specifically, ATT maps with vessel suppression were measured using our previously proposed fast and low-resolution multiple post-labeling delay PCASL method (Dai et al., 2012); a single post-labeling delay high-resolution PCASL with vessel suppression was performed to measure CBF both with the standard method (CBF without ATT correction) and accounting for the low-resolution ATT map (CBF with ATT correction). In this study, two PCASL sequences are used instead of a single high-resolution PCASL sequence with multiple post-labeling delays, to reduce the acquisition time and enable most robust measurements of ATT and CBF.

2. Methods

2.1. Study population

The investigation was carried out at the Cornell University MR Facility, utilizing a GE 3T MR 750 system, using a 32-channel receive-only phased-array head coil, a maximum gradient amplitude of 50 mT/m and a maximum slew rate of 200 T/m/s. A cohort of 131 participants (age range: 19–82 years, 89 female), comprising 44 young adults (range: 19–39 years; mean age: 27.07 ± 5.44 years, 31 female), 68 middle-aged adults (range: 40–69 years; mean age: 56.53 ± 9.41 years, 46 female), and 19 older adults (range: ≥70 years; mean age: 74.05 ± 3.57 years, 12 female), which were enrolled in the study. There was no significant difference (p = 0.89) in age between male (age = 48.86 ± 18.88 years) and female (age = 49.33 ± 19.39 years) participants in this study. The study was reviewed by the Cornell Institutional Review Board and all participants provided informed consent.

2.2. MRI acquisition

Each participant underwent a comprehensive imaging protocol, including the acquisition of T1-weighted Magnetization-Prepared Rapid Gradient-Echo (MPRAGE) images, which provide high-resolution anatomical details. Additionally, a low-resolution ATT map was obtained by acquiring a Pseudo-Continuous Arterial Spin Labeling (PCASL) difference images with a TR of 5 s, TE of 10.80 ms, labeling duration of 2 s, and five sequential post-labeling delays of 0.7, 1.3, 1.9, 2.5, and 3 s as that in our previous protocol (Dai et al., 2012). The study also implemented a single post-labeling delay PCASL sequence with a TR of 5 s, TE of 14.89 ms, labeling duration of 2 s, and post-labeling delay of 1.8 s, as in our earlier protocol (Dai et al., 2016), capturing 30 ASL image volumes and a reference M₀ image to allow for quantitative measurements of perfusion. The M0 image has the same acquisition parameters as the single-delay PCASL images. Both ATT and perfusion images were acquired with a 3D stacks of spiral RARE imaging sequence (reconstruction matrix size: 128 * 128; slice thickness: 4 mm, number of slices: 44, field of view (FOV): 24 cm, receiver bandwidth (rBW): 125 kHz). ATT and perfusion images were performed with one spiral interleave and 2 spiral interleaves, which produced an effective in-plane spatial resolution of 12.08 mm and 5.44 mm, respectively. Background suppression (optimal pulse timing determined using the algorithms in (Maleki et al., 2012; Dai et al., 2011)) and vessel suppression (suppressing vessels with a vessel velocity of 1cm/s and above) were also applied in both ATT (Dai et al., 2012) and perfusion (Dai et al., 2016) measurements as before. Specifically, to minimize contamination from motion and physiological noise, background suppression pulses suppress the gray and white matter to 0.3 % of their full magnetization signals for the ATT and perfusion measurements.

2.3. Image processing

The ASL difference (control – label) image time series were reconstructed using our custom reconstruction algorithm (Dai et al., 2016; Zhao et al., 2017). The ASL image realignment and co-registration were performed using SPM12 (Wellcome Trust Centre for Neuroimaging). The first 3D image of high-resolution ASL time series was removed for increased stability. The remaining 29 ASL time series images were realigned to correct for head motion and the mean of the motion-corrected ASL difference images was generated. The mean ASL difference image was co-registered with the M0 image. The low-resolution five post-labeling delay ASL images were also realigned. ATT maps, essential for accurately quantifying CBF, were derived voxel-by-voxel for each individual in this study. The process applied a noniterative algorithm to the realigned low-resolution ASL images encompassing five post-labeling delays (Dai et al., 2012). This approach ensured a high level of precision in capturing arterial transit. To further refine the analysis, CBF maps for each individual were computed with a standard quantification method without (CBF wo. ATT correction) and with ATT correction (CBF w. ATT correction) (Dai et al., 2012) using the one-compartment CBF kinetic model (Alsop and Detre, 1996; Buxton et al., 1998; Wang et al., 2002), Eq. (Claassen et al., 2021). The For CBF wo. ATT correction, the applied post-labeling delay of 1.8 s is assumed to be the same as the ATT in the entire brain. Given this assumption, CBF measurement errors are expected to be minimal for the voxels with ATT shorter than 1.8 s as far as the T1 of tissue and blood are similar (Alsop and Detre, 1996; Luh et al., 1999; Wong et al., 1998). For CBF w. ATT correction, the CBF map was calculated with the derived low-resolution ATT map using the following Equation.

$$CBF=\frac{\lambda \cdot \Delta M\cdot \text{exp}\left(\frac{\delta }{T_{1a}}\right)}{\left(2\cdot M_t^0\cdot \alpha \cdot T_{1t}\cdot \left[\text{exp}\left(-\frac{\max (w-\delta ,0)}{T_{1t}}\right)-\text{exp}\left(-\frac{\max (\tau +w-\delta ,0)}{T_{1t}}\right)\right]\right)}$$

where $\Delta M$ is the mean ASL difference image, $T_{1a}$ and $T_{1t}$ are the longitudinal relaxation times of blood and tissue (assumed to be 1.66 s (Lu et al., 2004) and 1.5 s (Ethofer et al., 2003), respectively), $M_t^0$ is the fully relaxed equilibrium magnetization of brain tissue, $\alpha$ is the labeling efficiency (assumed to be 0.8), $\lambda$ is the tissue-to-blood partition coefficient of water (assumed to be 0.9 ml of blood/g of tissue), $\delta$ is the arterial transit time, $\tau$ is the labeling duration, and $w$ is the post-labeling delay. The preprocessing pipeline for generating the ATT map, CBF map wo. ATT correction, and CBF map w. ATT correction is shown in Fig. 1a.

Fig. 1.

ASL preprocessing pipeline. (a) Generation of the ATT map, CBF map wo. ATT correction, and CBF map w. ATT correction. (b) Normalization of these maps into the standard MNI space.

To enhance the accuracy of inter-subject voxel-wise alignment of CBF maps and ATT maps with the standard MNI space, T1-weighted images were segmented into gray matter, white matter, CSF, and nonbrain images using SPM12. The first transformation was performed using SPM12 co-registration with the gray matter image as the reference image and the mean ASL difference image as the moving image. This initial step transforms the individual CBF/ATT maps to align with the corresponding gray matter image.

Building upon this foundation, the second transformation was calculated from the individual T1 images to the standard MNI space by taking gray matter/white matter templates in the MNI space as the reference images and individual gray matter/white matter images as the moving images. The second transformation step was performed using the Diffeomorphic Anatomical Registration Through Exponential Lie Algebra (DARTEL) method, a specialized toolbox within SPM12. DARTEL is adopted in this step because the nonlinear registration method enables precise inter-subject voxel-wise alignment, making it particularly effective in studies involving aging populations (Bhatti et al., 2020; Abdelkarim et al., 2019; Zimmerman et al., 2021; Bennett et al., 2009). The DARTEL procedure starts with Run DARTEL (create Templates), which estimate the deformations that best aligns the gray matter and white matter images with those in the template. Gray matter and white probability maps was first segmented from T1-weighted images and transformed into the MNI space. The initial template was calculated based on the average of transformed gray matter and white matter images from all subjects. The first iteration of registration warped each subject’s image to the initial template using a diffeomorphic transformation. Diffeomorphisms preserve the topology of the brain, which ensures that the images remain anatomically plausible during transformation. The second template can be generated by the average of the registered images. The new template is then used to refine the registration further. The process repeats iteratively for six times, with each round of transformations leading to a more precise alignment. The Run DARTEL (create Templates) step resulted in six templates and a flow field map that contains nonlinear transformation from each subject’s T1 image to the DARTEL template. The next DARTEL procedure is Normalized to MNI Space. The flow field map from each subject was applied to the subject’s CBF maps and ATT maps, which transformed those maps to the MNI space.

After the DARTEL procedure, ATT maps, CBF maps wo. ATT correction, and CBF maps w. ATT correction were transformed to the standard MNI space. To maintain the integrity of these maps, we preserve the concentrations during the resampling process. The CBF maps were smoothed using a 6 mm Gaussian kernel. This DARTEL step enhances the precision of spatial normalization, mitigating the impact of small anatomical variations and optimizing the alignment of CBF maps across the study cohort. The preprocessing pipeline for normalizing the ATT map, CBF map wo. ATT correction, and CBF map w. ATT correction is shown in Fig. 1b.

2.4. Statistical analysis

Differences in demographic data, including age, sex, education, handedness, race, and hypertension, among young, middle-aged, and older adults were compared using either ANOVA tests for continuous variables or χ² tests for categorical variables. Global hemodynamic variables, including ATT, CBF wo. ATT correction, and CBF w. ATT correction, for each subject in the whole brain or gray matter, were calculated as the mean value within each corresponding mask. The whole brain mask and the gray matter mask were generated using the SPM12’s standard brain mask and gray matter probability map with a 50 % threshold. Comparisons of each global hemodynamic variable among young, middle-aged, and older adults and association of each hemodynamic variable with age (or gender) were performed using ANOVA and simple linear regression models. We also performed multiple linear regression models by considering (1) age and gender as covariates, (2) age, gender, and age-gender interaction as covariates, (3) age, gender, and race as covariates, and (4) age, gender, and hypertension as covariates. Because of the small number of participants in Black/Native American/Latino, we combined all three such that race was considered as a categorical variable (white, Asian, other). To explore regional variations in ATT and CBF across distinct age groups encompassing young, middle-aged, and older adults, robust voxel-wise statistical comparisons were undertaken. These analyses were conducted employing the SPM12 One-way Analysis of Variance (ANOVA) framework, with gender considered as a covariate, accounting for potential confounding effects associated with unequal gender in each age category. The utilization of ANOVA allowed for the examination of age-related differences in ATT and CBF while examining and controlling for gender influences. To investigate the association of the changes of hemodynamic variables with age, the ATT maps, CBF maps wo. ATT correction, and CBF maps w. ATT correction were modeled using SPM12 on a voxel basis via multiple linear regression with age and gender as covariates. Both CBF maps w. ATT correction and globally scaled CBF maps wo. ATT correction were shown to decrease variability in an elderly group (Dai et al., 2017). Applying global scaling to CBF maps wo. ATT correction may yield comparable group-comparison results across young, middle-aged, and older adults as those obtained using CBF maps w. ATT correction. If this is the case in this study, acquiring ATT maps may not be essential. Globally scaled CBF maps wo. ATT correction, hereafter referred to as relative CBF maps, were generated by dividing CBF maps (without ATT correction) by their global mean (averaged across the entire brain mask) and then multiplying by a standard perfusion value (for instance, 60 ml/100g/min). This process ensures that each subject has the same global CBF mean, thereby reducing global CBF variance across subjects. Consequently, we conducted statistical analyses on relative CBF maps to directly compare them with CBF maps w. ATT correction. In addition, we also explored three voxel-wise multiple linear regression models each by including one additional covariate (age-gender interaction, race, and hypertension) besides age and gender.

Voxel-level p-value thresholds of 0.001 were used for ATT and CBF models. Either FWE-corrected voxel-level or FWE-corrected cluster-level p-value threshold of 0.05 were considered as significant, which was used to guard against false positives from multiple comparisons.

3. Results

Table 1 summarizes participants’ demographic data for young, middle-aged, and older adults. No differences in gender, education, and handedness were found between the three groups. The middle-aged and older adults’ groups had a lower prevalence of Asian people (p = 0.00039) and a higher prevalence of hypertension (p = 0.007).

Table 1.

Demographics of all subjects from three age groups.

	Young (18–39 y) N = 44	Middle-aged (40–69 y) N = 68	Older (≥70 y) N = 19	P values between 3 groups
Age, y	27.07 ± 5.44	56.53 ± 9.41	74.05 ± 3.57	0.00052
Sex, F/M	31/13	46/22	12/7	0.33
Education, y	17.54 ± 3.32	16.88 ± 4.24	17.32 ± 2.91	0.65
Handedness, R/L	41/3	64/4	17/2	0.50
Race, Asian/White/Black/Native American/Latino	13/26/5/0/0	2/45/15/3/3	0/18/1/0/0	0.0001
Hypertension*, Y/N	3/40 (N = 30)	22/33 (N = 55)	11/5 (N = 16)	0.0001

^*Systolic/diastolic blood pressure higher than 140/90 was considered as hypertensive.

3.1. Association of global ATT and CBF with age and gender

Before accounting for gender, global ATT was 1.39 ± 0.23 s, 1.52 ± 0.31 s, and 1.61 ± 0.21 s in gray matter, and 1.41 ± 0.22 s, 1.54 ± 0.30 s, and 1.63 ± 0.21 s in whole brain for young, middle-aged, and older adults, respectively. Global ATT was significantly longer in gray matter and whole brain for both middle-aged and older adults. After accounting for gender, significantly longer global ATT was still observed in gray matter (Fig. 2c) and whole brain (Fig. 2d) for both middle-aged and older adults; older age was correlated with longer global ATT in both gray matter (r = 0.37, p < 0.0001, Fig. 2a) and whole brain (r = 0.36, p < 0.0001, Fig. 2b). Before accounting for age, females exhibited shorter global ATT, but a similar age-related elongation rate compared to males (Supplementary Fig. S1). No significant difference in global ATT and its elongation rate with age was observed between male and female in both gray matter and whole brain after adjusting for age effect.

Fig. 2.

Changes of global ATT and CBF with age after adjusting for gender. ATT values were significantly increased with aging in both (a) gray matter and (b) whole brain. ATT values for young, middle-aged, and older adults were compared in (c) gray matter and (d) whole brain, respectively. CBF wo. ATT correction values were significantly decreased with aging in (e) gray matter and (f) whole brain. CBF wo. ATT correction values for young, middle-aged, and older adults were compared in (g) gray matter and (h) whole brain, respectively. CBF w. ATT correction values were significantly decreased with aging in (i) gray matter and (j) whole brain. CBF w. ATT correction values for young, middle-aged, and older adults were compared in (k) gray matter and (l) whole brain, respectively. No significant differences in ATT, CBF wo. ATT correction, and CBF w. ATT correction were observed between the middle-aged and older adults. The bars stand for standard deviations. * stands for 0.01 < P < 0.05, ** stands for 0.001 < P < 0.01, *** stands for P < 0.001, compared to the young group.

Before accounting for gender, global CBF wo. ATT correction was 68.05 ± 9.88 ml/100g/min, 58.24 ± 10.41 ml/100g/min, and 55.70 ± 11.74 ml/100g/min in gray matter, and 62.24 ± 8.56 ml/100g/min, 54.39 ± 8.77 ml/100g/min, and 52.36 ± 10.27 ml/100g/min in whole brain for young, middle-aged, and older adults, respectively. After accounting for gender effect, significantly reduced global CBF wo. ATT correction was still observed in gray matter (Fig. 2g) and whole brain (Fig. 2h) for both middle-aged and older adults; older age was correlated with decreased global CBF wo. ATT correction in both gray matter (r = −0.50, p < 0.0001, Fig. 2e) and whole brain (r = −0.48, p < 0.0001, Fig. 2f). Before accounting for age, females exhibited increased global CBF wo. ATT correction but similar age-related reduction rate compared to males (Supplementary Fig. S1). After adjusting for age effect, global CBF wo. ATT correction was significantly associated with gender, and females have higher global CBF than males in gray matter (p = 0.00081) and whole brain (p = 0.00047); no significant difference in the reduction rate of global CBF wo. ATT correction with age was observed between males and females in both gray matter and whole brain.

Before accounting for gender, global CBF w. ATT correction was 69.64 ± 11.02 ml/100g/min, 61.71 ± 10.43 ml/100g/min, and 61.52 ± 13.16 ml/100g/min in gray matter, and 64.03 ± 9.67 ml/100g/min, 58.43 ± 10.10 ml/100g/min, and 58.38 ± 11.85 ml/100g/min in whole brain for young, middle-aged, and older adults, respectively. After adjusting for gender, significantly reduced global CBF w. ATT correction was observed in gray matter for both middle-aged and older adults (Fig. 2k) and whole brain only for middle-aged adults (Fig. 2l); older age was correlated with global CBF w. ATT correction in gray matter (r = −0.36, p < 0.0001, Fig. 2i) and whole brain (r = −0.29, p = 0.0009, Fig. 2j). Before accounting for age, global CBF w. ATT correction was not significantly different between males and females; global CBF w. ATT correction reduced with increasing age in females but not in males (Supplementary Fig. S1). To determine whether the negative relationship between global CBF w. ATT correction and age in females was driven by their larger sample size (89 females vs. 42 males), we randomly selected 42 females from the group of 89. The negative relationship remained significant (Supplementary Fig. S2). Further research with a larger sample of males is needed to explore the relationship. This discrepancy may be influenced by the limited sample size in males, given that there were twice as many female participants as male participants. No significant difference in global CBF w. ATT correction, and difference in its reduction rate with age between males and females was observed in both gray matter and whole brain after adjusting for the age effect. The detailed beta and p values are listed in Supplementary Table S2. After accounting for age and gender, race and hypertension were found not associated with global ATT, global CBF wo. ATT correction, and global CBF wo. ATT correction.

3.2. Regional ATT changes with age and gender

In a comparison to young adults, middle-aged adults exhibited increased ATT values mainly in the rectus extending to middle, superior, inferior, and medial frontal regions (Fig. 3a), while older adults showed ATT increases in a wider area, mainly anterior cingulate extending to middle, superior, and medial frontal, postcentral, angular and supramarginal extending to inferior and superior parietal, and middle occipital regions (Fig. 3b). The regression analysis showed that ATT values increased with age in the anterior cingulate and rectus extending to middle, superior, inferior, and medial frontal, angular and supramarginal extending to inferior and superior parietal, middle, inferior, and superior temporal, middle, superior and inferior occipital regions (Fig. 3c) and ATT values were longer in male than in female in the middle temporal region (Fig. 3d). The clusters’ statistics is summarized in Supplementary Table 1. Additional analyses also revealed no significant association of hypertension and race with ATT.

Fig. 3.

Regions overlaid on a standard brain template in which ATT values were significantly longer in the (a) rectus extending to middle, superior, inferior, and medial frontal regions in middle-aged adults and (b) anterior cingulate extending to middle, superior, and medial frontal, postcentral, angular and supramarginal extending to inferior and superior parietal, and middle occipital regions in older adults compared to young adults; ATT values were positively associated with age in the (c) anterior cingulate and rectus extending to middle, superior, inferior, and medial frontal, angular and supramarginal extending to inferior and superior parietal, middle, inferior, and superior temporal, middle, superior and inferior occipital regions; ATT values were longer in male than in female in the (d) middle temporal regions. The color bar shows the range of t values.

3.3. Changes in regional CBF wo. ATT correction with age and gender

A significant reduction in CBF wo. ATT correction across various brain regions was observed when comparing middle-aged and older adults to young adults. Specifically, middle-aged adults exhibited decreased CBF mainly in the anterior cingulate, middle cingulate, insula, Rolandic, supplementary motor, and precentral extending to superior, middle, inferior and medial frontal, posterior cingulate, precuneus, supramarginal, angular and postcentral extending to inferior and superior parietal, Heschl extending to middle, superior, and inferior temporal, cuneus, lingual, calcarine, and fusiform extending to occipital, and cerebellar regions (Fig. 4a). In older adults, the reduction was observed in less extensive area, mainly including the insula extending to middle, superior, and inferior frontal, posterior cingulate, precuneus, angular, supramarginal extending to inferior and superior parietal, Heschl extending to middle and superior temporal, cuneus and calcarine extending to occipital, and cerebellar regions (Fig. 4b). Multiple linear regression analysis revealed that age was negatively associated with CBF in the anterior cingulate, middle cingulate, insula, and precentral extending to middle, superior, inferior, and medial frontal, posterior cingulate, precuneus, supramarginal, postcentral and angular extending to superior and inferior parietal, Heschl extending to middle, superior, and inferior temporal, cuneus, calcarine, and lingual extending to middle, superior and inferior occipital, and cerebellar regions (Fig. 4c) and CBF values were higher in female than in male in the superior and middle frontal, angular extending to inferior parietal, middle temporal, and calcarine extending to middle, superior and inferior occipital regions (Fig. 4d).

Fig. 4.

Regions overlaid on a standard brain template in which CBF values wo. ATT correction were significantly reduced in the (a) anterior cingulate, middle cingulate, insula, Rolandic, supplementary motor, and precentral extending to superior, middle, inferior and medial frontal, posterior cingulate, precuneus, supramarginal, angular and postcentral extending to inferior and superior parietal, Heschl extending to middle, superior, and inferior temporal, cuneus, lingual, calcarine, and fusiform extending to occipital, and cerebellar regions in middle-aged adults and (b) insula extending to middle, superior, and inferior frontal, posterior cingulate, precuneus, angular, supramarginal extending to inferior and superior parietal, Heschl extending to middle and superior temporal, cuneus and calcarine extending to occipital, and cerebellar regions in older adults compared to young adults; CBF values wo. ATT correction were negatively associated with age in the (c) anterior cingulate, middle cingulate, insula, and precentral extending to middle, superior, inferior, and medial frontal, posterior cingulate, precuneus, supramarginal, postcentral and angular extending to superior and inferior parietal, Heschl extending to middle, superior, and inferior temporal, cuneus, calcarine, and lingual extending to middle, superior and inferior occipital, and cerebellar regions; CBF values wo. ATT correction were higher in females than in males in the (d) superior and middle frontal, angular extending to inferior parietal, middle temporal, and calcarine extending to middle, superior and inferior occipital regions. Color bars show the range of t values.

To investigate whether global scaling can serve the same role as the ATT correction, we calculated the relative CBF map for each subject and found both higher and lower relative CBF when comparing among different age groups. More specifically, in middle-aged adults, significant lower relative CBF were observed in the anterior cingulate extending to middle, superior and medial frontal, angular, supramarginal, and precuneus extending to inferior and superior parietal, cuneus extending to superior occipital, and cerebellar regions (Fig. 5a), while in older adults, lower relative CBF were observed in the anterior cingulate extending to middle frontal, angular, precuneus, and supramarginal extending to inferior and superior parietal, cuneus extending to middle and superior occipital regions (Fig. 5b). Conversely, middle-aged exhibited higher relative CBF in the putamen, thalamus, pallidum, caudate, hippocampus, and amygdala regions (Fig. 5c); while older adults exhibited higher relative CBF in the putamen, thalamus, pallidum, caudate, hippocampus, and cerebellar regions compared to young adults (Fig. 5d). Additionally, compared to middle-aged adults, older adults exhibited higher relative CBF values in the superior frontal and supplementary motor regions (Fig. 5e). Following proportional scaling, age was shown with both negative and positive association with relative CBF. Specifically, age was negatively associated with relative CBF in the anterior cingulate and insula extending to middle, superior, inferior, and medial frontal, angular, supramarginal, and precuneus extending to superior and inferior parietal, middle temporal, cuneus and calcarine extending to middle and superior occipital, and cerebellar regions (Fig. 5f). By contrast, age was positively associated with relative CBF in the middle cingulate, supplementary motor, olfactory, putamen, thalamus, pallidum, caudate, amygdala, hippocampus/ parahippocampus and cerebellar regions (Fig. 5g). The clusters’ statistics are summarized in Supplementary Table 1. Additional analyses also revealed no significant association of hypertension and race with CBF wo. ATT correction.

Fig. 5.

Regions overlaid on a standard brain template in which relative CBF values were significantly reduced in the (a) anterior cingulate extending to middle, superior and medial frontal, angular, supramarginal, and precuneus extending to inferior and superior parietal, cuneus extending to superior occipital, and cerebellar regions in middle-aged adults, in the (b) anterior cingulate extending to middle frontal, angular, precuneus, and supramarginal extending to inferior and superior parietal, cuneus extending to middle and superior occipital regions in older adults compared to young adults, and significantly increased in the (c) putamen, thalamus, pallidum, caudate, hippocampus, and amygdala regions in middle-aged adults, (d) putamen, thalamus, pallidum, caudate, hippocampus, and cerebellar regions in older adults compared to young adults, and (e) superior frontal and supplementary motor regions in older results compared to middle adults; and global scaled CBF values were negatively associated with age in the (f) anterior cingulate and insula extending to middle, superior, inferior, and medial frontal, angular, supramarginal, and precuneus extending to superior and inferior parietal, middle temporal, cuneus and calcarine extending to middle and superior occipital, and cerebellar regions, and CBF values were positively associated with age in the (g) middle cingulate, supplementary motor, olfactory, putamen, thalamus, pallidum, caudate, amygdala, hippocampus/parahippocampus and cerebellar regions. Color bars show the range of t values.

3.4. Changes in CBF w. ATT correction with age and gender

Significant decreases in CBF w. ATT correction were observed in the anterior cingulate, middle cingulate, insula, and Rolandic extending to superior, middle, inferior, and medial frontal, angular, supramarginal, posterior cingulate, and precuneus extending to inferior and superior parietal, Heschl extending to superior temporal, cuneus, lingual, and calcarine extending to middle and superior occipital, and cerebellar regions for middle-aged adults (Fig. 6a), and in the anterior cingulate, middle cingulate and insula extending to superior, middle, inferior, and medial frontal, posterior cingulate, precuneus, angular, and supramarginal extending to inferior and superior parietal, Heschl extending to superior and middle temporal, cuneus and calcarine extending to middle, superior, and inferior occipital regions for older adults (Fig. 6b), compared to young adults. Multiple linear regression analysis revealed that age was negatively associated with CBF w. ATT in the anterior cingulate and insula extending to superior, inferior, and medial frontal, posterior cingulate, precuneus, angular, and supramarginal extending to inferior and superior parietal, Heschl extending to superior and middle temporal, cuneus and calcarine extending to middle, superior and inferior occipital, and cerebellar regions (Fig. 6c). No significant difference in CBF w. ATT correction was observed between females and males. The clusters’ statistics are summarized in Supplementary Table 1. Additional analyses also revealed no significant association of hypertension and race with CBF w. ATT correction.

Fig. 6.

Regions overlaid on a standard brain template in which CBF w. ATT correction was significantly reduced in the (a) anterior cingulate, middle cingulate, insula, and Rolandic extending to superior, middle, inferior, and medial frontal, angular, supramarginal, posterior cingulate, and precuneus extending to inferior and superior parietal, Heschl extending to superior temporal, cuneus, lingual, and calcarine extending to middle and superior occipital, and cerebellar regions in the middle-aged adults and (b) anterior cingulate, middle cingulate and insula extending to superior, middle, inferior, and medial frontal, posterior cingulate, precuneus, angular, and supramarginal extending to inferior and superior parietal, Heschl extending to superior and middle temporal, cuneus and calcarine extending to middle, superior, and inferior occipital regions in the older adults compared to the young adults; and CBF w. ATT correction was negatively associated with age in the (c) anterior cingulate and insula extending to superior, inferior, and medial frontal, posterior cingulate, precuneus, angular, and supramarginal extending to inferior and superior parietal, Heschl extending to superior and middle temporal, cuneus and calcarine extending to middle, superior and inferior occipital, and cerebellar regions. Color bars show the range of t values.

4. Discussion

The study employed two PCASL sequences: one to measure low-resolution ATT for more robust estimation; the other to capture high-resolution single post-labeling delay images to mitigate the loss of sensitivity in perfusion measurements. This approach significantly reduced acquisition time compared to using a single high-resolution PCASL sequence with multiple post-labeling delays. By separately analyzing age-related changes in ATT and CBF, the study offers a clearer, independent understanding of these hemodynamic properties as they change with aging. We found that global ATTs in gray matter and whole brain are directly associated with increasing age, which agrees well with previous studies using multiple post-labeling delays or comparing crushed versus non-crushed ASL signals (Dai et al., 2017; Liu et al., 2012; Campbell and Beaulieu, 2006; Mutsaerts et al., 2017). Our findings of increased global ATTs with aging may be related to increased vessel tortuosity in large carotid arteries and arterioles and increased damage of arteriole walls (Farkas and Luiten, 2001; Hutchins et al., 1996), The observed rates of ATT increase in gray matter and whole brain are 0.41 % and 0.38 % per year from the intercept values of 1.30 s and 1.34 s, respectively. The rate of ATT increase in gray matter is larger than the arterial-arteriole transit time increase of 0.175 % per year (2.1 % per decade) (Liu et al., 2012). This is expected because our study measured the ATT when the labeled blood flows into very small arteries (with a flow velocity 1cm/s or less), while most studies measured ATT to relatively larger arteries (no velocity cutoff or with a weaker velocity cutoff of 5cm/s). Therefore, our results add extra evidence that the influence of age on ATT remains significant even when the labeled blood reaches the small arteries. Specifically, the significantly longer ATT values observed in the orbitofrontal and rectus regions among middle-aged adults when compared to young adults suggest delayed blood supply in the regions involved in emotional regulation (Garcia-Cabezas and Barbas, 2017; Rolls, 2023). These regions have been reported with volume reduction in elderly patients with major depression (Ballmaier et al., 2004). This is consistent with literature suggesting that the prefrontal cortex is particularly vulnerable to the effects of aging (Raz et al., 2005; Fjell et al., 2009), and we added additional evidence that the prefrontal vulnerability is in the medial and ventral sectors and may be caused by its elongated blood circulation due to aging.

In contrast, the pattern observed in older adults—where longer ATT values extend to the default mode network (except the posterior cingulate cortex) and frontoparietal network regions—indicates a broader, age-related delayed blood supply. Prefrontal volume reduction (Gunning-Dixon and Raz, 2003), reduced integrity in the frontoparietal network (Veldsman et al., 2020), and poor white matter microvasculature in the default mode network and frontoparietal network (Brown et al., 2019) have been associated with age-related effects on executive function. Therefore, our findings support that elongated ATT in these brain networks may contribute to poorer executive performance with advanced aging (De Luca et al., 2003).

We found that global CBFs with and without ATT correction in gray matter and whole brain are inversely associated with increasing age. We observed the distinct correlations of age with true global CBF (i.e., CBF with ATT correction, rho = −0.36) and global ATT (rho = 0.37), suggesting separate effects of CBF and ATT. CBF and ATT have been independently associated with white matter hyperintensity, a feature of small vessel disease (Zhang et al., 2022). Our findings support that both ATT and CBF can provide an independent understanding of changes in brain hemodynamics with aging. A stronger correlation with age was observed with global CBF without ATT correction (rho = −0.50). A stronger correlation of global CBF without ATT correction with age is expected because it contains the combined effect of lower absolute CBF and longer ATT in older adults. The observed decline rates of global CBF in gray matter and whole brain are 0.41 % and 0.37 % per year (from the intercept values of 70.45 ml/100g/min and 63.92 ml/100g/min) when quantified without ATT correction, but 0.31 % and 0.26 % per year (from the intercept values of 68.94 ml/100g/min and 63.34 ml/100g/min) when quantified with ATT correction. This indicates that correcting for ATT is essential to accurately understand the decrease in true CBF with age. Our observed age-dependent decline in global CBF (without ATT correction) has been frequently reported with ASL (Liu et al., 2012; Mutsaerts et al., 2017; Asllani et al., 2009; Ambarki et al., 2015; Wagner et al., 2012; Zhang et al., 2017; Taneja et al., 2020) and agrees well with the prior reported decline rate of 0.38 % - 0.45 % per year (Chen et al., 2011; Zhang et al., 2018; Parkes et al., 2004). However, global CBF with ATT correction showed a smaller decline rate (0.31 % per year) compared to the reported 0.47 % decline per year for the ATT-corrected CBF (Juttukonda et al., 2021). The discrepancy may be caused by not suppressing large vessels in prior study’s ASL acquisitions. Further studies should assess how vessel suppression contributes to age-related CBF decline.

The ATT-corrected CBF reduction was observed with advanced age widely across multiple brain regions, including in the anterior cingulate and insula extending to superior, inferior, and medial frontal, posterior cingulate, precuneus, angular, and supramarginal extending to inferior and superior parietal, Heschl extending to superior and middle temporal, cuneus and calcarine extending to middle, superior and inferior occipital, and cerebellar regions, suggesting a global pattern of diminished perfusion that may underlie age-related cognitive decline. These findings are consistent with previous research indicating that CBF declines with age in the superior frontal (Parkes et al., 2004) and insula (Chen et al., 2011; Zhang et al., 2018; Leenders et al., 1990), inferior parietal and precuneus (Chen et al., 2011; Zhang et al., 2018; Beason-Held et al., 2009), occipital (Chen et al., 2011; Beason-Held et al., 2009), anterior cingulate (Zhang et al., 2018), and cerebellar (Zhang et al., 2018). The diminished blood flow in these regions could relate to a decrease in synaptic activity or a reduction in metabolic demand (Aanerud et al., 2012) and cerebral vasculature degeneration (Farkas and Luiten, 2001), potentially contributing to the age-related decline in sensorimotor processing and cognitive function. The broad distribution of affected areas corresponds with the diffuse pattern of cognitive changes often reported in aging, such as slower processing speed and reduced working memory capacity (Park and Reuter-Lorenz, 2009). The CBF reduction with advancing age in the medial frontal/anterior cingulate, precuneus, and angular regions, which form the default mode network, may be especially relevant to age-related cognitive decline. This finding aligns with a recent study (Pelizzari et al., 2020) reporting that the negative correlation between age and CBF in the predefined default mode network region observed in healthy controls (age range: 19 to 68 years), as demonstrated by multiple-delay PCASL imaging. The default mode network is also involved in the pathophysiology of MCI and AD (Li et al., 2016). A recent study reported that CBF in the medial frontal/anterior cingulate region, overlapping with our age-related CBF reduction in its spatial location, can predict fluid cognition (those related to performing tasks) in healthy subjects with a 4-year follow-up (De Vis et al., 2018). However, we did not analyze the association of CBF with cognitive performance in this study. The reduction of CBF in the occipital, cerebellar, temporal, and parietal regions with aging could reflect further progression of age-related cerebral degeneration. The occipital/cuneus and parietal lobes are integral to visual and spatial processing (Simpson et al., 2011), and declines here may be related to the common age-related impairments in visual-spatial abilities and attention (Wang et al., 2020). Older adults have been shown to recruit cuneus to compensate for reduced fluid cognition during visual problem-solving (Knights and al., 2024) and therefore reduced baseline CBF in this region will affect visual cognition in aging. Moreover, the cerebellum’s role in cognitive processes has been increasingly recognized (Stoodley and Schmahmann, 2009), and its diminished perfusion in the elderly might contribute to deficits in coordination and executive functions. The temporal region, important for memory and language, may also reflect the vulnerability of these functions to the aging process (Filippi et al., 2023).

Conversely, age was positively associated with relative CBF (wo. ATT correction) in the subcortical regions, including the putamen, thalamus, pallidum, caudate, middle cingulate, hippocampus/parahippocampus, amygdala regions, while age-associated CBF increases in these subcortical regions were not apparent for absolute CBF wo. ATT correction. These findings are generally consistent with no changes of subcortical CBF in aging when using absolute CBF values (Chen et al., 2011; Rusinek et al., 2011) and augmented subcortical CBF in normal aging populations (Zhang et al., 2017; Lee et al., 2009; Pagani et al., 2002), especially after correcting for global CBF values. It is worth noting that we observed preserved CBF in more subcortical regions. Our results suggested the preserved CBF with advanced ages in these subcortical regions and the observed higher relative CBF was caused by globally reduced CBF in the cortical regions with aging. The putamen and thalamus are key components of motor and cognitive circuits, respectively, and their preserved/relatively increased perfusion could be an adaptive response to maintain motor function and cognitive processing (Seidler et al., 2010). The hippocampus is crucial for memory formation, and its preserved/relatively increased perfusion may reflect an attempt to counteract memory decline (Dickerson and Augustinack, 2012; Eckert and Abraham, 2013), via neurovascular coupling, which is the relationship between local neural activity and subsequent changes in blood flow (Iadecola, 2004). This would suggest that the aging brain may attempt to maintain cognitive function by reallocating blood flow to crucial neural substrates involved in compensatory neural activity.

We observed significantly higher global CBF wo. ATT correction in females compared to males but no differences in global ATT and global CBF w. ATT correction between female and male. The findings of global CBF wo. ATT correction agrees well with the prior literature (Liu et al., 2012; Zhang et al., 2018; Parkes et al., 2004; Smith et al., 2019). The absence of gender differences in global ATT (not significant but with a longer ATT trend in males) in our study, is in contrast to those observed in the literature (Liu et al., 2012; Juttukonda et al., 2021), which may be due to cohort differences. Our low-resolution ATT mapping could potentially cause our missed gender effect. However, using similar ATT acquisitions, we observed the gender difference in regional ATTs in the elderly cohort in our prior study (Dai et al., 2017) and so it is unlikely caused by our low-resolution ATT acquisition. In addition, no significant interaction between age and gender in global ATT and global CBF was found, echoing previous findings that the slope of global ATT/CBF and age are independent of gender (Juttukonda et al., 2021; Chen et al., 2011; Zhang et al., 2018).

We have suggested that obtaining an extra ATT map may not be necessary for cross-sectional group comparisons of CBF maps because global scaling of CBF maps wo. ATT correction can reduce its variability (Dai et al., 2017). In this study, the age-related patterns observed using relative CBF maps wo. ATT correction (Fig. 5f) are roughly similar in distribution to those obtained using CBF maps w. ATT correction (Fig. 6c), though with a smaller spatial extent. However, using relative CBF maps wo. ATT correction results in an artifactually increased CBF with age, potentially distorting our understanding of age-related brain changes. Therefore, acquiring an additional ATT map is crucial to avoid ambiguous interpretations in aging studies.

As ATT prolongs with age, it can cause a misleading appearance of reduced CBF without ATT correction due to the timing choices in the ASL sequence. However, even after correcting for ATT, we still observed age-related hypoperfusion in the posterior cingulate, precuneus, supramarginal gyrus extending to the inferior parietal (lateral parietal), occipital, anterior cingulate gyrus extending to frontal regions. These findings suggest that the observed age-related changes are not merely methodological artifacts from incorrect ASL timing, but rather reflect a genuine physiological effect. It is worth noting that the observed age-related CBF pattern is similar to Alzheimer’s disease-related glucose metabolism/CBF patterns (Benzinger et al., 2013; Binnewijzend et al., 2013). Perfusion declined in the medial temporal cortex/hippocampus only when patients became demented. Similar glucose metabolism/CBF patterns have been reported in earlier AD stages, such as MCI patients (Asllani et al., 2008; Dai et al., 2009) or healthy APOE E4 carriers (Langbaum et al., 2010). The CBF in the frontal lobe and anterior cingulate cortex can predict cognitive performance in healthy individuals in a 4-year follow-up (De Vis et al., 2018). Cognition-related CBF patterns hold great promise to distinguish neurodegenerative diseases from normal aging.

Our study has limitations. First, the low-resolution ATT map can accurately estimate the voxel-wise arterial transit time but it is affected by partial volume effects (Dai et al., 2012). This results in a reduced ATT in the white matter regions due to the blurring of nearby gray matter perfusion. Additionally, our prior study showed that high-resolution multiple-delay imaging in the deep white matter regions tend to overestimate ATT values due to insufficient SNR, which hampers accurate ATT calculation (Dai et al., 2012). This may partially explain the longer white matter ATT observed in some studies. Another contributing factor is the assumption of a uniform gray matter T1 value for the entire brain in our signal-weighted delay model. Since white matter has a shorter T1 compared to gray matter, this assumption results in an underestimation of ATT in white matter regions. Together, these factors may contribute to smaller differences between the global ATT in gray matter and the whole brain. Second, CBF reduction with advanced age might be influenced by factors such as increased partial volume effects due to brain atrophy. Partial volume-corrected CBF may provide additional information about tissue-specific CBF changes with aging. Third, it is important to note that our findings are based on a relatively healthy cohort of highly educated participants. Therefore, caution is warranted when generalizing these observations to other cohorts, especially those with lower education levels. Fourth, although we tried to exclude subjects with cerebrovascular disease and neurodegenerative disease by inquiry of medical history and structural MRI, we did not directly assess or control for biological confounds affecting CBF variations, such as amyloid beta, APOE genotype, diabetes, and caffeine intake at the time of the study. Consequently, it is possible that some variations in CBF measures are not related to age but are instead induced by other conditions associated with age. Fifth, we had a small sample size in the older adults group. The large variance in this group may cause us to miss the ATT/CBF differences between middle-aged adults and older adults. Sixth, this is a cross-sectional study with large variation across subjects. Longitudinal studies are warranted to verify the age-related CBF and ATT changes and their association with age-related cognitive performance.

5. Conclusion

We identified age-related patterns in ATT and CBF, ensuring that CBF measurements were free from ATT contamination. ATT lengthens with age in the frontal, temporoparietal, and occipital regions, with a more pronounced elongation in males than females in the middle temporal regions. After correcting for ATT, we observed CBF decreases with age in the anterior cingulate, insula, posterior cingulate, precuneus, angular, supramarginal, frontal, parietal, superior and middle temporal, occipital, and cerebellar regions, while CBF remained stable in the inferior temporal and subcortical regions. However, without ATT correction, we observed artifactual decreases in the inferior temporal and precentral regions. These findings underscore the importance of ATT as an independent indicator of microvascular deficits and emphasize the necessity of ATT correction in CBF studies involving aging populations.

Supplementary Material

Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.neuroimage.2025.121098.

Data availability

Raw data were generated from MRI scanner. Reconstruction software is vendor’s proprietary product. Sharing of derived data will be supported by direct request to the PIs for different data sets. Before sharing data the PIs will make sure that all data are free of identifiers that could directly or indirectly link information to an individual or vulnerable group and that all sharing is compliant with institutional and IRB policies. The code supporting the findings of this study is available upon request.

Data availability

Data will be made available on request.