Participants were selected at the Neurorehabilitation Center of China Rehabilitation Research Center (CRRC) from March 2021 to August 2024. Strictly adhering to the following criteria, the study included thirty patients with PSCI as the experimental group, thirty patients with NPSCI and thirty age- and gender-matched healthy controls (HCs) as the control groups. This study was approved by the Medical Ethics Committee of the CRRC (No. 2024-086-02) and registered in the Chinese Clinical Trial Registry (ChiCTR2400089609).

For participants in the PSCI and NPSCI groups, who were all stroke patients, the shared inclusion criteria were as follows: (1) aged 35–75 years; (2) met diagnostic criteria of ischemic or hemorrhagic stroke (Warner et al., 2019; Greenberg et al., 2022), and was confirmed by computerized tomography (CT) or magnetic resonance imaging (MRI); (3) first-ever stroke, or stroke without significant sequelae; (4) stroke duration from 3 to 12 months; (5) stroke location in the unilateral supratentorial region; and (6) signed informed consent. Additionally, inclusion criteria for PSCI patients were also (1) met the PSCI diagnostic criteria (Wang K. et al., 2021); and (2) had mild to moderate cognitive impairment: the Mini-Mental State Examination (MMSE) < 27 or the Montreal Cognitive Assessment (MoCA) < 26, and MMSE ≥10. In contrast, the inclusion criteria for NPSCI patients were the MMSE ≥27 or the MoCA ≥26.

The shared exclusion criteria for stroke patients were as follows: (1) history of drug or alcohol abuse or substance misuse; (2) presence of severe depression, anxiety, or other psychiatric disorders; (3) presence of severe primary diseases in the circulatory, respiratory, digestive, urinary, endocrine, or hematopoietic systems that conventional medications cannot control; (4) presence of malignant hypertension or malignant tumors; (5) presence of severe infections, water and electrolyte imbalances, or acid–base disturbances; (6) presence of contraindications to MRI scans, such as any implanted metallic foreign bodies, electronic devices or claustrophobia; (7) pregnancy, breastfeeding, or planning to become pregnant; (8) concurrent participating in other clinical trials; and (9) presence of other conditions, as determined by the investigator, renders participants ineligible for the trial. Additionally, exclusion criteria for PSCI were also cognitive impairment resulting from not stroke, such as Alzheimer’s disease, frontotemporal dementia, lewy body dementia, vascular dementia, Parkinson’s disease dementia, traumatic brain injury, or other conditions.

For the HCs, the inclusion criteria were as follows: (1) aged 35–75 years; (2) physically healthy, with no acute or chronic diseases; (3) no history of major surgery or severe trauma; and (4) signed informed consent. The exclusion criteria were: (1) any current or past history of acute or chronic illnesses; (2) history of drug or alcohol abuse or substance misuse; (3) presence of severe depression, anxiety, or other psychiatric disorders; (4) presence of contraindications to MRI scans; (5) pregnancy, breastfeeding, or planning to become pregnant; (6) concurrent participating in other clinical trials; and (7) presence of other conditions, as determined by the investigator, renders participants ineligible for the trial.

In this study, MoCA scores were collected from PSCI patients within 3 days after the rs-fMRI scan. MoCA is a rapid and validated tool for assessing cognitive function, including visuospatial and execution, naming, memory, attention, language, abstraction, and delayed recall (Shi et al., 2018). MoCA score ranges from 0 to 30, with higher scores indicating better cognitive function.

Participants underwent MRI scanning using the same Philips Ingenia 3T MRI. To improve resolution, the scanner was equipped with a 32-channel head coil. Participants were instructed to close their eyes and stay awake during the scan. A rescan would be performed if significant head movement was detected during the scan, affecting data quality. Rs-fMRI images were collected with a gradient echo planar imaging (EPI) sequence: TR = 2000 ms, TE = 30 ms, field of view (FOV) = 224 × 224 mm², flip angle (FA) = 90°, matrix = 64 × 64, number of slices = 32, and voxel size = 3.5 × 3.5 × 4.35 mm³. High-resolution T1-weighted images were acquired using a three-dimensional magnetization-prepared rapid gradient echo (3D-MPRAGE) sequence: TR = 7 ms, TE = 3.2 ms, FOV = 256 × 256 mm², FA = 7°, matrix = 256 × 256 mm², number of slices = 192, and voxel size = 1 × 1 × 1 mm³.

In the MATLAB environment (MathWorks, Natick, MA, USA), REST software (version 1.8)¹ based on SPM 12² was used to preprocess the rs-fMRI data and calculate brain functional indicators. Preprocessing procedures were as follows: (1) converting image files from EPI DICOM to NIFTI format; (2) removing the first 10 time points; (3) slice timing correction; (4) realigning correction for head motion, with the maximum head motion in x, y, and z directions being less than 2 mm and maximum angular rotation being less than 2°for all participants; (5) coregistering functional images with the corresponding 3D T1-weighted structural images, and then spatial normalized into the standard Montreal Neurological Institute (MNI) space with 3 × 3 × 3 mm³ resample; (6) spatial smoothing using a Gaussian kernel of a 6 mm full-width-at-half-maximum (FWHM); (7) removing linear trends; (8) performing band-pass filtering (0.01–0.08 Hz) to eliminate the effects of low-frequency drift and high-frequency noise (Biswal et al., 1995); and (9) regressing out nuisance covariates, including six head motion parameters, global mean, white matter, and cerebrospinal fluid signals (Friston et al., 1996).

The fALFF was a functional indicator reflecting the intensity of spontaneous neural activity in local brain regions. Unlike the methods above, the preprocessing of fALFF was not band-pass filtered. It transformed the time series of each voxel to the frequency domain for the power spectrum by a fast Fourier transform (FFT) and then calculated the square root of each frequency in the power spectrum, where the averaged square root across 0.01–0.08 Hz was the ALFF (Zang et al., 2007). Afterwards, the fALFF was calculated as the ratio of the ALFF to the sum of amplitudes in the full frequency range (Zou et al., 2008). Finally, to improve the normality and comparability of the data, a Fisher Z transformation was performed on the fALFF to obtain the zfALFF map.

KCC-ReHo evaluated this brain region’s functional homogeneity by measuring the time series’s coherence among neighboring voxels within that specific region (Zhang et al., 2022). The preprocessing of KCC-ReHo was not spatially smoothed. Instead, spatial smoothing was performed after obtaining KCC-ReHo values to derive SKCC-ReHo. Similarly, a Fisher Z transformation was performed on the SKCC-ReHo to obtain the SzKCC-ReHo map.

Seed-based correlation analysis was used to study the effect of local neural activity alterations on whole-brain FC. Clusters with significant differences among groups in fALFF were set as seeds. Each seed’s time series of each voxel was extracted in a spherical region (radius = 5 mm), and all voxels within the seed were averaged to obtain the mean time series of the seed region. Furthermore, Pearson correlation coefficients between the mean time series of each seed region and that of each voxel in the whole brain were then calculated, which were transformed into z-scores using Fisher r-to-z transformation to obtain the zFC maps.

Demographic and clinical variables of all participants were analyzed using SPSS statistics 26.0 (IBM Corp. Armonk, NY, USA). Shapiro–Wilk test was used to determine the normality of variable distribution. Normally distributed continuous variables were presented as means [standard deviation (SD)], and the comparisons between groups were conducted using one-way analysis of variance (ANOVA) or Student’s t-test. Categorical variables were presented as frequencies (percentages), and group comparisons were conducted using the chi-square or Fisher’s exact tests. The significance level was set at p < 0.05 (two-tailed).

One-way ANOVA was used to calculate fALFF, KCC-ReHo, and FC. Least Significant Difference (LSD) was used as post hoc analysis for the significant clusters of the inter-group differences [False discovery rate (FDR) correction, cluster-level p < 0.05, voxel-level p < 0.001, two-tailed]. Additionally, clusters with significant differences in fALFF values between the PSCI and NPSCI groups were selected, and Pearson correlation was used to analyze the correlation between the fALFF values of these clusters and MoCA scores in PSCI patients (p < 0.05, two-tailed).

Table 1 showed no statistical differences in age, gender, and education level among the three groups (p > 0.05). Similarly, there were no statistical differences in stroke duration and type, lesion hemisphere, hypertension, hyperlipidemia, diabetes mellitus, smoking, alcoholism history, and National Institute of Health stroke scale (NHISS) scores (p > 0.05). On the contrary, there was a statistically significant difference in MMSE scores among the three groups (p < 0.001).

One-way ANOVA indicated significant differences in zfALFF values among the PSCI, NPSCI, and HC groups in the cingulate gyrus, putamen, inferior parietal lobe (IPL), left caudate, bilateral angular gyrus, and precuneus (Supplementary Table S1 and Figure 1). Further post hoc analysis was conducted with the PSCI group as the focus. Compared with the NPSCI group, the PSCI group exhibited decreased zfALFF values in the left caudate, right inferior temporal gyrus (ITG), and anterior cingulate cortex (ACC) (Table 2 and Figure 1A). Compared with the HC group, the PSCI group showed significantly decreased zfALFF values in the left putamen and left superior temporal gyrus (STG), while increased zfALFF values were observed in the right cerebellum_6 (Table 2 and Figure 1B).

Similarly, there were significant differences in SzKCC-ReHo values among the three groups (Supplementary Table S2 and Figure 2). Post hoc analysis with the PSCI group as the focus revealed that, compared with the NPSCI group, the PSCI group exhibited decreased SzKCC-ReHo values in the right middle frontal gyrus (MFG), while increased SzKCC-ReHo values in the left cerebellum_crus 1 (Table 3 and Figure 2A). Compared with the HC group, the PSCI group showed decreased SzKCC-ReHo values in the left postcentral lobe, while increased SzKCC-ReHo values were observed in the left cerebellum_4–5 (Table 3 and Figure 2B).

In this study, clusters with significant differences in zfALFF values among the three groups were set as seeds, including the cingulate gyrus, right angular gyrus (RAG), left angular gyrus (LAG), and precuneus. The results showed that, regardless of which seed was used for the zFC analysis, there were significant differences among the three groups (Supplementary Table S3 and Figure 3). Similarly, post hoc analysis with the PSCI group as the focus was conducted. Compared with the NPSCI group, when using the cingulate gyrus as the seed, the zFC value between the cingulate gyrus and the left inferior frontal gyrus (IFG)_triangular part decreased in the PSCI group. When using RAG as the seed, the zFC values between the RAG and two clusters, including the right superior frontal gyrus (SFG)_2 and left middle occipital gyrus (MOG), decreased in the PSCI group. When using LAG as the seed, the zFC values between the LAG and three clusters, including the right supramarginal gyrus (SMG), right amygdala, and right caudate, decreased in the PSCI group. When using the precuneus as the seed, the zFC values between the precuneus and two clusters, including the right SMG and right insula, decreased in the PSCI group (Table 4 and Figure 3).

Additionally, compared with the HC group, when using the cingulate gyrus as the seed, the zFC value between the cingulate gyrus and left Insula decreased in the PSCI group. When using RAG as the seed, the zFC values between the RAG and two clusters, including the precuneus and left cerebellum_crus 2, decreased in the PSCI group. When using LAG as the seed, the zFC values between the LAG and two clusters, including the left precuneus and right cerebellum_crus 1, decreased in the PSCI group. Furthermore, when using the precuneus as the seed, the zFC values between the precuneus and five clusters, including the SFG, right middle cingulate gyrus (MCG), bilateral angular gyrus, and right MFG_2, decreased in the PSCI group (Table 5 and Figure 4).

This study selected clusters with significant differences in zfALFF values between the PSCI and NPSCI groups, including the left caudate, left precentral gyrus, right ITG, and ACC, to explore the relationship between fMRI and cognition. We calculated the correlation between the zfALFF values of four clusters and the MoCA scores of PSCI patients separately. The results showed that only the zfALFF value of the ACC was positively correlated with the MoCA scores (Figure 5).

In the resting state, the PSCI group exhibited decreased zfALFF values in the left caudate, right ITG, and ACC compared with the NPSCI group, which is likely related to cognitive control and emotional regulation roles of these regions. Recent studies have also shown that the caudate is critical for working memory and attention control processes (Chiu et al., 2017; Samrani and Persson, 2024), while the ITG is important for advanced visual processing and semantic processing (Nobre et al., 1994; Schweinberger et al., 2002). Moreover, Bush et al. (2000) indicated significant activity in the ACC during cognitive control and emotional conflict, and Fleck et al. (2006) highlighted ACC plays a vital role in decision-making processes. Our study also revealed a positive correlation between MoCA scores and zfALFF values of the ACC, suggesting that the spontaneous neural activity intensity in the ACC is associated with cognitive dysfunction. When compared with the HC group, the PSCI group demonstrated decreased zfALFF values in the left putamen and left STG, which are involved in motor learning (Lee and Seo, 2016; Haber and Knutson, 2010) and sensory processing (Alloway et al., 2017).

Alterations in the coherence of specific brain regions have been observed in patients with PSCI (Liu et al., 2014). Yuan and Raz (2014) found that the MFG exhibited decreased neural synchrony after impairments in executive function. Consistent with these studies, our findings showed that the PSCI group exhibited decreased KCC-ReHo values in the right MFG. Additionally, the PSCI group also showed increased KCC-ReHo values in certain cerebellar regions, including the left cerebellum_crus 1 and left cerebellum_4–5. The cerebellum is part of cognitive and emotional circuits (Stoodley et al., 2016), and its increased synchrony may reflect compensatory mechanisms in response to cognitive decline. Wu et al. (2022) also demonstrated increased ReHo values in the cerebellum_4–5 in patients with mild cognitive impairment, further highlighting the cerebellum’s role in cognitive function.

Previous studies found that PSCI patients diminished FC in the default mode network (DMN), with significant changes also observed in the ACC and precuneus (Yue et al., 2023). In our study, when the cingulate gyrus was used as the seed, the PSCI group showed significantly decreased zFC values between the cingulate gyrus and the left IFG_triangular part, as well as the left insula. These findings suggested altered interactions between these regions, which are involved in cognitive control and emotional regulation (Seamans and Floresco, 2022). Notably, the weakened FC could contribute to cognitive deficits in PSCI patients, particularly in executive function (Dosenbach et al., 2006).

The PSCI group showed significant alterations in zFC values among multiple brain regions when using the angular gyrus as the seed. Among these, the zFC values were decreased between the RAG and several regions, including the right SFG_2, left MOG, precuneus, and left cerebellum_crus 2. These regions are crucial for cognitive functions, and decreased FC in these areas may reflect the cognitive decline observed in PSCI (Buckner et al., 2008). Furthermore, when the LAG was used as the seed, reduced zFC values in the PSCI group were also found between the LAG and several regions, such as the right SMG, right amygdala, right caudate, left precuneus, and right cerebellum_crus 1. These areas may be related to language processing, emotional regulation, and executive functions (Baghernezhad and Daliri, 2024). It is well known that the caudate and amygdala are part of the striatum, which belongs to the SN and is involved in memory and learning functions (Milardi et al., 2019). Additionally, Stoodley and Schmahmann (2009) indicated that the cerebellum plays a significant role in cognitive processing and emotional regulation, and weakened connectivity may be associated with the extensive cognitive deficits observed in PSCI. Furthermore, our study suggested that alterations in FC may be a crucial basis for cognitive dysfunction in PSCI patients.

Finally, when using the precuneus as the seed, the PSCI group also showed decreased zFC between the precuneus and several regions, such as the right SMG, right insula, SFG, and right MCG. The precuneus is a key node in the DMN and is involved in the limbic system. Thus, weakened connectivity in this area may reflect impairments in memory and spontaneous thinking in PSCI patients (Greicius et al., 2003; Cavanna and Trimble, 2006).

This study provides new insights into the neural mechanisms of brain function in PSCI patients, but it also has some limitations. First, there was some heterogeneity in the clinical characteristics of the participants. However, due to the small sample size in each group, we did not divide the subgroups by stroke location or severity to discuss their effects on brain function separately. Second, this study was a case–control study that only observed inter-group differences at a specific time and did not dynamically observe brain functional alterations in PSCI patients over time. In the future, a larger sample size and long-term follow-up should be conducted to further elucidate the neural mechanisms of PSCI and to provide evidence-based support for the cognitive function recovery of PSCI patients.