Multi-metric resting-state fMRI reveals brain network abnormalities underlying cognitive impairment in male patients during methamphetamine abstinence Authors: Wen Yang, Dezhi Hou, Yizhen Zeng, Bingbing Ling, Yun Zhu, Yu Xu, KunHua Wang, Yi Lu, Mei Zhu Published: 2026 ABSTRACT BackgroundMethamphetamine (MA) use disorder leads to persistent cognitive impairment, but the underlying neural mechanisms during abstinence remain unclear. This study aimed to investigate brain functional abnormalities in male methamphetamine-abstinent patients and their associations with cognitive impairment using multiple resting-state fMRI metrics.MethodsTwenty-five male MA-abstinent patients (Disease Group, DG) and twenty-nine healthy controls (Healthy Control Group, HC) were recruited, and rs-fMRI data were collected. Multimodal analysis methods, including regional homogeneity (ReHo), fractional amplitude of low-frequency fluctuations (fALFF), seed-based functional connectivity (FC), and independent component analysis (ICA), were employed to assess local spontaneous activity, hippocampal-centered functional connectivity, and large-scale resting-state network abnormalities. Cognitive function was evaluated using the Montreal Cognitive Assessment (MoCA). Group differences were compared using independent-samples t-tests, and brain-behavior relationships were analyzed using partial correlation and interaction models.ResultsCompared with the HC group, the DG group showed reduced fALFF in several regions, including the left fusiform gyrus and right inferior frontal gyrus pars triangularis, using a voxel-level threshold of p < 0.005 combined with cluster-level FWE correction at p < 0.05. Reduced ReHo was also observed in the right inferior frontal gyrus pars triangularis using a voxel-level threshold of p < 0.05 combined with cluster-level FWE correction at p < 0.05. Seed-based functional connectivity analysis revealed widespread enhanced right hippocampal functional connectivity in the DG group using a voxel-level threshold of p < 0.005 combined with cluster-level FWE correction at p < 0.05. ICA identified abnormal functional connectivity in the default mode, auditory, and somatomotor networks, with all findings surviving cluster-level FWE correction at p < 0.05. Right hippocampus–right inferior temporal gyrus connectivity was strongly positively correlated with MoCA total score in the DG group (r = 0.765, p < 0.001), and this association was significantly stronger in the DG group than in the HC group.ConclusionMale MA-abstinent patients exhibit multi-layered brain functional abnormalities involving local activity, functional connectivity, and large-scale networks. These abnormalities, together with the disease-specific associations between right hippocampal functional connectivity and cognitive performance, suggest a potential role for right hippocampal networks in cognitive impairment during methamphetamine abstinence in men. Introduction Methamphetamine (MA) is a potent central nervous system stimulant, and methamphetamine use disorder has become a global public health crisis [1]. Long-term MA use not only leads to severe addictive behaviors but also precipitates widespread cognitive impairments, including deficits in attention, memory, and executive function [2, 3]. Critically, these cognitive deficits persist during abstinence, profoundly impacting patients’ recovery, social reintegration, and significantly increasing the risk of relapse [4]. Therefore, a thorough understanding of the neural mechanisms underlying MA-related cognitive impairment during abstinence is crucial for developing effective intervention strategies [5]. However, current research predominantly focuses on the acute effects of MA, leaving the persistent cognitive impairment during abstinence and its underlying brain network basis largely unsystematized. Resting-state functional magnetic resonance imaging (rs-fMRI), by detecting spontaneous blood oxygen level-dependent (BOLD) signal fluctuations, provides a non-invasive and reliable method for revealing the brain’s intrinsic functional organization [6]. This technique has been widely applied in the study of various psychiatric and neurological disorders, and in the addiction field, it has uncovered functional dysregulation in several key brain networks, such as the Default Mode Network (DMN), Salience Network (SN), and Central Executive Network (CEN) [7, 8]. Abnormal connectivity patterns within these networks are often associated with impaired cognitive control, heightened impulsivity, and increased craving in individuals with addiction [9, 10]. However, existing research is largely limited to single imaging metrics or specific brain networks, lacking a multi-layered perspective from “local activity” to “intra-network connectivity” to “inter-network interactions” to systematically elucidate the patterns of brain functional reorganization during MA abstinence. Furthermore, while the hippocampus and prefrontal cortex play central roles in addiction memory and cognitive control, their functional connectivity patterns and disease-specific associations with cognitive impairment in the context of long-term MA abstinence remain insufficiently clarified. One brain region of particular interest in addiction research is the hippocampus. The hippocampus was selected as a seed region for functional connectivity analysis based on converging evidence from preclinical and human studies [11]. Chronic methamphetamine exposure induces structural and functional alterations in the hippocampus, a region critical for memory consolidation and contextual learning [12]. In individuals with methamphetamine use disorder, resting-state fMRI studies have reported altered hippocampal connectivity with prefrontal and temporal regions, and these abnormalities are associated with cognitive deficits and craving severity [13]. The hippocampus was selected as a seed region for functional connectivity analysis because converging evidence from preclinical and human studies supports its central role in addiction-related associative memory and contextual learning (11–13). In the present study, we focused on the right hippocampus a priori because previous human studies suggest relative functional lateralization, with the right hippocampus being more strongly implicated in contextual and spatial memory processing [14, 15]. Given the role of cue-associated contextual memory in stimulant use disorders, focusing on right hippocampal connectivity was intended to characterize one theoretically relevant pathway of network reorganization during MA abstinence. To address these research gaps, this study employed a multi-modal rs-fMRI analysis strategy, integrating regional homogeneity (ReHo), fractional amplitude of low-frequency fluctuations (fALFF), seed-based functional connectivity (FC), and independent component analysis (ICA) methods. The aim was to systematically investigate brain functional abnormalities in male MA-abstinent patients and their relationship with cognitive impairment. The specific research objectives included: (1) identifying changes in local brain activity and synchronization in abstinent patients; (2) characterizing abnormal functional connectivity within key brain networks (such as the DMN, somatomotor network, and auditory network); (3) exploring the reconfigured functional connectivity patterns centered around the right hippocampus; and (4) uncovering the disease-specific associations between these neuroimaging features and multi-dimensional cognitive performance. Based on existing literature and our understanding of MA’s neurobiological effects, we hypothesized that: (1) MA-abstinent patients would exhibit general cognitive impairment; (2) local spontaneous activity and functional connectivity would be abnormal in several key brain regions (particularly those related to reward, memory, and executive function, such as the hippocampus and prefrontal cortex); and (3) these brain functional abnormalities would be significantly correlated with the degree of cognitive impairment in MA-abstinent patients, and this correlational pattern would likely be disease-specific. This study was designed to characterize multi-level functional brain abnormalities in MA-abstinent patients and to examine whether hippocampal-centered connectivity patterns are associated with cognitive impairment. We expected that these findings would help clarify the neuroimaging correlates of cognitive dysfunction during MA abstinence and inform future hypothesis-driven studies. In this study, we adopted a complementary two-level analytical framework. First, we conducted exploratory analyses using ReHo and fALFF to provide a global profile of local functional brain activity alterations in MA-abstinent patients. Second, to test a specific hypothesis-driven pathway, we performed a seed-based functional connectivity (FC) analysis focusing on the right hippocampus. This approach allows us to examine how a critical hub for addiction-related memory (the hippocampus) interacts with broader cortical regions that may exhibit local functional impairments. Given that only male participants were included, the present study was designed to characterize male-specific patterns of brain functional reorganization during MA abstinence. Materials and methods Participants and clinical assessment This study was registered at the Chinese Clinical Trial Registry (Registration No. ChiCTR2100051931) on 2021/10/09 (http://www.chictr.org.cn/). All participants provided written informed consent prior to participation. The study recruited 30 male MA-abstinent patients (Disease Group, DG) meeting the diagnostic criteria for methamphetamine use disorder according to the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5), and 30 age- and demographic-matched healthy male controls (Healthy Control Group, HC). All participants were right-handed, aged 18–60 years, had no history of major physical illness, neuropsychiatric disorders, or contraindications for MRI scanning. The inclusion criteria for MA-abstinent patients (DG) were: (1) meeting the DSM-5 diagnostic criteria for methamphetamine use disorder, confirmed by structured clinical interviews conducted by two experienced psychiatrists; and (2) currently in MA abstinence. In this study, “abstinence” was defined as self-reported continuous cessation of methamphetamine use, verified by negative results on random urine toxicology screens at enrollment and on the day of MRI scanning. All patients were scanned after at least three months of self-reported abstinence and within the first year following cessation (i.e., between approximately 3 and 12 months of abstinence), corresponding to the protracted abstinence phase. Patients reporting less than three months of abstinence or with unstable medical conditions were excluded. Because participants were unable to reliably recall the exact calendar date of their last use, precise day-level abstinence duration was not available for inclusion as a covariate in statistical analyses; however, the cohort was homogeneous with respect to being in the protracted abstinence time window. To ensure research specificity, patients with a history of use disorders on other psychoactive substances were excluded. Healthy controls (HC) had no history of psychoactive substance use disorder and no psychiatric diagnoses. All participants completed the Montreal Cognitive Assessment (MoCA) to comprehensively evaluate overall cognitive function and various subdomains (e.g., executive function, memory, attention). Unfortunately, due to the complexity of the MA user sample and research constraints, we were unable to systematically collect detailed information on methamphetamine exposure severity, including cumulative dosage, frequency of use, duration of use, and other indices of addiction severity, for the DG group. Consequently, these variables were not available for dose-dependent or severity-stratified statistical analyses in the present study. fMRI data acquisition All participants underwent imaging data acquisition using a 3.0 T magnetic resonance scanner. Routine cranial MRI was performed prior to scanning to exclude obvious structural brain abnormalities. Throughout the scanning procedure, all participants were instructed to remain still, awake, and keep their eyes open. Cushions were used to restrict head movement, and earplugs were used to reduce noise. Resting-state functional images were acquired using an echo-planar imaging sequence with the following parameters: TR = 2000 ms, TE = 30 ms, flip angle = 90°, voxel size = 3.5 × 3.5 × 3.5 mm³, field of view = 224 × 224 mm², 35 axial slices per volume, and 230 time points per volume. High-resolution three-dimensional T1-weighted structural images were acquired as anatomical references with the following parameters: TR = 7.2 ms, TE = 3 ms, flip angle = 10°, voxel size = 1.0 × 1.0 × 1.0 mm³, field of view = 230 × 256 mm². Data preprocessing Preprocessing followed standard procedures: Raw DICOM data were converted to NIfTI format using the dcm2niix tool. All preprocessing steps were performed using SPM12 and DPABI toolboxes, as follows: Time point discarding: The first 10 time points were removed to eliminate initial unstable magnetic field signals. Slice-timing correction: Interpolation was used for interleaved slice acquisition (odd slices first). Head motion correction: Images were realigned to the mean functional image, and subjects with head motion exceeding 3 mm translation or 3° rotation were excluded. Frame-wise displacement (FD) information was generated during preprocessing to quantify volume-to-volume head motion and was used as part of the head-motion quality-control inspection. Head-motion quality control was primarily based on the conventional maximum displacement threshold of 3 mm translation or 3°rotation. In addition, Friston 24 head motion parameters were regressed out during nuisance covariate regression to reduce residual motion-related effects. Spatial normalization: Individual functional images were normalized to Montreal Neurological Institute (MNI) standard space using a unified segmentation algorithm and resampled to 3 × 3 × 3 mm³ voxels. Spatial smoothing was applied using Gaussian kernels with different full-width at half-maximum (FWHM) values according to the specific analytical purpose. A 6-mm FWHM kernel was used for fALFF and ICA analyses to improve signal-to-noise ratio and enhance the stability of group-level spatial patterns and independent components. For seed-based functional connectivity analysis, a relatively smaller 4-mm FWHM kernel was used to preserve spatial specificity around the hippocampal seed and its voxel-wise connectivity patterns while still reducing high-frequency spatial noise. For ReHo analysis, no smoothing was applied before calculation because ReHo measures local synchronization among neighboring voxels, and prior smoothing could artificially increase local similarity. Thus, smoothing procedures were tailored to the methodological characteristics of each rs-fMRI metric. Denoising: Linear detrending was performed, and linear trends of voxel time series, Friston 24 head motion parameters, white matter, and cerebrospinal fluid signals were regressed out. Band-pass filtering: For ReHo, FC, and ICA time series, a band-pass filter of 0.01–0.08 Hz was applied (not applied before fALFF calculation). Multi-modal functional metrics calculation To systematically elucidate brain functional reorganization, local metrics (ReHo and fALFF) were first employed to identify spontaneous neural activity changes across the whole brain. Subsequently, the right hippocampus was selected as an a priori ROI for seed-based FC analysis. This dual-level strategy was designed to bridge the gap between regional functional deficits and long-range network dysconnectivity. Local activity and synchronization Fractional Amplitude of Low-Frequency Fluctuations (fALFF): Reflects the intensity of spontaneous neuronal activity, obtained by calculating the ratio of the power spectrum in the 0.01–0.08 Hz band to the power spectrum of the entire band. Regional Homogeneity (ReHo): Measures the synchronization of local neuronal activity, obtained by calculating Kendall’s coefficient of concordance between each voxel and its 26 neighboring voxels’ time series. These metrics were calculated using the RESTplus toolbox. Seed-based voxel-wise functional connectivity The right hippocampus was selected as the seed region a priori based on evidence that the hippocampus is critically involved in contextual memory and addiction-related associative processing, and that the right hippocampus may show relative dominance in contextual/spatial memory functions. This hypothesis-driven choice was intended to examine one theoretically relevant hippocampal pathway during MA abstinence, although potential involvement of the left hippocampus cannot be excluded. The average time series from the right hippocampal seed region was extracted, and Pearson correlation coefficients between this seed time series and the time series of all other brain voxels were calculated to generate individual voxel-wise functional connectivity maps. The resulting correlation coefficients were transformed into Fisher’s z values to improve normality and to facilitate subsequent group-level statistical comparisons and brain-behavior correlation analyses. This analysis was performed using RESTplus. Independent component analysis (ICA) and large-scale brain networks Group-level ICA was performed using the GIFT toolbox. The steps were as follows: Data dimensionality reduction and decomposition: All subject data were first reduced to 30 dimensions via principal component analysis, then decomposed into 20 independent components using the Infomax algorithm. Stability verification: ICASSO toolbox was used for 100 repeated calculations, and the most stable components were selected. Component identification and reconstruction: Resting-state network components were identified by spatial matching with the resting-state network templates provided in the GIFT toolbox, including canonical templates for the default mode, auditory, and somatomotor networks, supplemented by visual inspection of their anatomical distributions and canonical spatial patterns. Specifically, components corresponding to the default mode network, auditory network, and somatomotor network were identified based on their spatial similarity to the template networks and their anatomical distribution. Each subject’s component spatial maps and time series were then obtained using the GICA back-reconstruction method. Statistical analysis Demographic and behavioral data Independent-samples t-tests (for continuous variables) or chi-square tests (for categorical variables) were performed using SPSS 26.0, with a significance level set at p < 0.05. Group comparisons of imaging data Two-sample t-tests were performed in SPM12 using a multi-level statistical threshold strategy. For fALFF and hippocampal seed-based functional connectivity analyses, a voxel-level threshold of p < 0.005 combined with cluster-level family-wise error (FWE) correction at p < 0.05 was applied. For ReHo analysis, a voxel-level threshold of p < 0.05 with cluster-level FWE correction at p < 0.05 was applied. For ICA-derived component comparisons, voxel-level thresholds were specified for each component analysis as reported in Table 5, and all results were corrected at the cluster level using FWE correction (p < 0.05). Brain-behavior correlation analysis Brain-behavior correlation analysis was performed in MATLAB R2018b. To investigate the relationship between right hippocampal functional connectivity and cognitive function, we first calculated partial correlations between right hippocampal functional connectivity values and MoCA total scores, as well as scores for each cognitive subdomain, separately within the DG and HC groups. In all partial correlation analyses, age and years of education were strictly controlled as covariates to exclude their potential confounding effects on brain-behavior associations. All partial correlation results were corrected for multiple comparisons using the False Discovery Rate (FDR) method (p < 0.05). Furthermore, to test whether these brain-behavior correlations were group-specific (i.e., whether significant differences existed between the DG and HC groups), a General Linear Model (GLM) was constructed. The cognitive score was used as the dependent variable, and group (categorical variable), functional connectivity value (continuous variable), and the interaction term between group and functional connectivity (Group×Functional Connectivity) were included as primary predictors in the model. Additionally, age and years of education were included as covariates in the GLM to further control for the influence of these demographic factors. For significant interaction terms in the model, simple slope analysis was performed to explore the specific correlation patterns within each group (DG and HC). Visualization of results All statistically significant brain regions were visualized using SPM12, xjView, Brainnet Viewer and MRIcroGL software. Images were rendered onto the Montreal Neurological Institute standard brain template, with statistical values indicated by a “cool-to-warm” color scale. Spatial maps of identified independent components and group differences were generated and rendered by the GIFT toolbox. Scatter plots for brain-behavior correlations were generated using MATLAB R2018b, with data distribution and regression lines presented separately for the healthy control group and the disease group. Results Demographic and behavioral results The final analysis sample included 25 male methamphetamine-abstinent patients (Disease Group, DG) and 29 healthy controls (Healthy Control Group, HC). Five subjects from the DG group and one from the HC group were excluded due to excessive head motion or poor data quality. The two groups showed no significant differences in age (t = -0.199, p = 0.843) or years of education (t = 1.050, p = 0.299). However, the MoCA total score in the DG group was significantly lower than that in the HC group (t = 7.395, p < 0.001; Cohen’s d = 2.02), indicating a large between-group difference in global cognitive performance. This impairment was also reflected in lower scores on several MoCA subdomains, including visuospatial/executive function, attention, abstraction, and delayed recall (all p < 0.01). Given the screening nature of the MoCA, these subdomain differences should be interpreted cautiously as preliminary indicators rather than definitive domain-specific deficits (Table 1; Fig. 1). Table 1Demographic and cognitive-behavioral characteristics of participantsItemHC group(n = 29)DG group(n = 25)t/χ valuep valueEffect size(Cohen’s d)Age(years)35.86 ± 8.0236.36 ± 10.37-0.1990.843—Education(years)10.55 ± 1.5710.08 ± 1.731.0500.299—MoCA score28.34 ± 2.2422.28 ± 3.707.395< 0.0012.02Visuospatial and Executive Function4.69 ± 0.602.84 ± 1.346.678< 0.001—Naming2.90 ± 0.312.76 ± 0.601.0760.287—Attention5.59 ± 0.574.80 ± 1.223.0960.003—Language2.90 ± 0.312.64 ± 0.701.7830.080—Abstraction1.93 ± 0.261.40 ± 0.654.074< 0.001—Delayed Recall4.66 ± 0.612.20 ± 1.637.513< 0.001—Orientation5.69 ± 0.545.64 ± 1.250.1940.847—Note: Data are presented as mean ± standard deviation. HC: Healthy Control group; DG: Methamphetamine Abstinent group; MoCA: Montreal Cognitive Assessment. Independent-samples t-tests were used for continuous variables, with p < 0.05 considered statistically significant. For the DG group, abstinence was verified by negative urine toxicology screens at enrollment and scanning. All patients had self-reported abstinence of at least 3 month and were scanned within the first year following cessation (verified abstinence window: approximately 3–12 months), corresponding to the protracted abstinence phase. Precise day-level data were not available due to unreliable recall of exact cessation dates Table 1Demographic and cognitive-behavioral characteristics of participantsItemHC group(n = 29)DG group(n = 25)t/χ valuep valueEffect size(Cohen’s d)Age(years)35.86 ± 8.0236.36 ± 10.37-0.1990.843—Education(years)10.55 ± 1.5710.08 ± 1.731.0500.299—MoCA score28.34 ± 2.2422.28 ± 3.707.395< 0.0012.02Visuospatial and Executive Function4.69 ± 0.602.84 ± 1.346.678< 0.001—Naming2.90 ± 0.312.76 ± 0.601.0760.287—Attention5.59 ± 0.574.80 ± 1.223.0960.003—Language2.90 ± 0.312.64 ± 0.701.7830.080—Abstraction1.93 ± 0.261.40 ± 0.654.074< 0.001—Delayed Recall4.66 ± 0.612.20 ± 1.637.513< 0.001—Orientation5.69 ± 0.545.64 ± 1.250.1940.847—Note: Data are presented as mean ± standard deviation. HC: Healthy Control group; DG: Methamphetamine Abstinent group; MoCA: Montreal Cognitive Assessment. Independent-samples t-tests were used for continuous variables, with p < 0.05 considered statistically significant. For the DG group, abstinence was verified by negative urine toxicology screens at enrollment and scanning. All patients had self-reported abstinence of at least 3 month and were scanned within the first year following cessation (verified abstinence window: approximately 3–12 months), corresponding to the protracted abstinence phase. Precise day-level data were not available due to unreliable recall of exact cessation dates Demographic and cognitive-behavioral characteristics of participants Note: Data are presented as mean ± standard deviation. HC: Healthy Control group; DG: Methamphetamine Abstinent group; MoCA: Montreal Cognitive Assessment. Independent-samples t-tests were used for continuous variables, with p < 0.05 considered statistically significant. For the DG group, abstinence was verified by negative urine toxicology screens at enrollment and scanning. All patients had self-reported abstinence of at least 3 month and were scanned within the first year following cessation (verified abstinence window: approximately 3–12 months), corresponding to the protracted abstinence phase. Precise day-level data were not available due to unreliable recall of exact cessation dates Fig. 1Group differences in MoCA total scores and sub-item scores. Note: The y-axis represents cognitive scores in points. HC, healthy controls; DG, methamphetamine-abstinent group. Error bars represent the standard error of the mean. Asterisks indicate significant group differences (*p < 0.05, **p < 0.01, ***p < 0.001; independent-samples t-test) Group differences in MoCA total scores and sub-item scores. Note: The y-axis represents cognitive scores in points. HC, healthy controls; DG, methamphetamine-abstinent group. Error bars represent the standard error of the mean. Asterisks indicate significant group differences (*p < 0.05, **p < 0.01, ***p < 0.001; independent-samples t-test) Resting-state fMRI analysis results Local brain activity and synchronization Compared to the DG group, the HC group showed significantly higher fALFF in the left fusiform gyrus (Fusiform_L), left calcarine cortex (Calcarine_L), right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R), and left supramarginal gyrus (SupraMarginal_L) (all clusters passed a voxel-level p < 0.005 and a cluster-level Family-Wise Error (FWE) correction p < 0.05; Table 2; Fig. 2). Regarding regional homogeneity (ReHo), the HC group showed significantly higher ReHo values within a large cluster centered in the right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R) and extending to adjacent temporal regions compared with the DG group (voxel-level p < 0.05, cluster-level FWE-corrected p < 0.05; Table 3; Fig. 3). Table 2Brain regions showing decreased fALFF in MA-abstinent patients (DG) compared to healthy controls (HC)Peak region of brainMNI Peak CoordinatesT valueCluster SizeXYZFusiform_L-36-45-214.9531136Calcarine_L-12-4564.4056121Frontal_Inf_Tri_R3912243.8441133SupraMarginal_L-60-24394.2273144Note: Results show brain regions where fALFF values were significantly lower in the MA-abstinent group (DG) compared to the healthy control group (HC) (DG < HC). Statistical thresholds were set at voxel-level p < 0.005 and cluster-level Family-Wise Error (FWE) corrected p < 0.05. Coordinates are based on the Montreal Neurological Institute (MNI) standard space. Only significant clusters passing correction and their peak voxel information are listed. fALFF: fractional amplitude of low-frequency fluctuations Table 2Brain regions showing decreased fALFF in MA-abstinent patients (DG) compared to healthy controls (HC)Peak region of brainMNI Peak CoordinatesT valueCluster SizeXYZFusiform_L-36-45-214.9531136Calcarine_L-12-4564.4056121Frontal_Inf_Tri_R3912243.8441133SupraMarginal_L-60-24394.2273144Note: Results show brain regions where fALFF values were significantly lower in the MA-abstinent group (DG) compared to the healthy control group (HC) (DG < HC). Statistical thresholds were set at voxel-level p < 0.005 and cluster-level Family-Wise Error (FWE) corrected p < 0.05. Coordinates are based on the Montreal Neurological Institute (MNI) standard space. Only significant clusters passing correction and their peak voxel information are listed. fALFF: fractional amplitude of low-frequency fluctuations Brain regions showing decreased fALFF in MA-abstinent patients (DG) compared to healthy controls (HC) Note: Results show brain regions where fALFF values were significantly lower in the MA-abstinent group (DG) compared to the healthy control group (HC) (DG < HC). Statistical thresholds were set at voxel-level p < 0.005 and cluster-level Family-Wise Error (FWE) corrected p < 0.05. Coordinates are based on the Montreal Neurological Institute (MNI) standard space. Only significant clusters passing correction and their peak voxel information are listed. fALFF: fractional amplitude of low-frequency fluctuations Fig. 2Brain regions showing significantly decreased fALFF in MA-abstinent patients (DG) compared to healthy controls (HC). Note: Compared to the healthy control group (HC), the methamphetamine abstinence group (DG) exhibited significantly decreased fALFF in the left fusiform gyrus, left calcarine cortex, right inferior frontal gyrus (pars triangularis), and left supramarginal gyrus (DG < HC; two-sample t-test). Brain images are displayed in coronal, sagittal, and axial views within MNI space Brain regions showing significantly decreased fALFF in MA-abstinent patients (DG) compared to healthy controls (HC). Note: Compared to the healthy control group (HC), the methamphetamine abstinence group (DG) exhibited significantly decreased fALFF in the left fusiform gyrus, left calcarine cortex, right inferior frontal gyrus (pars triangularis), and left supramarginal gyrus (DG < HC; two-sample t-test). Brain images are displayed in coronal, sagittal, and axial views within MNI space Table 3Brain regions showing decreased ReHo in MA-abstinent patients (DG) compared to healthy controls (HC)Peak region of brainMNI Peak CoordinatesT valueCluster SizeXYZFrontal_Inf_Tri_R5721123.82091236Note: Results show brain regions where regional homogeneity (ReHo) values were significantly lower in the MA-abstinent group (DG) than in the healthy control group (HC) (DG < HC). Statistical thresholds were set at voxel-level p < 0.05 and cluster-level family-wise error (FWE)-corrected p < 0.05. Coordinates are based on the Montreal Neurological Institute (MNI) standard space. This significant cluster was centered in the right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R) and extended to adjacent temporal regions. ReHo, regional homogeneity Table 3Brain regions showing decreased ReHo in MA-abstinent patients (DG) compared to healthy controls (HC)Peak region of brainMNI Peak CoordinatesT valueCluster SizeXYZFrontal_Inf_Tri_R5721123.82091236Note: Results show brain regions where regional homogeneity (ReHo) values were significantly lower in the MA-abstinent group (DG) than in the healthy control group (HC) (DG < HC). Statistical thresholds were set at voxel-level p < 0.05 and cluster-level family-wise error (FWE)-corrected p < 0.05. Coordinates are based on the Montreal Neurological Institute (MNI) standard space. This significant cluster was centered in the right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R) and extended to adjacent temporal regions. ReHo, regional homogeneity Brain regions showing decreased ReHo in MA-abstinent patients (DG) compared to healthy controls (HC) Note: Results show brain regions where regional homogeneity (ReHo) values were significantly lower in the MA-abstinent group (DG) than in the healthy control group (HC) (DG < HC). Statistical thresholds were set at voxel-level p < 0.05 and cluster-level family-wise error (FWE)-corrected p < 0.05. Coordinates are based on the Montreal Neurological Institute (MNI) standard space. This significant cluster was centered in the right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R) and extended to adjacent temporal regions. ReHo, regional homogeneity Fig. 3Brain regions showing significantly decreased ReHo in MA-abstinent patients (DG) compared to healthy controls (HC). Note: The depicted cluster indicates significantly lower regional homogeneity (ReHo) in the DG group than in the HC group (DG < HC). Statistical thresholds were set at voxel-level p < 0.05 and cluster-level family-wise error (FWE)-corrected p < 0.05. The significant cluster was centered in the right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R) and extended to adjacent temporal regions. Brain images are displayed in coronal, sagittal, and axial views within MNI space Brain regions showing significantly decreased ReHo in MA-abstinent patients (DG) compared to healthy controls (HC). Note: The depicted cluster indicates significantly lower regional homogeneity (ReHo) in the DG group than in the HC group (DG < HC). Statistical thresholds were set at voxel-level p < 0.05 and cluster-level family-wise error (FWE)-corrected p < 0.05. The significant cluster was centered in the right inferior frontal gyrus triangular part (Frontal_Inf_Tri_R) and extended to adjacent temporal regions. Brain images are displayed in coronal, sagittal, and axial views within MNI space Voxel-wise functional connectivity (voxelFC) Using the right hippocampus as the seed region, voxel-wise functional connectivity maps were compared between groups using two-sample t-tests. Group differences were thresholded at voxel-level p < 0.005 combined with cluster-level FWE correction at p < 0.05. With the contrast defined as HC minus DG, no regions showed significantly stronger connectivity in the HC group. In contrast, the DG group exhibited significantly enhanced right hippocampal functional connectivity with several regions, including the right inferior temporal gyrus, left inferior occipital gyrus, right thalamus, left angular gyrus, left inferior parietal lobule, right middle frontal gyrus, and left superior medial frontal gyrus (Table 4; Fig. 4). Table 4Group differences in seed-based functional connectivity with the right hippocampus (DG > HC)Peak region of brainMNI Peak CoordinatesT valueCluster SizeXYZTemporal_Inf_R63-51-15-4.115767Occipital_Inf_L-42-66-3-3.437116Thalamus_R9-219-3.394711Angular_L-42-6639-3.526144Parietal_Inf_L-51-2448-3.412128Frontal_Mid_R422145-3.316241Frontal_Sup_Medial_L03048-3.73532Note: Group differences in seed-based functional connectivity were identified using the right hippocampus as the seed region. Statistical thresholds were set at voxel-level p < 0.005 and cluster-level FWE-corrected p < 0.05. The contrast was defined as HC minus DG; therefore, negative T-values indicate significantly stronger functional connectivity in the DG group than in the HC group. HC, healthy controls; DG, methamphetamine-abstinent group; FWE, family-wise error Table 4Group differences in seed-based functional connectivity with the right hippocampus (DG > HC)Peak region of brainMNI Peak CoordinatesT valueCluster SizeXYZTemporal_Inf_R63-51-15-4.115767Occipital_Inf_L-42-66-3-3.437116Thalamus_R9-219-3.394711Angular_L-42-6639-3.526144Parietal_Inf_L-51-2448-3.412128Frontal_Mid_R422145-3.316241Frontal_Sup_Medial_L03048-3.73532Note: Group differences in seed-based functional connectivity were identified using the right hippocampus as the seed region. Statistical thresholds were set at voxel-level p < 0.005 and cluster-level FWE-corrected p < 0.05. The contrast was defined as HC minus DG; therefore, negative T-values indicate significantly stronger functional connectivity in the DG group than in the HC group. HC, healthy controls; DG, methamphetamine-abstinent group; FWE, family-wise error Group differences in seed-based functional connectivity with the right hippocampus (DG > HC) Note: Group differences in seed-based functional connectivity were identified using the right hippocampus as the seed region. Statistical thresholds were set at voxel-level p < 0.005 and cluster-level FWE-corrected p < 0.05. The contrast was defined as HC minus DG; therefore, negative T-values indicate significantly stronger functional connectivity in the DG group than in the HC group. HC, healthy controls; DG, methamphetamine-abstinent group; FWE, family-wise error Fig. 4Brain regions showing enhanced functional connectivity with the right hippocampus in the DG group (DG > HC). All results were FWE-corrected at the cluster level (p < 0.05). Note: Seed-based voxel-wise functional connectivity analysis was performed using the right hippocampus as the seed region. The DG group showed significantly enhanced connectivity with temporal, occipital, thalamic, parietal, and frontal regions compared with the HC group. Results were thresholded at voxel-level p < 0.005 and cluster-level FWE-corrected p < 0.05. HC, healthy controls; DG, methamphetamine-abstinent group; FWE, family-wise error Brain regions showing enhanced functional connectivity with the right hippocampus in the DG group (DG > HC). All results were FWE-corrected at the cluster level (p < 0.05). Note: Seed-based voxel-wise functional connectivity analysis was performed using the right hippocampus as the seed region. The DG group showed significantly enhanced connectivity with temporal, occipital, thalamic, parietal, and frontal regions compared with the HC group. Results were thresholded at voxel-level p < 0.005 and cluster-level FWE-corrected p < 0.05. HC, healthy controls; DG, methamphetamine-abstinent group; FWE, family-wise error Independent component analysis (ICA): Abnormalities in resting-state network functional connectivity Group-level ICA identified several resting-state networks and revealed abnormalities in their intrinsic functional connectivity (Table 5). For the auditory network, the HC group showed significantly stronger functional connectivity in the bilateral precentral/postcentral gyri compared to the DG group (voxel-level p < 0.001, cluster-level FWE corrected; Fig. 5A). For the posterior component of the default mode network (DMN-Posterior), the HC group exhibited stronger connectivity in the right middle temporal gyrus, while for the anterior component of the default mode network (DMN-Anterior), the DG group showed stronger connectivity in the left superior frontal gyrus (all voxel-level p < 0.001, cluster-level FWE corrected; Fig. 5B and C). For the somatomotor network, the HC group showed significantly stronger functional connectivity in the right superior frontal gyrus compared to the DG group (voxel-level p < 0.05, cluster-level FWE corrected; Fig. 5D). Table 5Group differences in intrinsic functional connectivity of resting-state networksComponentContrastThresholdSignificant Brain Region(Peak Voxel)MNI Coordinates(X, Y, Z)tvalueCluster SizeAuditoryHC > DGVoxel p < 0.001, FWE Cluster CorrectedLeft Precentral Gyrus-48, -3, 305.1862AuditoryHC > DGVoxel p < 0.001, FWE Cluster CorrectedRight Postcentral Gyrus51, -6, 274.3043DMN-PosteriorHC > DGVoxel p < 0.001, FWE Cluster CorrectedRight Middle Temporal Gyrus42, -66, 184.2641DMN-AnteriorDG > HCVoxel p < 0.001, FWE Cluster CorrectedLeft Superior Frontal Gyrus-30, 39, 394.6239SomatomotorHC > DGVoxel p < 0.05, FWE Cluster CorrectedRight Superior Frontal Gyrus21, -9, 663.81824 Table 5Group differences in intrinsic functional connectivity of resting-state networksComponentContrastThresholdSignificant Brain Region(Peak Voxel)MNI Coordinates(X, Y, Z)tvalueCluster SizeAuditoryHC > DGVoxel p < 0.001, FWE Cluster CorrectedLeft Precentral Gyrus-48, -3, 305.1862AuditoryHC > DGVoxel p < 0.001, FWE Cluster CorrectedRight Postcentral Gyrus51, -6, 274.3043DMN-PosteriorHC > DGVoxel p < 0.001, FWE Cluster CorrectedRight Middle Temporal Gyrus42, -66, 184.2641DMN-AnteriorDG > HCVoxel p < 0.001, FWE Cluster CorrectedLeft Superior Frontal Gyrus-30, 39, 394.6239SomatomotorHC > DGVoxel p < 0.05, FWE Cluster CorrectedRight Superior Frontal Gyrus21, -9, 663.81824 Group differences in intrinsic functional connectivity of resting-state networks Fig. 5Note: Group differences in intrinsic functional connectivity identified in four independent components: (A) Auditory Network: HC > DG; (B) DMN-Posterior: HC > DG; (C) DMN-Anterior: DG > HC; (D) Somatomotor Network: HC > DG. All results were FWE-corrected at the cluster level (p < 0.05) and rendered in MNI space Note: Group differences in intrinsic functional connectivity identified in four independent components: (A) Auditory Network: HC > DG; (B) DMN-Posterior: HC > DG; (C) DMN-Anterior: DG > HC; (D) Somatomotor Network: HC > DG. All results were FWE-corrected at the cluster level (p < 0.05) and rendered in MNI space Brain-behavior correlation analysis: Relationship between right hippocampal functional connectivity and cognitive function In MA-abstinent patients, the functional connectivity strength between the right hippocampus and several high-level cognitive cortical regions, particularly the right inferior temporal gyrus, was significantly positively associated with overall and specific cognitive performance after controlling for age and years of education (Table 6; Fig. 6). Further interaction analyses indicated that several of these brain–behavior associations were significantly stronger in the DG group than in the HC group, suggesting disease-specific patterns of right hippocampal network coupling and cognitive performance (Table 7). These group-specific brain–behavior patterns are further illustrated in Fig. 7, which presents scatter plots of selected right hippocampal functional connectivity measures and their associations with cognitive performance across multiple domains. Table 6Significant brain-behavior correlations between right hippocampus functional connectivity and cognitive domains (FDR-corrected)Seed RegionTarget RegionCognitive DomainPartial rp (uncorrected)p (FDR)SignificanceHippocampus_RTemporal_Inf_RMoCA0.765< 0.0010.00035***Hippocampus_ROccipital_Inf_LAbstraction0.661< 0.0010.00685**Hippocampus_RAngular_LAttention0.634< 0.0010.00938**Hippocampus_RTemporal_Inf_RAbstraction0.5670.003120.01871*Hippocampus_RTemporal_Inf_RAttention0.5800.002350.01871*Hippocampus_ROccipital_Inf_LMoCA0.5780.002470.01871*Hippocampus_RFrontal_Mid_RMoCA0.5720.002820.01871*Hippocampus_RAngular_LMoCA0.5230.007360.03864*Note: All analyses were performed within the DG group, controlling for age and years of education. Only significant results surviving FDR correction are shown. FDR, false discovery rate; MoCA, Montreal Cognitive Assessment. *p < 0.05, **p < 0.01, ***p < 0.001 after FDR correction Table 6Significant brain-behavior correlations between right hippocampus functional connectivity and cognitive domains (FDR-corrected)Seed RegionTarget RegionCognitive DomainPartial rp (uncorrected)p (FDR)SignificanceHippocampus_RTemporal_Inf_RMoCA0.765< 0.0010.00035***Hippocampus_ROccipital_Inf_LAbstraction0.661< 0.0010.00685**Hippocampus_RAngular_LAttention0.634< 0.0010.00938**Hippocampus_RTemporal_Inf_RAbstraction0.5670.003120.01871*Hippocampus_RTemporal_Inf_RAttention0.5800.002350.01871*Hippocampus_ROccipital_Inf_LMoCA0.5780.002470.01871*Hippocampus_RFrontal_Mid_RMoCA0.5720.002820.01871*Hippocampus_RAngular_LMoCA0.5230.007360.03864*Note: All analyses were performed within the DG group, controlling for age and years of education. Only significant results surviving FDR correction are shown. FDR, false discovery rate; MoCA, Montreal Cognitive Assessment. *p < 0.05, **p < 0.01, ***p < 0.001 after FDR correction Significant brain-behavior correlations between right hippocampus functional connectivity and cognitive domains (FDR-corrected) Note: All analyses were performed within the DG group, controlling for age and years of education. Only significant results surviving FDR correction are shown. FDR, false discovery rate; MoCA, Montreal Cognitive Assessment. *p < 0.05, **p < 0.01, ***p < 0.001 after FDR correction Fig. 6Right hippocampal functional connectivity networks associated with cognitive performance in the DG group. (A) Abstraction domain; (B) attention domain; (C) global cognition indexed by the MoCA total score. Edges represent significant positive partial correlations surviving FDR correction after controlling for age and years of education. Node abbreviations: RHipp, right hippocampus; RITG, right inferior temporal gyrus; LIOG, left inferior occipital gyrus; LAG, left angular gyrus; RMFG, right middle frontal gyrus Right hippocampal functional connectivity networks associated with cognitive performance in the DG group. (A) Abstraction domain; (B) attention domain; (C) global cognition indexed by the MoCA total score. Edges represent significant positive partial correlations surviving FDR correction after controlling for age and years of education. Node abbreviations: RHipp, right hippocampus; RITG, right inferior temporal gyrus; LIOG, left inferior occipital gyrus; LAG, left angular gyrus; RMFG, right middle frontal gyrus Table 7Significant interaction effects between group and right hippocampus functional connectivity on cognitive performanceTarget RegionCognitive DomainInteraction β (95% CI) p HC: rDG: rAngular_LAttention3.92 (1.17, 6.67)0.0060.2090.634**Temporal_Inf_RMoCA9.30 (2.09, 16.51)0.0130.5330.765***Occipital_Inf_LAbstraction1.75 (0.40, 3.10)0.0120.2840.661**Temporal_Inf_RAttention3.33 (0.66, 6.00)0.0160.2030.580*Temporal_Inf_RAbstraction1.42 (0.08, 2.76)0.0380.5210.567*Note: The interaction β-value represents the difference in the slope of the brain-behavior relationship between the DG and HC groups (DG group minus HC group). CI: Confidence Interval. Significance of simple slopes: *p < 0.05, **p < 0.01, ***p < 0.001 Table 7Significant interaction effects between group and right hippocampus functional connectivity on cognitive performanceTarget RegionCognitive DomainInteraction β (95% CI) p HC: rDG: rAngular_LAttention3.92 (1.17, 6.67)0.0060.2090.634**Temporal_Inf_RMoCA9.30 (2.09, 16.51)0.0130.5330.765***Occipital_Inf_LAbstraction1.75 (0.40, 3.10)0.0120.2840.661**Temporal_Inf_RAttention3.33 (0.66, 6.00)0.0160.2030.580*Temporal_Inf_RAbstraction1.42 (0.08, 2.76)0.0380.5210.567*Note: The interaction β-value represents the difference in the slope of the brain-behavior relationship between the DG and HC groups (DG group minus HC group). CI: Confidence Interval. Significance of simple slopes: *p < 0.05, **p < 0.01, ***p < 0.001 Significant interaction effects between group and right hippocampus functional connectivity on cognitive performance Note: The interaction β-value represents the difference in the slope of the brain-behavior relationship between the DG and HC groups (DG group minus HC group). CI: Confidence Interval. Significance of simple slopes: *p < 0.05, **p < 0.01, ***p < 0.001 Fig. 7Group-specific patterns of right hippocampal functional connectivity and its relationship with cognitive function. Note: The scatter plots illustrate five representative brain–behavior associations that were significant in the DG group after FDR correction. These associations also showed significant group-by-functional connectivity interaction effects, as summarized in Table 7. Blue and red points/lines represent the HC and DG groups, respectively. Age and years of education were controlled as covariates in the correlation analyses. HC, healthy controls; DG, methamphetamine-abstinent group; FC, functional connectivity; MoCA, Montreal Cognitive Assessment; FDR, false discovery rate. All reported p-values are FDR-corrected Group-specific patterns of right hippocampal functional connectivity and its relationship with cognitive function. Note: The scatter plots illustrate five representative brain–behavior associations that were significant in the DG group after FDR correction. These associations also showed significant group-by-functional connectivity interaction effects, as summarized in Table 7. Blue and red points/lines represent the HC and DG groups, respectively. Age and years of education were controlled as covariates in the correlation analyses. HC, healthy controls; DG, methamphetamine-abstinent group; FC, functional connectivity; MoCA, Montreal Cognitive Assessment; FDR, false discovery rate. All reported p-values are FDR-corrected Discussion This study employed a multimodal resting-state functional magnetic resonance imaging (rs-fMRI) analysis framework to systematically characterize brain functional abnormalities associated with cognitive impairment in methamphetamine (MA) abstinence patients. Previous research has confirmed that MA use disorder can lead to persistent cognitive dysfunction, which remains difficult to fully recover from during the abstinence period [2–4]. By integrating local brain activity metrics (fALFF, ReHo), seed-based functional connectivity, and large-scale brain network analysis, we found evidence of hierarchical brain functional reorganization in MA abstinence patients, with the right hippocampus playing a crucial hub role in the relationship between brain functional abnormalities and cognitive impairment. Importantly, because only male participants were included, these findings should be interpreted as reflecting male-specific patterns of brain functional abnormalities and brain-behavior associations during MA abstinence. Abnormal local brain activity and synchronicity reflect persistent cortical functional impairment Our study found that MA abstinence patients exhibited significant abnormalities in fALFF and ReHo values in multiple brain regions. Specifically, compared to the healthy control group (HC), the DG group showed significantly reduced fALFF in the left fusiform gyrus, left calcarine cortex, right inferior frontal gyrus, pars triangularis, and left supramarginal gyrus (Table 2; Fig. 2), and significantly reduced ReHo in the right inferior frontal gyrus, pars triangularis (Table 3; Fig. 3). These results suggest a general reduction in local neural activity and synchronicity in these regions in MA abstinence patients. fALFF reflects the intensity of spontaneous local neural activity [16], while ReHo measures the synchronicity of local neuronal activity [17]. The abnormal activity and synchronicity disturbances in these brain regions may be related to MA’s damage to the central dopaminergic system [18], and even other neurotransmitter systems(such as the glutamatergic system) [19]. The dopamine system plays a crucial role in regulating cortical excitability and neuronal synchronicity [20], and its dysfunction can directly affect local activity in relevant brain regions. The inferior frontal gyrus [21, 22], fusiform gyrus [23], calcarine cortex [24] and supramarginal gyrus [25]are widely involved in higher cognitive functions, including executive control, visual information processing, and multimodal information integration. Reduced local activity in these regions suggests that their functional efficiency may be impaired [26, 27], thus constituting an important neural basis for cognitive impairment during MA abstinence. Notably, previous studies have also reported reduced ReHo or fALFF in frontal and temporal regions in MA abstainers [28, 29], with some abnormalities persisting even after long-term abstinence. The results of this study are highly consistent with these findings, further supporting the persistence of brain functional abnormalities in MA abstinence patients. Furthermore, some studies have found both structural and functional abnormalities in frontal and temporal regions in MA-related psychosis patients [30], suggesting cross-diagnostic consistency in MA-related brain damage. This evidence collectively indicates that the neurotoxic effects of MA use disorder may cause long-term impacts on brain structure and function, thereby affecting cortical information processing efficiency. However, we also need to consider the dynamic nature of brain functional changes after MA abstinence. Longitudinal studies have found that after long-term abstinence, ALFF and ReHo values in the right middle frontal gyrus may actually significantly increase in MA abstainers, and positively correlate with impulsivity scores [31]. This apparent “contradiction” may suggest that brain functional changes after MA abstinence are not a linear recovery process, and different brain regions may experience different adaptive or damaging changes at different stages of abstinence. This cross-sectional study primarily reflects the “reduced” state of abstinence patients compared to healthy controls; future longitudinal studies will help delineate this dynamic trajectory and the association of other cognitive or behavioral indicators (such as decision-making ability, emotional regulation) with these dynamic changes. Together, the fALFF and ReHo findings shown in Figs. 2 and 3 indicate convergent local cortical abnormalities, particularly involving the right inferior frontal gyrus, in MA-abstinent patients. Widely enhanced hippocampal-centric functional connectivity as inefficient compensatory hyperconnectivity As shown in Fig. 4; Table 4, the right hippocampus exhibited enhanced functional connectivity with multiple temporal, parietal, frontal, and subcortical regions in the DG group. Our findings support the value of the present integrated analytical approach. While ReHo and fALFF analyses revealed localized reductions in spontaneous activity and regional synchronization in cortical regions, the seed-based functional connectivity analysis further demonstrated enhanced connectivity between the right hippocampus and multiple cortical and subcortical regions in MA-abstinent patients. Specifically, increased connectivity was observed between the right hippocampus and the right inferior temporal gyrus, left angular gyrus, left inferior parietal lobule, right middle frontal gyrus, left superior medial frontal gyrus, as well as the right thalamus (Table 4). Taken together, these findings suggest that the hippocampus may serve as an important hub in large-scale functional reorganization during MA abstinence. The hippocampus plays a central role in memory consolidation, contextual learning, and episodic processing [32, 33]. Enhanced connectivity between the hippocampus and association cortices in the temporal, parietal, and frontal lobes may therefore reflect increased recruitment of memory-related pathways when broader cortical efficiency is compromised [34, 35]. This interpretation is broadly consistent with previous evidence suggesting that, under conditions of neural injury or dysfunction, increased functional coupling may emerge as a compensatory response aimed at maintaining cognitive output [34, 35]. In addition, prior studies in MA abstainers have reported dissociations between local intrinsic activity and broader network-level alterations, including abnormal static and dynamic ALFF patterns across multiple brain regions [36]. The coexistence of reduced local activity and enhanced long-range connectivity observed in the present study is consistent with this broader pattern of network reorganization. Importantly, however, we do not interpret the observed hippocampal hyperconnectivity as evidence of successful or fully adaptive compensation. Rather, the present findings are more parsimoniously understood within a framework of inefficient compensatory hyperconnectivity. In this framework, increased hippocampal-cortical connectivity may reflect the brain’s attempt to preserve cognitive performance in the setting of widespread cortical dysfunction, but this reorganization is likely achieved in a less flexible, more resource-demanding, and potentially more vulnerable manner [35]. Thus, the observed hyperconnectivity may not indicate restored network efficiency; instead, it may reflect a state in which cognitive processing becomes increasingly dependent on a limited set of hippocampal-centered pathways. Given the cross-sectional design, the present data cannot determine whether this pattern initially emerges as an adaptive response and later becomes maladaptive over time. Nevertheless, our results suggest that framing this pattern as either purely compensatory or purely pathological would be overly simplistic. A prominent finding of this study was that stronger functional connectivity between the right hippocampus and the right inferior temporal gyrus was positively associated with global cognitive performance, as indexed by the MoCA total score, in the MA-abstinent group, and that this association was significantly stronger than that observed in healthy controls. The corresponding brain–behavior associations are illustrated in Figs. 6 and 7, with the statistical results summarized in Tables 6 and 7. The inferior temporal gyrus is involved in high-level visual processing, object recognition, and memory-related information integration [37, 38]. Its functional coupling with the hippocampus may therefore be associated with the integration of perceptual and mnemonic processes relevant to cognitive performance. However, the fact that this brain-behavior association was markedly stronger in MA-abstinent patients than in healthy controls suggests not simply preserved function, but a greater reliance on this specific pathway under conditions of broader network inefficiency. In healthy individuals, cognitive performance is typically supported by distributed and relatively redundant large-scale networks [39, 40]. Under such conditions, the integrity of cognition does not depend excessively on any single pathway. In contrast, in MA abstainers, widespread cortical dysfunction and altered network architecture may reduce this redundancy, making cognitive performance more dependent on a restricted subset of hippocampal-cortical connections [41, 42]. From the perspective of neural reserve and compensation, such abnormally strengthened brain-behavior coupling may indicate not resilient network function, but rather a vulnerable compensatory configuration in which limited remaining pathways carry a disproportionate functional burden [43, 44]. Similar patterns of hyperconnectivity associated with network inefficiency or vulnerability have also been reported in addiction-related conditions and other neuropsychiatric disorders [45, 46]. Therefore, the increased hippocampal-centered connectivity observed here is best interpreted not as a marker of healthy compensation, but as a form of inefficient compensatory reorganization that may temporarily support cognition while also reflecting reduced flexibility and increased network burden. Taken together, these findings indicate that the right hippocampus may function as a key hub in the reorganization of brain networks during MA abstinence. More broadly, across local activity (ReHo/fALFF), hippocampal functional connectivity, and ICA-derived network abnormalities, the present results converge on hippocampus-centered network reorganization as a plausible core mechanism underlying cognitive vulnerability during MA abstinence. In this view, local cortical hypoactivity, enhanced hippocampal coupling, and large-scale network imbalance should be understood as interconnected manifestations of a broader functional reorganization process rather than isolated abnormalities. Overall imbalance in the default mode network and sensorimotor network Independent component analysis (ICA) revealed functional connectivity abnormalities in multiple large-scale brain networks in MA abstinence patients. These ICA-derived abnormalities are illustrated in Fig. 5A–D and summarized in Table 5. The default mode network (DMN), as a core network for self-referential processing, episodic memory, and future planning, its functional imbalance is widely present in various mental illnesses [47]. This study found that in the posterior component of the DMN, the HC group had stronger connectivity in the right middle temporal gyrus, while in the anterior component of the DMN, the DG group had stronger connectivity in the left superior frontal gyrus. This “anterior connectivity enhancement, posterior connectivity reduction” dissociated pattern suggests impaired DMN internal coordination, which may be closely related to abnormal self-referential processing, enhanced rumination, and impaired cognitive control in MA abstainers [48]. This impaired coordination may lead to difficulty switching between introspection and external tasks, affecting cognitive flexibility. Additionally, functional connectivity in the auditory network and sensorimotor network also showed reduction. The HC group had significantly stronger functional connectivity in the bilateral precentral/postcentral gyri compared to the DG group; the HC group had significantly stronger functional connectivity in the right superior frontal gyrus compared to the DG group (Table 5). This suggests that MA-related brain damage is not limited to higher-level cognitive systems but also affects sensory processing and motor control networks. This widespread neural network abnormality is consistent with previous reports of widespread abnormalities in multiple resting-state networks in individuals with MA use disorder [30, 45], further supporting the view of whole-brain scale network imbalance in this study. Functional dysregulation of these networks may lead to impaired sensory information processing, reduced motor coordination, and attention deficits, thereby further exacerbating cognitive dysfunction during MA abstinence. To further examine the convergence across different rs-fMRI metrics, we qualitatively compared the spatial distribution of significant findings from fALFF, ReHo, and ICA analyses. A partial overlap was observed between fALFF and ReHo results in the right inferior frontal gyrus, pars triangularis, suggesting that this region may show both reduced local spontaneous activity and impaired regional synchronization in MA-abstinent patients. In contrast, the ICA-derived network abnormalities did not show complete voxel-wise overlap with the fALFF or ReHo peak regions, but they involved anatomically and functionally related frontal and temporal network regions. These findings indicate that the three approaches captured complementary levels of functional alteration, ranging from local activity and regional synchronization to large-scale network imbalance. Exploratory associations between MoCA subdomains and hippocampal connectivity The brain-behavior correlation analysis suggested exploratory associations between MoCA subdomain scores and distinct hippocampal-cortical connectivity measures. However, because these subdomain scores were derived from a brief screening instrument rather than a comprehensive neuropsychological battery, these findings should be interpreted cautiously and not as definitive evidence of domain-specific neural mechanisms. First, overall cognitive function (MoCA total score) showed the strongest positive correlation with the strength of functional connectivity between the right hippocampus and the right inferior temporal gyrus (r = 0.765) (Fig. 6C). The inferior temporal gyrus plays a key role in high-level visual processing, object recognition, and semantic memory [49–51]. The strength and disease specificity of this association indicate that in MA abstinence patients, overall cognitive performance was closely associated with information exchange between the hippocampus and the visual association cortex in the temporal lobe. This may reflect that, against a backdrop of reduced functional efficiency across widespread cortical areas, patients are forced to over-recruit this relatively “preserved” or “compensatorily strengthened” pathway to complete complex tasks requiring the integration of memory and visual information. However, this “strong coupling” may essentially represent a fragile compensatory state: once this connection fails due to overload or further damage, overall cognitive function may decline precipitously. Second, impairment in abstract thinking ability was significantly correlated with enhanced functional connectivity between the right hippocampus and the left inferior occipital gyrus(Fig. 6A). The inferior occipital gyrus is involved in early visual form processing [52, 53], The enhancement of its connectivity with the hippocampus may suggest that patients abnormally rely on converting abstract problems into concrete visual mental imagery during abstract reasoning tasks. Although this “detour” neural strategy may be associated with task performance to some extent, it is likely inefficient and consumes neural resources that could otherwise be allocated to other cognitive processes, thereby limiting the flexibility and complexity of abstract thinking overall. This study also found that attention function was associated with functional connectivity between the right hippocampus and the left angular gyrus (Fig. 6B). The angular gyrus is a key hub connecting the default mode network (DMN) and the dorsal attention network (DAN), responsible for integrating memory, attention, and spatial information [54]. Pathological enhancement of hippocampus-angular gyrus connectivity may disrupt the normal anti-correlated relationship between the DMN and DAN [55], This pattern may be associated with reduced ability to suppress DMN-mediated introspective thoughts (e.g., memories of drug craving) when focusing on external tasks, manifesting as distractibility and difficulty sustaining attention. This pattern may tentatively suggest altered network interaction relevant to attentional difficulties in MA abstinence, but this interpretation remains preliminary and requires validation using more specific attention measures. These preliminary patterns may be useful for hypothesis generation regarding brain-cognition relationships in MA abstinence, but they should not be interpreted as establishing precise domain-specific mapping. It indicates that the neural damage induced by MA is not a uniform “global downregulation of function” but rather selectively disrupts the cognitive system in a network-specific manner. Different cognitive functions, depending on their underlying neural substrates, exhibit differential vulnerability to the neurotoxicity of MA. The association pattern discovered in this study offers a new perspective on the “cognitive reserve” theory: in MA abstainers, limited neural resources may be reallocated to form several highly specialized yet rigid “cognitive compensatory pathways” to support specific cognitive dimensions. While such compensation may sustain behavioral performance to some degree, it does so at the cost of network flexibility, robustness, and overall efficiency, and may represent a neural correlate of cognitive vulnerability. From a clinical translation perspective, these findings may help generate hypotheses for future cognitive rehabilitation and neuromodulation studies. Rather than supporting direct treatment recommendations, the present results suggest that hippocampal-centered connectivity abnormalities may be relevant targets for further investigation. In particular, these connectivity patterns may serve as candidate imaging markers for patient stratification or longitudinal monitoring, although their predictive value and therapeutic relevance remain to be established in larger prospective studies. Future work should test whether these fine-grained associations are reproducible and whether they are modifiable through cognitive training or neuromodulatory approaches.Although several FC–cognition associations reached statistical significance, the variation in cognitive scores across FC values was relatively limited in some scatter plots, except for the association involving Occipital_Inf_L. This was particularly evident for MoCA subdomain scores. This may be partly attributable to the restricted score range and ceiling effects of MoCA subdomains, as well as the modest sample size. Therefore, these results should be interpreted as exploratory evidence of subtle brain–behavior associations rather than as indicating large clinical changes in cognitive performance. Future studies using comprehensive neuropsychological batteries are needed to validate the clinical magnitude and domain specificity of these associations. Clinical significance Given the disease-specific associations observed in this study, right hippocampal-centered networks may represent candidate neuroimaging markers of cognitive vulnerability during MA abstinence. However, these findings should be interpreted cautiously, as the present study was cross-sectional and cannot establish causality. Future longitudinal and interventional studies are needed to determine whether these network abnormalities have predictive value or can inform treatment targeting. Taken together, these findings highlight the potential translational value of hippocampal network alterations. The hippocampus is a core hub within memory-related and reward-associated networks, and its functional connectivity patterns have been consistently implicated in substance use disorders. In the context of methamphetamine abstinence, hippocampal-centered network alterations may reflect persistent dysfunction in contextual memory processing and cue–reward associations, which are closely linked to craving and relapse vulnerability. Therefore, abnormal hippocampal functional connectivity may serve as a potential imaging biomarker for characterizing disease severity and cognitive impairment during abstinence. Importantly, hippocampal network alterations may also have translational relevance as a potential therapeutic target. Interventions that modulate hippocampal circuit activity, including cognitive training, exercise, and neuromodulation approaches, may contribute to the restoration of network balance and cognitive function. In this regard, the observed associations between hippocampal connectivity and cognitive performance suggest that hippocampal networks may provide a useful target for monitoring treatment response and recovery trajectories in methamphetamine abstinent populations. Limitations While this study yielded important findings, several limitations should be acknowledged. First, the sample size was relatively small (HC group 29, DG group 25), which may limit statistical power, reduce generalizability, and preclude more detailed subgroup analyses. In addition, all participants were male; therefore, the present findings should be interpreted specifically as reflecting neural patterns in male MA-abstinent patients and cannot be generalized to female patients or mixed-sex populations. Future studies should include female participants to determine whether the observed alterations are sex-specific or shared across sexes.Second, this study employed a cross-sectional design, which cannot reveal the dynamic evolution of brain functional abnormalities during MA abstinence or establish causal relationships. Future longitudinal studies are needed to track brain functional changes and cognitive recovery across different abstinence stages.Third, detailed methamphetamine use history was not systematically available, including cumulative dosage, frequency of use, duration of use, and other indices reflecting addiction severity. Consequently, dose-dependent analyses and severity-stratified analyses could not be performed. This limitation is important because inter-individual variability in prior MA exposure and addiction severity may influence both neuroimaging findings and cognitive performance, thereby introducing heterogeneity into the patient group and limiting the interpretability of abstinence-related effects.Fourth, although a family history of psychiatric disorders was excluded, undetected comorbidities might still exist, which could confound the results. Fifth, the Montreal Cognitive Assessment (MoCA) was used as a global cognitive screening instrument rather than a comprehensive domain-specific neuropsychological battery. Accordingly, the MoCA total score is more appropriate for indexing overall cognitive status, whereas individual subdomain scores have more limited psychometric precision, restricted score ranges, and may be particularly vulnerable to ceiling effects in healthy populations. This was evident in the present study, where healthy controls showed near-ceiling performance on the abstraction subscale (1.93 ± 0.26 out of 2), which may have constrained variability and influenced correlation estimates. Therefore, all analyses involving MoCA subdomains should be considered exploratory and hypothesis-generating rather than definitive evidence of domain-specific cognitive impairment or domain-specific brain-behavior relationships. Future studies should incorporate more comprehensive and standardized neuropsychological measures to validate the specificity of the cognitive patterns suggested here. In addition, the seed-based connectivity analysis focused only on the right hippocampus. Although this choice was hypothesis-driven, inclusion of bilateral hippocampal ROIs would provide a more comprehensive assessment of hippocampal network alterations and help determine whether the observed effects are lateralized or reflect broader hippocampal dysconnectivity. Finally, while the multimodal analysis in this study provided a comprehensive perspective, how to integrate these different levels of information more deeply to construct a unified neurophysiological model remains a challenge for future research. In addition, all participants were male; therefore, the present findings should be interpreted specifically as reflecting neural patterns in male MA-abstinent patients and cannot be generalized to female patients or mixed-sex populations. Future studies should include female participants to determine whether the observed alterations are sex-specific or shared across sexes. Conclusion and future outlook This study comprehensively characterized abnormalities in local brain activity, voxel-level functional connectivity, and large-scale resting-state networks in male MA-abstinent patients, and identified disease-specific associations between right hippocampal functional connectivity and cognitive performance. These findings suggest that right hippocampal-centered networks may be associated with cognitive performance during MA abstinence through an inefficient compensatory reorganization pattern. Future research should focus on the following aspects: Larger-scale, longitudinal studies: To track dynamic changes in brain functional abnormalities to reveal neural adaptation and recovery processes during MA abstinence. Multimodal data integration: To combine more multimodal data, such as structural imaging, metabolic imaging, neurotransmitter levels, and gene polymorphisms, to provide a more comprehensive neurobiological view. Application of artificial intelligence and machine learning: To develop cognitive impairment prediction models based on multimodal fMRI data, aiming for early diagnosis and precise intervention. Neuromodulation intervention research: To conduct neuromodulation intervention studies targeting specific abnormal brain regions and networks, hoping to provide new treatment strategies for cognitive rehabilitation in MA abstinence patients.