Thirty male C57BL/6 mice (10–12 weeks old) were used with approval from the Institutional Animal Care and Use Committee of Sungkyunkwan University. Because sex-dependent differences have are well documented in CORT levels (Fahrenkrug et al., 2012; Gong et al., 2015; Lindhardt et al., 2022; Thanos et al., 2009) and FC connectivity (Biswal et al., 2010; Satterthwaite et al., 2015; Tian et al., 2011; Wang et al., 2012; Wu et al., 2013; Zuo et al., 2010), only male mice were included in this study to reduce variability and thereby minimize the number of animals required. All experiments were performed in accordance with the standards for humane animal care and in compliance with the guidelines of the Animal Welfare Act and the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

The head-post (Figure 1A), made of acrylonitrile butadiene styrene (ABS), was 0.75–1.3 mm thick, and was slightly curved to ensure tight contact with the skull (Choi et al., 2025). A detailed head-post implantation protocol is described in our previous publications (Choi et al., 2025; Shim et al., 2022). In short, animals were initially anesthetized with isoflurane (ISO) anesthesia at 3 ~ 4%, then maintained at 1.5 ~ 2.0% during the stereotaxic surgery for head-post placement. After placing the mouse in a stereotactic frame, the hair was shaved, and 2% lidocaine hydrochloride was subcutaneously injected under the scalp. All soft tissues were removed and cleaned with saline-soaked cotton buds. After ensuring the skull was completely clean, the ~0.2-g head-post was placed on the exposed skull. All the exposed regions, including the head-post, were covered with a smooth layer of dental cement (Model #: C&B, Sun Medical Co, Shiga, Japan) 3 ~ 4 times to reduce distortion in the MRI images caused by air. After the surgery, the mice were allowed to recover from anesthesia and then returned to their cages.

Two weeks after head-post implantation, mice were randomly assigned to three groups (n = 10 per group): no habituation (Group 1), habituation outside the magnet (Group 2), and habituation inside the magnet during fMRI scanning (Group 3). To monitor stress-related physiological changes, body weight was measured on days 1–8, 10 (2 days post-habituation), 12 (4 days post-habituation), 16 (8 days post-habituation), and 23 (15 days post-habituation), with day 1 corresponding to the first day of habituation.

To restrain each mouse in Groups 2 and 3, we used a custom-designed cradle with a body frame and a detachable body cover that conformed to the mouse’s body and included open slots at the bottom to allow paw movement (Figure 1A), thereby reducing stress (Choi et al., 2025). Isoflurane anesthesia was briefly administered to minimize stress during the positioning process, with induction at 5% ISO and maintenance at 2% ISO via a nose cone. After the mouse was positioned in the cradle, its mouth was secured to a tooth bar integrated into the nose cone. The entire procedure lasted less than 5 min. Daily 2-h habituation or scanning sessions were then conducted under awake conditions on days 1–8, followed by an additional scanning session on day 23 (15 days post-habituation) (Figure 1B).

For Group 2, MRI acoustic noise was not presented, as previous studies have shown that EPI scanning noise is not a major stressor in C57BL/6 mice (Almeida et al., 2022; Xu et al., 2022). This allowed multiple mice to be habituated simultaneously without the need for a dedicated soundproof mock scanner room.

To repeatedly measure plasma CORT levels in the same mice, we followed the blood collection and CORT assay procedures described by Xu et al. (2022). Plasma CORT concentrations were measured on days −3, 1, 3, 5, and 8 within 20 min after each habituation or scanning session for Groups 2 and 3, and again on day 16 (8 days post-habituation) to confirm the return of CORT levels to baseline (Figure 1B). For Group 1 mice, CORT measurements were performed on the same schedule. CORT measurements were conducted in the morning for Groups 1 and 2, whereas for Group 3 they were performed in the afternoon on day −3 and during the daytime on all other days.

Briefly, under 1–1.5% isoflurane anesthesia, approximately 40 μL of blood was collected from the saphenous vein using heparin-coated glass capillaries sealed at one end (Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany). One mouse in Group 3 died due to anesthesia complications during blood sampling; thus, nine mice were included in the final analysis for that group.

Blood-filled capillaries were placed in 1.5 mL Eppendorf tubes on ice and centrifuged (Model: 5424R, Eppendorf, Hamburg, Germany) at 10,000 rpm for 10 min. The clear plasma layer (15–20 μL) was carefully transferred to PCR tubes (PCR-02-C, AxyGen, Union City, CA, USA) and stored at −20 °C until further analysis. After the completion of all experiments, plasma samples were analyzed for CORT concentration using an enzyme-linked immunosorbent assay (ELISA) kit (DRG-4164, DRG Instruments GmbH, Marburg, Germany).

Two-hour MRI experiments were conducted daily on days 1–8 using a 15.2 T MRI system (BioSpec, Bruker, Ettlingen, Germany), with an additional fMRI session performed on day 23. The system was equipped with an 11 cm horizontal-bore magnet and an actively shielded gradient coil (6 cm inner diameter, maximum strength 100 G/cm, rise time 110 μs). A 12 mm inner-diameter surface coil was used for both transmission and reception. The mouse brain was positioned at the magnet isocenter, and field inhomogeneity was minimized using the MAPSHIM protocol with an ellipsoidal shim volume covering the cerebrum (ParaVision 360, Bruker BioSpin, USA).

Detailed MRI procedures have been described previously (Choi et al., 2025). Anatomical images were acquired using a fast low-angle shot (FLASH) sequence with the following parameters: matrix size = 256 × 128, field of view (FOV) = 15.0 × 7.5 mm² (0.059 × 0.059 mm² in-plane resolution), 20 slices (0.5 mm thickness), repetition time/echo time (TR/TE) = 390/3.4 ms, and number of averages = 2. Resting-state functional images were obtained using gradient-echo single-shot echo-planar imaging (EPI) with the following parameters: matrix size = 96 × 48, FOV = 15.0 × 7.5 mm² (0.156 × 0.156 mm² in-plane resolution), 20 slices (0.5 mm thickness), TR/TE = 1000/11.7 ms, and 600 volumes (10 min per run). Each daily session consisted of five 10-min runs (representative EPI images for all animals are shown in Supplementary Figure 1).

During all experiments, physiological signals were continuously monitored using a physiological monitoring system (Model 1,030, SA Instruments, Inc., Stony Brook, NY, USA) and recorded using a data acquisition system (Biopac Systems, Inc., Goleta, CA, USA). Body motion and respiration rate were monitored using a pressure sensor (Model RS-301) placed on the ventral surface.

Since respiration rate is indicative of stress level, respiration (breaths per minute, BPM) was estimated from pressure sensor signals recorded using the Biopac system, which captured respiration-induced chest movements. The average BPM for each animal was calculated from a 2000-s segment (approximately 33 min) acquired near the midpoint of each session by counting the sinusoidal respiratory cycles per minute in the signal.

Slice-time-corrected rsfMRI data were used to estimate six head-motion parameters (three translations and three rotations) during the image realignment step. Subsequently, frame-wise displacement (FD) was calculated as a head-motion index by summing the absolute displacements of the six motion parameters, assuming a spherical mouse brain with a radius of 5 mm (Power et al., 2012). Among the 9 animals across 9 daily sessions, three sessions showed unusual motion along the phase-encoding direction, likely due to technical issues related to phase locking and unstable shimming in the Bruker software; therefore, these sessions were excluded from further analyses. For each run, an FD time course was generated. To assess the impact of habituation, motion histograms and mean FD values across days and runs were computed.

All FLASH anatomy data and BOLD time series EPI data for each mouse were analyzed using several tools and software packages, including the Analysis of Functional Neuroimages package (Cox, 1996), FMRIB Software Library (FSL), Advanced Normalization Tools (ANTs) (Avants et al., 2011), and Matlab codes (MathWorks, USA). EPI images were co-registered to anatomical images and subsequently to the Allen Mouse Brain Atlas, following procedures described in detail in our previous studies (Choi et al., 2025). Briefly, individual EPI images were first aligned to their corresponding T₂^*-weighted anatomical images using linear transformation. These co-registered images were then normalized to a mouse brain template using nonlinear transformations and subsequently aligned to the Allen Mouse Brain Atlas.

Preprocessing of the acquired data included slice-timing correction, spike removal, motion correction, regression of motion components (12 regressors) as well as global and CSF signals, spatial smoothing with a 0.5 mm full width at half maximum, detrending, and band-pass filtering (0.01–0.25 Hz). The use of 12 motion parameters (six first-order parameters and their six temporal derivatives) followed prior studies on motion correction (Chen et al., 2020; Dinh et al., 2021). CSF regions were automatically selected based on the criterion that voxel intensities exceeded half of the maximum intensity. Although white matter signals are often regressed in rsfMRI analyses, we did not include them here because white matter could not be reliably identified in our data.

Brain region-specific ROIs for quantitative analysis were defined according to the Allen Mouse Brain Atlas (https://atlas.brain-map.org/). Three analyses were performed:

All results are presented as the mean and standard deviation (SD). The statistical significance was tested using ANOVA and t tests. For all statistical tests, significance was considered according to an FDR-corrected p-value <0.05.

To assess stress levels during the habituation process, we measured body weight and plasma CORT in three experimental groups, and respiratory rate in the fMRI group. Mean baseline body weights on day 1 were 27.1 ± 2.35 g (n = 10), 28.4 ± 2.36 g (n = 10), and 27.9 ± 3.35 g (n = 9) for the control (Group 1), outside-the-magnet habituation (Group 2), and fMRI habituation groups (Group 3), respectively (Figure 2A). In the control group, body weight remained stable for the first 6 days and then gradually increased, although an earlier onset of weight gain had been expected. This delayed increase was likely due to stress associated with repeated blood sampling for CORT analysis. Both habituation groups showed a similar ~10% loss of body weight during the 8-day habituation period, followed by weight gain after habituation ended, indicating comparable stress levels between the two groups.

CORT levels were measured before, during, and one week after the 8-day habituation period (Figure 2B). Baseline CORT levels were 160.1 ± 100.5 ng/mL, 169.1 ± 75.7 ng/mL, and 216.9 ± 145.6 ng/mL (mean ± SD) for Group 1, Group 2, and Group 3, respectively. Although no statistically significant differences were observed among groups, baseline CORT levels in Group 3 tended to be higher than those in Groups 1 and 2, potentially reflecting circadian-related variability associated with differences in sampling time and/or increased variability due to the limited sample size. Two animals exhibited baseline CORT levels >400 ng/mL, underscoring the high sensitivity of CORT to handling stress.

Habituation increased CORT levels by approximately twofold in the restraint group and two- to fourfold in the fMRI habituation group. The average CORT levels from days 1–8 were 108.8 ± 55.6 ng/mL (n = 10 animals × 4 days), 301.8 ± 89.3 ng/mL (n = 10 animals × 4 days), and 509.8 ± 257.0 ng/mL (n = 9 animals × 4 days) for Group 1, Group 2, and Group 3, respectively. During the habituation period, the fMRI habituation group consistently exhibited higher CORT levels than the restraint group, except on day 3. The highest CORT levels were observed on day 1 in the fMRI group. One week after the habituation procedures, CORT levels returned to baseline in all groups, indicating that the 8-day habituation regimen did not induce lasting chronic stress.

Respiratory rates in the magnet-habituation group were 238.6 ± 25.4 breaths per minute (BPM) during the first fMRI scan and averaged 216.5 ± 7.9 BPM across sessions (n = 74 days sessions in 9 animals). No significant differences were observed across days (Figure 2C). Overall, respiratory rates remained stable at approximately 210–220 BPM, which falls within the normal range reported in the literature (210–270 BPM) (Friedman et al., 2004; Han et al., 2001; Pamenter and Powell, 2016; Schwarte et al., 2000; Tsukamoto et al., 2015). This range is consistent with our previous observation using the same cradle setup (227.3 ± 28.3 BPM) (Choi et al., 2025), but lower than values obtained with an earlier version of the cradle (260–300 BPM) (Dinh et al., 2021). Despite exhibiting normal respiratory rates, awake mice undergoing fMRI scanning still exhibited signs of stress based on body weight loss and elevated CORT levels.

Because stressed mice tend to move more, habituation is commonly used to reduce head motion, particularly in fMRI studies. In our study, although CORT levels remained elevated throughout the 8-day fMRI scanning period, it is still possible that habituation reduced motion during later sessions. Therefore, we compared FD across days and across runs within each session.

Figures 3A,C shows FD time courses averaged across five runs from nine animals on the first fMRI day (without prior habituation) and on day 8 (after seven habituation sessions). FD fluctuations were observed to a similar extent on both days. Histograms illustrate the distributions of FD values for all runs (600 volumes/run × 5 runs × 9 animals) on day 1 (Figure 3B) and day 8 (Figure 3D). The average FD values were 28.0 ± 21.1 μm and 33.8 ± 24.9 μm for day 1 and day 8, respectively, indicating that head motion during fMRI was comparable without or with one week of habituation.

Average FD values were plotted across days 1–8 and day 23 (15 days after the end of habituation) (Figure 3E), as well as across the five consecutive runs within each session (Figure 3F). The average FD value in day 8 was significantly higher than the average FD value in Day 4 (p-value = 0.02) and 6 (p-value = 0.01). No decreasing trends in head motion were observed across successive sessions or runs, indicating that head motion remained stable when the restraint system was well designed to reduce stress levels, such as by allowing free paw movements.

To assess whether rsfMRI maps were reliably detected in our data, correlation maps were generated from rsfMRI data on day 1 using seed regions in the bilateral ACC, left SSp, left hippocampal region, and left striatum. Correlation maps obtained from the ACC seed resembled the DMN, while maps generated from the SSp, hippocampal region, and striatum exhibited robust bilateral connectivity (see Figures 4A,B for DMN and somatosensory network examples; and Supplementary Figure 2 for hippocampal region and striatum). These rsfMRI connectivity patterns closely resembled those reported in the literature (Gutierrez-Barragan et al., 2022).

To assess the specificity of FC, we adopted an index proposed by Grandjean et al. (2023), defined as the ratio of bilateral SSp connectivity to the correlation between the SSp and ACC. As expected, correlations between interhemispheric SSp were higher than those between SSp and the ACC (Figure 4C; individual sessions). FC specificity was calculated the logarithm of the absolute value of the ratio between bilateral SSp and SSp-ACC correlation coefficients, and remained consistent across sessions with different levels of habituation (Figure 4D). An FC specificity value of 1.0 indicates that bilateral SSp connectivity is approximately 2.7 times stronger than SSp-ACC connectivity.

We next examined whether resting-state functional connectivity changed during the habituation process, despite the similar CORT levels, head motion and FC specificity values observed across days. Cross-correlation analysis was performed among 20 ROIs in both hemispheres, with 10 ROIs per each hemisphere. Figure 5A shows the FC matrices from days 1 and 8. No obvious changes were detected over time (Figure 5B). For improved visualization, −log (p-value) was plotted, where a p-value of 0.05 without FDR correction corresponds to ~3.0. Inter-hemispheric functional connectivity, a hallmark feature of rsfMRI FC, was also evaluated across days (Figure 5C), and no significant changes were observed. These results indicate that resting-state FC remained stable throughout habituation.

Multiple stressors are associated with awake mouse fMRI, including head and body restraint to minimize motion artifacts, as well as acoustic and vibrational noise. To reduce stress under these conditions, two primary strategies have been explored: (1) using less restrictive restraint systems, and (2) performing habituation under conditions that closely mimic the actual experimental environment.

Head motion is an important indicator of both stress and fMRI data quality. In our experiments, head motion remained similar across all sessions, with FD values of approximately 29 μm, regardless of habituation. This value is slightly higher than that reported in our previous study (20 μm FD) (Choi et al., 2025). The exact source of this difference is unclear but may reflect variability across animal batches. Although acclimation is generally expected to reduce head motion (Dinh et al., 2021), we did not observe decreases in head motion over days, consistent with the lack of changes in stress-related measures. This is consistent with recent observations comparing 4-day and 13-day habituated mice [Figure 2C in Laxer et al. (2025)].

To reduce motion-related artifacts in fMRI, data scrubbing is often applied in awake fMRI analyses. However, because our goal was to compare time-dependent rsfMRI functional connectivity across days, no data scrubbing was applied, nor sophisticated data processing procedures were adopted. In our dataset, both FC specificity and FC between ROIs remained stable before, during, and after the habituation period. These findings suggest that the observed stability of functional connectivity reflects consistent underlying neural networks across the habituation process, and that habituation may primarily serve to reduce stress and head motion rather than to improve connectivity measures.

In the present study, a fixed habituation protocol was applied. However, gradual habituation protocols are generally expected to induce less stress. For example, Xu et al. (2022) and Poplawsky et al. (2023) gradually increased restraint duration and acoustic noise levels over periods of 2 weeks and 4 weeks, respectively. Accordingly, implementing a longer and more gradual habituation protocol within the MRI environment may be beneficial for further reducing stress.

Stress can induce physiological changes such as hypoxia or tachycardia, which may confound blood oxygenation level–dependent (BOLD) signals. However, oxygen saturation and heart rate were not measured in this study due to technical constraints. Despite potential physiological variability, FC remained stable across our experiments.

CORT levels are known to be sensitive to both sex and circadian rhythms (Fahrenkrug et al., 2012; Gong et al., 2015; Lindhardt et al., 2022; Thanos et al., 2009). Because only male mice were included in this study, our findings cannot be readily generalized to females, and future studies incorporating female mice are warranted. In addition, CORT measurements were not performed at a fixed circadian time point due to the requirement for MRI scans throughout the day. Consequently, the large variability observed in CORT levels is likely attributable, at least in part, to differences in sampling times.