Lower urinary tract (LUT) control depends on a complex and extensive network of brain regions. Although much of the pathophysiological mechanisms of overactive bladder syndrome (OAB) are unknown, it is likely that causative factors reside in the neural network (1). The periaqueductal gray (PAG) is located in the brainstem and occupies a pivotal role in the bidirectional communication between higher cortical and subcortical brain areas and the LUT. Afferent LUT sensory information arrives in the brain at the level of the PAG and, from here, is relayed to other brain areas involved in LUT control. Once the decision to void is made, cortical areas send a signal to PAG, which in turn activates the pontine micturition center (Barrington’s nucleus) in order to start voiding.

Functional neuroimaging enables mapping of the brain structures involved in LUT control. Brain parcellation methods enable the division of brain structures into a number of functional sub-regions, or parcels that have high within-cluster similarity of the blood-oxygen-level-dependent (BOLD) signal. By means of resting-state fMRI, it is possible to calculate the similarity in BOLD fluctuations between voxels and use that information to functionally parcellate brain areas. The duration of the scan determines how representative the data is true for “resting-state” fluctuations, and therefore influences the accuracy of parcellation. Longer scan durations has been shown to improve the reliability of resting-state fMRI connectivity patterns, plateauing at 12 minutes or longer for cortical areas (2). However, given the physiology of bladder filling, assuming an empty bladder state for longer periods of time is not tenable. Recent neuroimaging work reported that when comparing the consistency of empty and full bladder parcellation maps of PAG there were significant differences between patients and controls (3).

Therefore, our study aims to ascertain the robustness of parcellations derived from truncated time windows, and if this metric differs between patients and controls. We systematically parcellate the PAG during an empty bladder state using increasing time intervals and hypothesize that the rate at which parcellations resemble the ground truth parcellation differs between patients and controls.

This study was designed and conducted in line with the Declaration of Helsinki and was approved by our local ethics committee. Written informed consent was obtained from each of our participants before any study-related procedures took place. For this analysis, we included data from 6 female controls and 4 female OAB patients. We conducted an empty bladder resting-state fMRI scan during which we collected 420 T2*-weighted multiband echo planar imaging volumes (mb-EPI sequence, acceleration factor = 2, MB-factor = 2, TR =1400ms, TE = 22ms, resolution = 1.1 x 1.1 x 1.1mm). For each participant, we acquired 40 slices covering the supramedullary portion of the brainstem (including PAG). In addition, we ran a T1‐weighted whole‐brain anatomical scan using an MP2RAGE sequence. Data were preprocessed using BrainVoyager and normalized to MNI space with an additional manual step to optimize alignment of the brainstem to the MNI template.

To assess whether shorter scan durations produce comparable outcomes of the clustering algorithm across both groups, we segmented the entire scan length (420 volumes) into smaller datasets. Each dataset consisted of increments of 20 volumes, starting with the first set of 20 volumes and increasing by 20 volumes until reaching the final set of 420 volumes.
We generated parcellation maps of PAG per time increment obtaining 21 different maps. To achieve this, PAG voxels were chosen utilizing a mask delineating this region in the MNI template, after which a correlation matrix on a voxel-by-voxel basis was created. Subsequently, we employed the Louvain module detection algorithm to parcellate this correlation matrix into clusters with higher in-cluster connectivity than between-cluster connectivity. For each parcellation, we ran the algorithm for 200 iterations and selected the parcellation with the largest modularity value (Q-value) for further processing.

We used a permutation test to assess whether patients deviated from controls, running 250 permutations in which we randomly shuffled patient and control labels. The statistical significance level was set at p ≤ 0.05 after correcting for multiple comparisons.

In Figure 1, the correlation coefficient of PAG parcellations for each increment compared to the full length-scan is illustrated. In Figure 2, we can observe the mean coefficient correlation (± SE) of each group (patients and healthy participants). 
We observed that patients had higher correlations compared to controls across time windows. We demonstrated that correlations between PAG clusters in empty bladder state for the increase in increments and the correlation with the 420 volume full resting-state parcellation scan were significantly higher in patients compared to controls than could be expected based on chance (p < 0.005).

Our results indicate that the rate at which OAB patient PAG parcellations increase, in correlation to the ground truth as a function of dataset size, is considerably higher than in controls. Previous research showed that the consistency between empty bladder and full bladder parcellations differs between healthy and OAB patients (2). Here, we show that there is a difference in the effect of scan duration on parcellation results between controls and OAB. Therefore, we can conclude that the empty bladder state in OAB patients is transient compared to controls, as measured with resting-state BOLD signal in the brainstem.

Our results provide additional support that PAG activity patterns during rest in OAB patients differ to those of controls. This emphasizes the relevance of investigating the PAG to better understand lower urinary tract symptoms (LUTS), including OAB. In our figures, it is illustrated that patients reach a plateau quicker than healthy participants do. We propose to examine longer scans to investigate this trend further across different bladder states.

Our results show that the correlations between PAG parcellations based on short duration resting-state scans to a long duration resting-state parcellation were significantly higher in OAB patients compared to healthy participants. These differences indicate that resting-state BOLD fluctuations and functional connectivity patterns in the brainstem differ between OAB patients and controls.

Clarkson, B. D., Karim, H. T., Griffiths, D. J., & Resnick, N. M. (2018). Functional connectivity of the brain in older women with urgency urinary incontinence. Neurourology and urodynamics, 37(8), 2763-2775.
Fernández Chadily, S., de Rijk, M. M., Janssen, J. M., van den Hurk, J., & van Koeveringe, G. A. (2023). Assessment of brainstem functional organization in healthy adults and overactive bladder patients using ultra-high field fMRI. Biomedicines, 11(2), 403.
Birn, R. M., Molloy, E. K., Patriat, R., Parker, T., Meier, T. B., Kirk, G. R., ... & Prabhakaran, V. (2013). The effect of scan length on the reliability of resting-state fMRI connectivity estimates. Neuroimage, 83, 550-558.