Classification of Vigilance Level Based on Sympathetic Biomarkers: An fMRI-Pupillometry Study

2026 

Abstract Submission 

Elif Can¹, Pinar S Ozbay¹

¹Bogazici University, Istanbul, Turkey

Elif Can  
Bogazici University
Istanbul, Turkey

Pinar S Ozbay  
Bogazici University
Istanbul, Turkey

Our aim is to leverage pupil size, a marker of sympathetic activity, within a resting-state fMRI dataset, utilizing it as a proxy to classify subjects into two distinct subsets: those exhibiting high arousal and those with low arousal levels. Furthermore, we intend to support this classification by identifying and correlating observed sympathetic patterns within the data, thus shedding light on the relationship between autonomic nervous system, cerebrospinal fluid (CSF) flow and cognitive states.

In our study, we utilized the "Yale Resting State fMRI/Pupillometry: Arousal Study" dataset (Lee et. al., 2022), comprising 27 subjects with an average age of 26.52 years, including 16 females and 11 males, of which 25 were right-handed. T1-weighted anatomical images were acquired using a magnetization prepared rapid gradient echo (MPRAGE) pulse sequence with the following parameters: repetition time (TR) = 2,400 ms, echo time (TE) = 1.22 ms, flip angle = 8°, slice thickness = 1 mm, in-plane resolution = 1 × 1 mm, matrix size = 256 × 256, field-of view (FOV) = 256 mm, 208 contiguous slices acquired in the sagittal plane. Functional T2*-weighted BOLD images were acquired using a multiband gradient echo-planar imaging (EPI) pulse sequence (TR = 1000 ms, TE = 30 ms, flip angle = 55°, multiband acceleration factor = 5, slice thickness = 2 mm, in-plane resolution = 2 × 2 mm2, matrix size= 100 × 100, FOV = 220 mm, 75 contiguous slices acquired in the axial-oblique planes parallel to AC-PC line). Correlation maps were generated by calculating the cross-correlation between each voxel within the brain and; i)z-scored pupil size, ii)CSF signal from the 4th ventricle, and iii) whole brain average signal, across lags of +/- 10 TR. We also utilized group level ICA for identifying network patterns in two arousal groups.

The z-scored pupil size served as a proxy for arousal (Pais-Roldan et. al., 2020), facilitating the classification of subjects into two distinct arousal groups. A positive z-score denotes values above the average. The number of occurrences of positive and negative z-scored pupil sizes was then tallied for each subject. Subjects exhibiting an overall higher number of positive z-scores were categorized as 'high arousal,' and conversely, those with a prevalence of negative z-scores were designated as 'low arousal' subjects. The high arousal group consisted of 17 subjects, while the low arousal group included 10 subjects. In our findings, we discovered a distinguished negative correlation in the ventricular (CSF) area, in contrast to a more positive correlation pattern within the gray matter (Fig. 1). In addition, we observed a negative correlation pattern in the insula region for the high arousal group, a pattern which did not exist for the low arousal group. Within the pupil size & fMRI correlation maps, we noted a more pronounced (negative) correlation between pupil size and the visual area in subjects experiencing high arousal (Fig. 1). Conversely, when examining the whole-brain average correlation, we observed a stronger pattern in the ventricular (CSF) and gray matter regions, which aligns with previous observations during heightened sympathetic activity, such as during light sleep (Ozbay et al., 2018) or deep breaths (Picchioni, Ozbay et al., 2020). Our exploration of ICA has yielded that, within the low arousal group exists a substantial ventricle-GM contrast as evident in component 7 (Fig. 2), emphasizing the pronounced neural differences associated with diminished arousal states. Moreover, in component 12 of the high arousal group, a distinctive negative activation profile emerged within the insula region, accompanied by a positive activation in the middle temporal gyrus.

In summary, our study enhances our understanding of the complex relationship between brain dynamics, pupil size, cognitive states, and CSF flow, revealing mechanisms governing arousal levels.

Task-Independent and Resting-State Analysis ¹

Multi-Modal Imaging ²

Data analysis

fMRI CONTRAST MECHANISMS

FUNCTIONAL MRI

MRI