Phase of quasi-periodic patterns in the brain predicts performance on psychomotor vigilance task in humans
Anzar Abbas1, Waqas Majeed2, Garth Thompson3, and Shella Keilholz4

1Neuroscience Program, Emory University, Atlanta, GA, United States, 2School of Science and Engineering, Lahore University of Management Sciences, Lahore, Pakistan, 3Radiology and Biomedical Imaging, Yale University, New Haven, CT, United States, 4Biomedical Engineering, Emory University and Georgia Institute of Technology, Atlanta, GA, United States

Synopsis

Functional organization of brain networks plays an important role in behavior. Analysis of the dynamics of two functional networks – the default mode (DMN) and task positive (TPN) networks – has shown a dependency of task performance on relative network activation. Fluctuations between these two networks have been seen to occur in humans in a continuous, quasi-periodic fashion. However, the nature of these quasi-periodic patterns (QPPs) and their effect on behavior is not well understood. We show that QPPs do not differ between resting state and task-based scans and that the phase of these QPPs can serve as predictors of performance on the psychomotor vigilance task (PVT).

Purpose

We aimed to compare the occurrence of QPPs between resting state and task-based functional scans by looking at their frequency, intervals, and correlation strength. Next, we aimed to understand whether the intrinsic properties of a QPP would serve as a predictor of task performance. We hypothesized that the QPPs would be similar between both resting state and task-based scans and that the phase of a QPP would affect performance on a psychomotor vigilance task.

Methods

MRI data collection, preprocessing, and DMN and TPN mask creation was performed as in Thompson et al1. QPPs were identified using the pattern-finding algorithm described in Majeed et al2. QPP templates, ~20 seconds in length, were acquired by using varying starting points in the algorithm until the phase of the resulting template showed progression of BOLD activation from DMN to TPN, as shown in Figure 1. Next, occurrences of QPPs were accurately located within the functional scans using a thresholded timecourse of the QPP template strength. The average number of QPPs per scan, the average interval between QPPs, and the average strength of correlation were compared between resting state and PVT scans. Finally, performance on the PVT was compared with the phase of the QPP that the task onset occurred in by comparing reaction times for task onsets that occurred in the first half of the QPP with reaction times for task onsets that occurred in the second half.

Results

There was no significant difference seen in the average number of QPPs per scan, the average interval between QPPs, and the average strength of correlation of QPPs with the functional scan between resting state scans and PVT scans (Figure 2). Comparison of reaction times for PVT task onsets that occurred in the first half of the QPP versus the second half of the QPP showed a significant difference in mean reaction time (p < 0.05). Performance was slower in the first half of the QPP when DMN activation was high compared to the second half of the QPP where TPN activation was dominant (Figure 3).

Discussion

Our results are consistent with previous studies showing that analysis of spatiotemporal dynamics of low frequency BOLD fluctuations indeed results in observation of quasi-periodically repeating patterns in humans that comprise of repeated shifts between the default mode network and the task positive network (Figure 1). The basic properties of QPPs, such as number of QPPs per scan, average QPP interval, and average strength of correlation with original functional scans shows that the nature of QPPs does not differ between resting state and task-based scans (Figure 2). This contributes to the hypothesis that QPPs may be responsible for large-scale modulations of cortical excitability regardless of brain state, with QPPs seen even in anesthetized animals2, 3. Analysis of relationship between the phase of the QPP and PVT reaction time begins to establish the link between QPPs and behavior/task performance. The distribution of the reaction times as a function of QPP phase clearly showed a tendency toward greater variability in the first half of the QPP, where the DMN is dominant. This motivated a division of the QPP into two halves, one with DMN dominance and one with TPN dominance. Reaction times are faster and less variable in the second half of the QPP, when TPN activation is dominant, suggesting higher vigilance during TPN activation and confirming QPP phase as a predictor of task performance. These findings are consistent with previous work that looked at the relative activation in the DMN and TPN rather than the QPP phase, suggesting that the QPP could be one contributor to the variations observed in the network activity. This is especially significant as QPPs have been shown to correlate with infraslow electrical activity in rats4. In consequence, though QPPs are unlikely to directly reflect cognitive processing due to their slow and quasi-periodic occurrence, their effect on large-scale cortical excitability may be tied to attention and arousal levels.

Conclusion

Quasi-periodic patterns involving DMN and TPN activation are a robust occurrence in fMRI data regardless of brain state, suggesting a role in large-scale functional network organization. QPPs can have an effect on attention and task performance through modulations of cortical excitability, with a higher activation of the TPN in the latter phase of the QPP leading to faster reaction times and increased performance on the psychomotor vigilance task.

Acknowledgements

No acknowledgement found.

References

1. Thompson G, Magnuson M, Merritt M, et al. Short-Time Windows of Correlation Between Large-Scale Functional Brain Networks Predict Vigilance Intraindividually and Interindividually. Hum Brain Mapp 2012;34:3280-3298.

2. Majeed W, Magnuson M, Hasenkamp W, et al. Spatiotemporal dynamics of low frequency BOLD fluctuations in rats and humans. Neuroimage 2011;54:1140-1150.

3. Majeed W, Magnuson M, Keilholz S. Spatiotemporal dynamics of low frequency fluctuations in BOLD fMRI of the rat. J Magn Reson Imaging 2009;30:384-393.

4. Thompson G, Pan W, Magnuson M, et al. Quasi-periodic patterns (QPP): Large-scale dynamics in resting state fMRI that correlate with local infraslow electrical activity. Neuroimage 2014;84:1018-1031.

Figures

Figure 1: QPP templates were acquired using multiple starting points until a progression of activation from the DMN to TPN was seen. This shows BOLD activation of DMN and TPN in a QPP from all scans used in study (background) overlaid with the average QPP template from all scans.

Figure 2: Comparing QPPs between resting state and PVT scans with standard deviation. (a) Average number of QPPs per scan, (b) average QPP interval, and (c) average correlation strength of QPP showed no significant difference between resting state and task data (p = 0.2495, 0.5392, 0.6625 respectively)

Figure 3: (a) Relationship of QPP Phase versus reaction time. Blue and red circles represent task onsets occurring with higher DMN and TPN activation respectively. Dashed and dotted lines represents mean reaction time and standard error in that half respectively. (b) Mean reaction times in first half versus second half of QPP with standard error (p <0.05).



Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)
1192