Introduction

Widely used to treat various neurological disorders^1,2, deep brain stimulation (DBS) is currently evolving to target new areas in the brain. Medial septal nucleus (MSN) has projections to the hippocampus, making it an appealing target to treat memory impairments³ and induce neurogenesis in the hippocampal formation^4,5. Functional magnetic resonance imaging (fMRI) with Multi-Band SWeep Imaging with Fourier Transform (MB-SWIFT) is a powerful tool to track brain response during the DBS, however, its low temporal resolution restricts discovering temporal features of activation. In this work, we demonstrate a novel resampling strategy that is applicable to radial k-space sampling such as used in MB-SWIFT and that allows tracking repeating events in the brain with high temporal resolution. Here, we aimed at mapping brain activation in response to DBS of the MSN in rats.

Materials and Methods

The study was conducted in 7 male Sprague-Dawley rats. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Minnesota.
The resampling approach relies on the recognition that similar repeating events can be tracked with higher temporal resolution than a single event by combining data from several events. A different portion of the k-space can be selected when acquiring a data point from each event and can then be combined with other k-space portions to cover the entire k-space. To achieve that, each event should be shifted so that different portions of k-space are measured each time for every time point. To improve time resolution n times when keeping k-space coverage the same, we would need at least n repetitions of the event (Fig. 1). After reordering the spokes, the data is reconstructed normally, as if it was acquired from one event.
We applied this method to DBS of the rat brain. Animals were anesthetized with isoflurane and implanted with an electrode in the MSN. Subsequently, the anesthesia was switched to urethane. We used MB-SWIFT imaging sequence with the following parameters: TR 0.97 ms, bandwidth 192 kHz, matrix size 643, FOV 3.5×3.5×6.4 cm3, flip angle 5°. The stimulation paradigm consisted of 10-s stimulations and 2-minute rest. A single 3D k-space consisted of 2000 spokes and the total time of volume acquisition was 2 s. Therefore, to achieve 200-ms time resolution, stimulus was repeated 10 times. The stimulus shifts were applied in random order to avoid the effect of global signal change. The resting time was randomized between 55 and 65 volumes. This measurement was repeated 3 times for each animal.
The spokes for each time point were aggregated to achieve the whole k-space coverage. For each animal, three 118-s – long measurements with 200-ms time resolution were composed out of the three repeated acquisitions with 10 stimulus periods. These three measurements were then averaged.
To exclude habituation to the stimulus (Fig. 2), we checked the consistency of the response and activation amplitudes for each animal. We chose several regions of interest (ROIs) and compared time-to-peak between these ROIs. (Fig. 3). In addition, we performed a group tensor ICA analysis to see all the different responses in the brain (Fig. 4).

Results

DBS in the MSN caused widespread activation in several brain regions, including the hippocampus and some cortical structures. We found that the results were consistent for each animal, and for the three repeated scans (Fig. 2). No habituation to the stimulus during the one-hour experiment was observed.
Time-to-peak and amplitude of activation varied between the different brain regions (Fig. 3). In general, activation closer to the electrode and in the areas directly connected to the MSN lasted longer than in the areas more distant from the stimulus site.
Group ICA was able to detect several stimulus-related components with different time courses (Fig. 4).

Discussion

The goal of this development was to establish a framework for high temporal resolution method using MB-SWIFT technique for monitoring fMRI response to DBS. Increasing temporal resolution can allow investigating the mechanisms of action of DBS in a broader temporal scale than it was possible so far using conventional fMRI.
With the new method, we found that time-to-peak and shapes of DBS responses differ between the brain regions. Noticeably, cortical areas showed the fastest time-to-peak interval, while the areas directly connected to the area of stimulation exhibited the longest time-to-peak response. It is important to note that temporal differences can either reflect differences in brain activation or brain region-dependent differences in hemodynamic response function.
Notably, this method can also be applied to naturally occurring events, not only to external stimuli. This will require more repetitions of the event for k-space coverage and potentially more sophisticated reconstruction for partially sampled k-space, as natural events occur more randomly.

Conclusion

A novel resampling SWIFT-fMRI method allows tracking repeating events in the brain with high temporal resolution. It is easily applicable and useful for DBS studies, where the responses are stable and differ significantly between the brain regions. However, the method has the potential to be used with recurring events of any type.

Acknowledgements

This work was supported by the National Institutes of Health U01-NS103569-01, the Center for Magnetic Resonance Research NIH core grant P41EB027061, the EU H2020 Marie Skłodowska RISE project #691110 (MICROBRADAM), Erkko Foundation