Introduction

Optogenetic techniques are valuable for selectively exciting neurons targeted by a wavelength-specific fluorophore that responds only to laser light with a specific wavelength¹, the effects of which can be detected using BOLD fMRI². However, the lasers used to excite the optical fluorophores may produce non-negligible thermal effects in neural tissues³. BOLD fMRI is conventionally based on relatively long TE gradient-recalled echo (GRE) acquisitions which are also suitable for real-time temperature monitoring, so the opportunity arises to simultaneously monitor the neural effects of optogenetic neuromodulation and concomitant tissue heating caused by the laser. We describe a processing pipeline to calculate temperature maps from BOLD images and detect small temperature rises associated with laser stimulation using general linear modeling and based on the proton resonance frequency (PRF) shift with temperature⁴.

Methods

Animal Preparation: Figure 1 illustrates the experimental setup. A squirrel monkey transfected using AAV5/9-CaMKIIa-ChR2 in secondary somatosensory cortex (S2) was pre-anesthetized with ketamine and anesthesia was maintained with isoflurane (0.9-1.5%) delivered in a 70:30 N₂O/O₂ mixture during functional data acquisition. ChR2 has a peak excitation wavelength of 470 nm (blue light). An optical fiber with tip area 0.0314 mm² was inserted into the left side of S2. All procedures were conducted in accordance with NIH guidelines and were approved by the IACUC at Vanderbilt University.
Data Acquisition: fMRI and structural images were collected on a Varian DirectDrive^TM/Agilent MR spectrometer horizontal 9.4-T magnet using a birdcage head coil. BOLD fMRI images were acquired with a 2-shot T2*-weighted 2D GRE-EPI sequence with TE/TR = 10ms/1500ms, in-plane resolution = 0.86 x 0.86 mm², slice thickness = 1 mm, 24 slices and 290 volumes. High-resolution T2*-weighted structural images were acquired with in-plane resolution 0.12 x 0.12 mm² in the same session.
Laser Stimulation Protocol: During imaging, blue (473 nm) and green (561 nm) light emissions with power 30 mW and 20 Hz pulses were delivered in an interleaved fashion. After acquiring 10 baseline volumes (3s / volume), 30 second-duration blocks of interleaved blue and green light were repeated 28 times. The protocol was repeated with three laser duty-cycles of 3s/30s (10%), 6s/30s (20%), and 9s/30s (30%) for each pulse.
Temperature Map Calculation and Processing: Figure 2 shows how the images were processed to calculate temperature changes. Complex-valued BOLD-fMRI images were reconstructed from raw data with EPI phase correction. Then, the hybrid referenceless and multi-baseline subtraction thermometry method^5,6 was used to calculate a voxel-wise temperature time series based on the PRF shift, while removing dynamic background phase variations due to motion, respiration, and cardiac pulsation. Next, a general linear model was fit to the temperature time series (no spatial smoothing) at each voxel for the stimulation presentation paradigm using FSL⁷ to quantify the temperature changes during the stimulation. No convolution function was used in the model. Finally, a statistical threshold was applied to identify the voxels that exhibited significant temperature changes (p < 0.05, FDR corrected).

Results

Figure 3 shows a movie of calculated temperature through time, using conventional baseline subtraction and hybrid multi-baseline thermometry. Compared with baseline subtraction thermometry, hybrid multi-baseline thermometry reduced confounding phase variations caused mainly by respiration that otherwise obscure true temperature changes caused by the laser. Figure 4 plots the mean temperatures through time in 2 x 2-voxel ROI of size 1.72 x 1.72 mm² around the probe. The blue and green columns in the background indicate when the blue and green lasers were on. For 3-second/10% duty cycle illumination there is very little apparent temperature change due to the lasers, but the temperatures start to peak for 6-second/20% duty cycle illumination and temperature peaks closely follow laser illumination in the 9-second/30% duty cycle case. Figure 5 shows z-statistic temperature ‘activation’ maps and parametric temperature change maps overlayed on structural images, for each duty cycle and laser color. The parametric temperature change maps represent the average amount by which the temperature increased due to the laser in each voxel. The white arrows indicate the position of the laser probe. Similar to the ROI temperature curves, little activation is seen for the 3-second illumination protocol for both blue and green light, while the 3-second and 9-second protocols show stronger activation and temperature rises.

Discussion

Optogenetic stimulation in combination with BOLD fMRI provides a powerful method for neuron type-selective neuromodulation with simultaneous read-out of effects, both locally and over the whole brain. At the same time, the lasers used to stimulate the optogenetic effect may cause heating that could confound the results. In this work, we demonstrated how BOLD-fMRI images can be used to simultaneously detect small temperature changes on the order of tenths of a degree Celsius caused by the lasers used for optogenetic neuromodulation. At 30-mW power level, we found that detectable temperature rises appeared with laser protocols with duty cycles of 20% or higher, and that the two laser wavelengths induced similar levels of heating. Because control green light does not excite neurons as blue light does, the observed temperature changes are likely only attributable to PRF shifts caused by the laser⁸. These results and the technique could be useful for the design of optogenetic stimulation protocols and for monitoring confounding heating effects.

Acknowledgements

We acknowledge the support of NIH grants U18 EB 029351 (HEAL), R01 EB028773, and R01 NS078680.