Functional magnetic resonance imaging (fMRI) has become the most important technique to monitor blood flow responses that follow neural activity. However, the spatial resolution of most MR scanners is restricted to voxel sizes larger than the size of cortical columns (~200 µm), which makes investigations of neurovascular coupling at submillimeter scale difficult. In contrast, optical imaging of intrinsic signals (IOI) has been used for more than 20 years to record hemodynamic responses at high spatial and temporal resolutions, allowing to map columnar structures like individual barrels in rodent cortices or orientation columns in the primary visual cortex of primates and other species. However, it lacks the third dimension that fMRI is able to provide. In order to get deeper insights into the origin of the BOLD signal we have created a setup that allows us to combine the benefits of both techniques.

Setup: An optical imaging setup was designed that is completely capable of functioning inside our 14.1 T MR scanner (Fig. 1), based on a custom-made 5.5 megapixel (2560 x 2160 pixel) board-level camera equipped with a scientific, highly sensitive CMOS sensor, capable of recording images at framerates of up to 120Hz. A tandem macroscope was assembled of two objectives (Nikon Nikkor 50mm F1.2 and Zeiss Sonnar 135mm F1.8) with an effective magnification of 2.7. Both objectives were stripped of all mechanical elements, leaving only the lenses and the lens mounting (mainly aluminum), which was replaced by plastic in the regions close to the rat brain to avoid artifacts. The cortical surface was illuminated by red or green light (LEDs peak wavelengths: 530nm, 630nm) that was guided to the cortical surface by eight 1mm diameter PMMA fibers.

Animal preparation: Rats were anesthetized with urethane. The skull was removed to form a cranial window of about 6x6 mm² above the somatosensory forepaw region. The dura was left unharmed. Neuronal activity inside S1 was triggered by forepaw stimulation, applying 9 mA currents with 9 Hz for four seconds through two electrodes inserted in the forepaw.

Optical Imaging: Reflectance images from the cortical surface were recorded at green and red light illumination in consecutive stimulus blocks, each consisting of 15 stimulus trials. 800 frames were recorded per single trial at 29 Hz, 30 ms exposure /frame and 2560x2160 pixel resolution. The signal-to-noise ratio was enhanced by averaging all 15 trials of a single block and additional 2x2 pixel binning. To calculate single condition maps we used a frame-0-subtraction: Twenty camera frames preceding stimulus onset were averaged to form the frame zero. Each frame following stimulus onset and its ten nearest neighbours were averaged and used to calculate percentage signal changes. For topographic localization of functional domains, reflectance maps were further filtered for high and low frequencies in the spatial domain.

fMRI: A single loop surface coil around the cranial window served to transmit and receive MRI signals. Functional images were acquired from four 0.5 mm slices parallel to the brain surface with EPI with a spatial resolution of (0.31 mm)², a matrix size of 64², and a TR of 1 s. To investigate the effect of the rapid gradient switching in EPI, additional functional scans were performed using a FLASH sequence with long TE. The data were processed with home-written Matlab routines.

Intrinsic optical signals were of excellent quality and allowed for clear identification of activated cortex regions (Fig. 2) in both red and green light, indicating changes in oxygenation and CBV, respectively. Reflectances up to 0.15% were observed. RF-transmission during the MRI scans had no effect on the images, while the rapidly switched EPI gradients cause slight vibrations in the optical images, which get stronger for increasing spatial resolution of the fMRI scans, but disappear when using a FLASH sequence. fMRI signals were not significantly affected by the presence and operation of the camera. BOLD signal changes of up to 15% were observed in all slices.Concurrent fMRI and intrinsic optical imaging with high spatial resolution is possible even in the magnetic field of an ultra-high field animal scanner. In contrast to previous approaches, which rely on optical fibers to guide the signal to a camera outside the magnetic field [1], our setup uses a magnetic field proof camera and optics, allowing for higher sensitivity and spatial resolution. The excellent quality of both optical and BOLD signals will allow for a detailed investigation of the spatial and temporal formation of the BOLD signal and open up novel possibilities to explore brain function by a combination of those two modalities.