Wide field multispectral imaging of light backscattered by brain tissues provides maps of hemodynamics changes (total
blood volume and oxygenation) following activation. This technique relies on the fit of the reflectance images obtain at
two or more wavelengths using a modified Beer-Lambert law1,2. It has been successfully applied to study the activation
of several sensory cortices in the anesthetized rodent using visible light1-5. We have carried out recently the first
multispectral imaging in the olfactory bulb6 (OB) of anesthetized rats. However, the optimization of wavelengths choice
has not been discussed in terms of cross talk and uniqueness of the estimated parameters (blood volume and saturation
maps) although this point was shown to be crucial for similar studies in Diffuse Optical Imaging in humans7-10. We have
studied theoretically and experimentally the optimal sets of wavelength for multispectral imaging of rodent brain
activation in the visible. Sets of optimal wavelengths have been identified and validated in vivo for multispectral imaging
of the OB of rats following odor stimulus. We studied the influence of the wavelengths sets on the magnitude and time
courses of the oxy- and deoxyhemoglobin concentration variations as well as on the spatial extent of activated brain
areas following stimulation. Beyond the estimation of hemodynamic parameters from multispectral reflectance data, we
observed repeatedly and for all wavelengths a decrease of light reflectance. For wavelengths longer than 590 nm, these
observations differ from those observed in the somatosensory and barrel cortex and question the basis of the reflectance
changes during activation in the OB. To solve this issue, Monte Carlo simulations (MCS) have been carried out to assess
the relative contribution of absorption, scattering and anisotropy changes to the intrinsic optical imaging signals in
somatosensory cortex (SsC) and OB model.
|