There has been recently a renewed interest in using Autofluorescence imaging (AF) of NADH and flavoproteins (Fp) to map brain activity in cortical areas. The recording of these cellular signals provides complementary information to intrinsic optical imaging based on hemodynamic changes. However, which of NADH or Fp is the best candidate for AF functional imaging is not established, and the temporal profile of AF signals is not fully understood. To bring new theoretical insights into these questions, Monte Carlo simulations of AF signals were carried out in realistic models of the rat somatosensory cortex and olfactory bulb. We show that AF signals depend on the structural and physiological features of the brain area considered and are sensitive to changes in blood flow and volume induced by sensory activation. In addition, we demonstrate the feasibility of both NADH-AF and Fp-AF in the olfactory bulb.
© 2009 Optical Society of America
Endogenous optical signals arise from molecular mechanisms linked to neuroenergetics and are routinely used to map sensory activation of neural networks. These signals can be divided into two groups according to the vascular or the cellular compartment that they sample in the brain tissue. Intrinsic Optical signals (IOS) are tightly coupled to changes in vascular dynamics. They are produced during activation by a local decrease in red light reflectance due to changes in blood oxygenation level and blood volume . Since the mid 80’s, IOS imaging has provided with unique insights into the functional architecture of the visual, auditory, somatosensory and olfactory systems . On the other hand, cellular fluorescence or autofluorescence (AF) signals monitor the activity of energetic cycles in vivo because they are related to changes in the redox state of metabolic intermediates such as Nicotinamide Adenine Dinucleotide (NADH) and flavoproteins (Fp). NADH-AF has been used for a long time in optical spectroscopy  and imaging  experiments. Recent improvements in the sensitivity of scientific imaging CCD cameras have allowed in vivo imaging of NADH-AF with adequate temporal resolution [4-6] and the use of Fp-AF for functional brain imaging [7,8]. AF signals imaging has allowed visualization of activity-dependent changes in many sensory structures [7-13]. However, to our knowledge, AF recordings have not been reported in the olfactory bulb (OB) despite the great interest of this structure to the study of neurovascular and neuroenergetics coupling following sensory stimulation [14,15]. In particular, the OB is constituted of remarkably well-defined functional modules, the olfactory glomeruli, which are easily accessible for functional optical imaging and can be stimulated by odorants at controlled physiological intensities and durations.
Although potentially interesting for high resolution imaging, AF signals are not fully understood because of three incompletely addressed questions. First, which of NADH or Fp is the best candidate for in vivo AF signals imaging is unclear. Second, the spatial location in depth and extension of the source of the recorded AF signals within brain tissues have not been studied so far. Third, the quantification of the putative contamination of AF signals by activity-dependent hemodynamic changes (leading to changes in light absorption and scattering) is still lacking for in vivo studies. In this view, attempts were made to separate the AF signals and hemodynamic responses by pharmacological studies in brain slices [8,16] or in vivo [8,12]. These studies have shown that some components of AF signals are clearly not due to vascular effects. However, quantitation of the vascular contamination to in vivo AF signals is difficult since the selective inhibition of hemodynamic responses by nitric oxide synthase is only partial as underlined in reference .
To date, no theoretical quantitative approach has been applied to describe optical changes during brain activation and their influence on sensory-evoked AF signals recorded in vivo. In the present article, we have carried out Monte Carlo (MC) simulations to study the physical aspects of AF signals recordings following sensory activation in the somatosensory cortex (SsC) and OB.
2. Materials and methods
We performed standard MC simulations of optical photons travel through biological tissues  in anatomofunctional models of the SsC and OB. For NADH and Fp, we considered photons with the following wavelengths: NADH excitation at 350 nm and emission at 440 nm, and Fp excitation at 440 nm and emission at 530 nm. Since AF recordings have already been reported in the rat SsC in vivo, we first performed MC simulations in the SsC and then made a comparison with the OB.
2.1 Geometrical properties of the somatosensory cortex model
In rodents, each whisker is represented in a localized column of the neocortex which is organized in 4 distinct layers classically numbered from L1 to L4 in the dorsoventral axis from the surface (see Fig. 1). Mechanical stimulation of a whisker activates a specific thalamic relay, which terminates in layer 4 of the SsC named the “barrel” because of its cylindrical shape. After initiation in the barrel, the activation then propagates to layers L3/L2 of the SsC . On the basis of anatomical data [18,19], we chose an average thickness of 300 µm for each layer with a diameter of 300 µm for the base of the cylindrical column. Considering the vascularization of these layers, the density of capillaries is similar in L2, L3 and L4 and higher than in L1 . On the basis of these anatomofunctional data, we have built a model of SsC constituted of 2 layers: Layer LII with a width of 900 µm represents the neocortical column and includes L2, L3 and L4, and layer LI represents cortical tissues above LII (Fig. 1).
2.2 Geometrical properties of the olfactory bulb
The OB is organized in six different layers: the Olfactory Nerve Layer (ONL), the Glomerular Layer (GL), the external plexiform layer, mitral cell layer, internal plexiform layer and granular cell layer . Our OB model is constituted of 3 layers. The first is the ONL, formed by the unmyelinated axons of olfactory receptor neurons located in the olfactory epithelium. The ONL innervates the GL, which is constituted of spherical structures of roughly 100 µm diameter, the olfactory glomeruli. As described by Chaigneau et al. in their in vivo two-photon microscopy study , the upper boundary of the GL is located on average at 140 µm below the surface. Considering these anatomical data, we have chosen the GL as the second layer of our model and defined it as a layer of contiguous 100 µm diameter spheres located between 140 and 240 µm of depth. Finally, as the remaining layers present similar anatomical and vascularization features, their optical properties were assumed to be identical and were fused into a single “Underneath the Glomerular Layer” (UGL). UGL is the third layer of our OB model and represents all tissues between 240 and 1200 µm of depth.
2.3. Optical properties of the tissues
The propagation of light in tissues is ruled by the local optical properties described by four coefficients: the index of refraction n, the absorption coefficient µa, the scattering coefficient µs and the anisotropy factor g. In our model, n is fixed for all tissues to 1.4 . Absorption and scattering optical properties throughout different layers are homogeneous and are wavelength-dependent. For each layer, the optical parameters are summarized in table 1 (SsC data), table 2 (OB data), and tables 3 and 4 (influence of hemodynamic changes on absorption coefficient µa).
In Eq (1), [Hb]t (in mole/l) is the total hemoglobin concentration in tissues. εHbR and εHbO2 are the molar extinction coefficients (in cm-1/mole/l) for deoxyhemoglobin and oxyhemoglobin, respectively and were taken from the Oregon Medical Laser Center website . S is the oxygen saturation defined as [HbO2]/[Hb]t and was set to 0,6 . Values of [Hb]t for LII and GL were respectively set to 10 g/l and 7.5g/l in accordance with local blood volume estimated in rat cortical tissues  and in the GL . Values of [Hb]t for ONL and UGL were both derived from 2-photon vascularization images of the OB  and set to 5.4 g/l. The value of [Hb]t for LI was derived from vascularization images obtained by dye injection in penetrating arterioles in the SsC  and was set to 7.5 g/l. These values of [Hb]t are considered to represent the basal state in the absence of activity-dependent hemodynamic changes.
In addition to Hb absorption, a lower but significant contribution (about 2% to 12%) to the total absorption coefficient is due to cell bodies and axons, which contain endogenous chromophores, including NADH and Fp. Here we made the assumption that the cellular absorption reflects the absorption by the endogenous fluorophores. Values of the cellular absorption coefficient µa-cell were taken from a recent study of absorption in tissues devoid of blood .
The total absorption coefficient was calculated for each wavelength with Eq (2).
Variations of cerebral blood flow and volume following activation lead to an increase in total hemoglobin concentration (Δ[Hb]t) in all the layers of the SsC and OB models. The blood volume increases not only in the capillaries within the deep layers (e.g. GL and UGL in the OB) but also in the surface arterioles that penetrate the top layers (e.g. the ONL in the OB). In our models this induces an increase of the absorption coefficient in all layers following activation as shown in tables 3 and 4. To simulate dynamic changes in blood volume following activation, we chose a maximum Δ[Hb]t of 30% according to the literature [26-29] with two intermediate increases of 10 and 20%.
2.3.2 Scattering and anisotropy
The scattering and anisotropy properties of tissues depend on their anatomical structure. The nerve fibers of the ONL are organized in bundle and the diameters of axons are small comparatively to cell bodies leading to scattering properties similar to white matter, i.e. a high scattering coefficient µs. Contrarily to ONL, the other layers GL, LI and LII contain mainly cell bodies and blood vessels so that their optical properties are similar to those of grey matter, i.e. a moderate µs. To our knowledge, there is no reported data on optical properties of brain tissues in the living rodent at the wavelengths that we consider in this article. Consequently, we fixed values of µa-cell(λ), µs(λ) and g according to measurements from human brain tissues .
2.4 Principles of MC simulations
MC simulations of photons travel through biological tissues rely on the principles described by Prahl et al , which have been further validated in the widely used MC code MCML . Each photon packet is described by its Cartesian coordinates, direction cosines and variable weight. Photons are launched with perpendicular incidence to the brain surface and propagate in the tissues in successive steps with varying length. To avoid repetition in the large random sequences of number necessary for the propagation of photons in MC simulations, we use a pseudo random number generator named “Mother-of-all” . At the end of each step, photons can either be absorbed or scattered. The probabilities of scattering or absorption are randomly sampled according to the properties of the tissues and the Henyey Greenstein phase function. No Russian roulette termination of photon was used to preserve the accuracy of MC simulations for the determination of absorption coordinates. The geometry is semi infinite so that photons travelling deeper than 1200 µm from the surface are no longer propagated and are considered as “Transmitted” photons. The photons directly reflected by the surface and the photons backscattered after having penetrated the tissues are both considered and counted as “Reflected” photons. In our MC simulations 108 photons were launched to achieve acceptable statistical accuracy on the recorded parameters
2.4.1 Absorption of excitation photons
For both SsC and OB models and for the wavelengths of excitation of NADH (350 nm)  and Fp (440 nm) , the coordinates of absorbed photons were recorded. First, the vascular absorption was studied using µa-Hb as absorption coefficient for both SsC and OB models. Second, the total absorption for each layer, including absorption by cellular fluorescent chromophores, was studied using µa as the total absorption coefficient. The matrixes of vascular absorption probability and the matrixes of total absorption probability were derived from these simulations.
2.4.2 Emission and detection of fluorescence photons
Fluorescence matrixes were obtained by subtracting the total hemoglobin matrix of absorption from the total hemoglobin and tissues matrix of absorption. The resulting matrices reflect the heterogeneous sources of fluorescence for the OB and SsC models and were used as isotropic sources for either NADH or Fp fluorescence. In the absence of in vivo data, a non-realistic fluorescence quantum efficiency of 1 was considered for both NADH and Fp. Emitted photons were propagated in tissues and their coordinates and direction cosines recorded after each step. Then, for each location in the 3-dimensional models of OB and SsC, we derived the probability of a photon scattering to that location from the functional layers GL and LII. Fluorescence photons detected are those exiting the brain with a direction cosines encompassed within the numerical aperture of the optical apparatus, which was set to 0.14.
3.1 Absorption and backscattering of excitation photons at 350 and 440 nm in the SsC
The percentage of excitation photons absorbed, reflected and transmitted in our SsC model are presented in Fig. 2(a). Reflected photons represent about 44% of the excitation photons at 350 nm and only 19% at 440 nm. The contribution of LII to the total amount of absorbed photons is the most important with 40% and 54% of excitation photons at 350 nm and 440 nm, respectively. This phenomenon is mainly due to the volume of LII that occupies 75% of the SsC volume in our model. If we further focus on the different layers of LII, we observe that the largest amount of excitation photons at each wavelength is absorbed between 300 and 600 µm underneath the surface, i.e. in L2. For instance, at 440 nm, we can see that 46% of the excitation photons are absorbed in L2, whereas only 8% are absorbed in L3 and almost no photons are absorbed in L4.
As shown on Fig. 2(b), the distribution of absorbed photons in SsC expressed in percentage of the total absorption versus depth increases almost linearly through LI and increases steeply after penetration in the upper layer of LII (L2) to reach its maximum at 350 µm at either 350 or 440 nm. The amount of photons locally absorbed decreases through L2, L3 and L4 where it is almost negligible. Although similar absorption profiles are observed for both wavelengths, the increase of photons absorbed locally in L2 is more moderate at 350 compared to 440 nm.
The percentage of excitation photons that will lead to fluorescence in L2 at both excitation wavelengths are presented in Fig. 2(c). Since L2 absorbs most of the excitation photons at 350 and 440 nm, only weak fluorescence will occur in L3 and no fluorescence will occur in L4. In L2, 1.75% of the 350 nm excitation photons that are absorbed will lead to fluorescence, whereas this percentage is 0.60% for excitation photons at 440 nm.
3.2 Absorption and backscattering of excitation photons at 350 and 440 nm in the OB.
The percentage of excitation photons absorbed, backscattered and transmitted in our OB model are presented in Fig. 3(a). Due to the highly scattering structure of the ONL, almost 70% and 56% of the total excitation photons are backscattered at 350 and 440 nm. Although the GL and ONL have similar volumes, we observe that the local amount of absorbed photons in the GL is twice higher than in the ONL (2% versus 4% at 350 nm and 3% versus 7% at 440 nm). The UGL occupies 80% of the total OB volume. Consequently, the contribution of UGL to the total absorption is the most important with about 20% and 33% of absorbed photons at 350 and 440 nm respectively.
Figure 3(b) presents the distribution of absorbed photons in OB expressed as percentage of the total absorption versus depth. For 350 and 440 nm wavelengths, the amount of photons locally absorbed increases similarly through ONL and, after a sharp increase, reaches its maximum of about 4% at 190 µm in the GL. After exiting the GL, the amount of photons locally absorbed remains stable until 600 µm and then decreases progressively.
Focusing on the percentage of absorbed excitation photons which lead to fluorescence in the functional layer GL (Fig. 3(c)), we observe that there is more fluorescence at 350 than at 440 nm. In the GL, 0.42% of the excitation at 350 nm leads to fluorescence compared to only 0.11% at 440 nm. Similar to layer L2 in the SsC, the contribution to cellular absorption in the GL is higher at 350 nm, (µa-cell=3.4 cm-1 at 350 nm versus 1.5 cm-1 at 440 nm), whereas the contribution of hemoglobin is higher at 440 nm (µa-Hb=60.8 cm-1 at 440 nm versus 30.2 cm-1 at 350 nm). As a consequence, excitation at 350 nm leads to four fold more fluorescence photons in the GL compared to excitation at 440 nm.
3.3 Origin and intensity of the detected fluorescence signals at 440 and 530 nm emission wavelengths in the SsC.
Figure 4(a) shows that most of the detected fluorescent photons have their origin in L2. Considering the same amount of fluorescent photons emitted at 440 or 530 nm, we observe that 1% of 530 nm fluorescent photons are detected compared to 0.14% of 440 nm photons (Fig. 4(b), left column). This leads to an intensity ratio of 7 for the detected fluorescence as illustrated on the corresponding images at the surface of the SsC for both fluorescence signals (Fig. 4(c), left column). This intensity ratio is a consequence of the influence of the higher absorption coefficient of hemoglobin at 440 nm (µa-Hb=60.8 cm-1 at 440 nm versus 10.6 cm-1 at 530 nm for LII), which is enhanced by the depth of the fluorescence emission in the SsC.
3.4 Origin and intensity of the detected fluorescence signals at 440 and 530 nm emission wavelengths in the OB.
The profile of the percentage of the detected GL fluorescent photons as a function of their depth of emission mimics the absorption profile of excitation photons with a maximum at 190 µm depth for both absorption (Fig. 3(b)) and detected fluorescence (Fig. 4(a)). On the right column of Fig. 4(b) we observe that 0.8% of fluorescent photons emitted from the GL are detected at 440 nm compared to 1.4% at 530 nm. Like in the SsC model, the absorption properties of OB are higher at 440 nm than at 530 nm (µa-Hb=43.7 cm-1 at 440 nm versus 7.6 cm-1 at 530 nm), which explains that more fluorescent photons are detected at 530 nm. However, as shown on the corresponding OB surface images (Fig. 4(c)), right column), the Fp/NADH-AF intensities ratio is 2, notably less than in the SsC.
3.5 Influence of increases in [Hb]t on the intensity of AF recorded in the SsC and OB.
Activation of SsC and OB networks is followed by an increase in [Hb]t. This in turn modifies the absorption properties of the brain tissues and should lead to a decrease in the detected fluorescent photons. We have studied the effect of increases in [Hb]t (Δ[Hb]t between a basal state and an activated state) on the relative intensity of AF signals. Results are presented on Fig. 5 where increases in absorbed fluorescent photons by SsC and OB are plotted as a function of Δ[Hb]t. In the SsC, for Δ[Hb]t=30%, 50% more fluorescent photons are absorbed at 440 nm and 10% more at 530 nm compared to the basal state (Δ[Hb]t=0%). Corresponding simulated images are shown in Fig. 5(b). The difference between the two wavelengths of emission is already notable at Δ[Hb]t=10%: 20% of initially detected photons are absorbed at 440 nm versus 3% at 530 nm (Fig. 5(a)). Comparatively to the SsC, AF signals in the OB are less affected by Δ[Hb]t. The decrease in intensity for NADH-AF is less than 5% with Δ[Hb]t=10% but 16% with Δ[Hb]t=30%. Interestingly, the Fp-AF signal is almost unaffected by increases in [Hb]t. It decreases only by 2% with Δ[Hb]t=30%.
In the present report, we have used MC simulations in realistic models of the SsC and OB to bring new elements into the debate concerning functional imaging with AF signals. Notably, we showed that for the SsC, Fp-AF is more intense compared to NADH-AF and reflects the propagation of activation to the superficial layers rather than its onset in deeper layers while on the OB both NADH-AF and Fp-AF recordings seem to be feasible. We demonstrate that there is a significant impact of activity-dependent hemodynamic changes on AF time course. In this section, we will discuss these results after a brief summary of the controversial results on activity-dependent AF signals from major reports in the literature.
4.1 Controversial issues on AF signals imaging
Both NADH and Fp-AF signals have been recorded in slices and in intact animals and their relative intensities and time course have been studied through pharmacological experiments [8,12,16]. NADH-AF monitoring in vitro and in vivo has more than 50 years of history, the milestone for in vivo studies being the classical paper by Chance et al. in 1962 . Studies in hippocampal slices showed similar intensities of NADH and Fp-AF signals following electrical stimulation [35,36]. Although NADH AF recordings have been extensively used to study the activity of the respiratory chain in vivo [for a review see 6], recent in vivo studies of brain activation in rodents have rather focused on Fp-AF [8,9,11-13,36]. Successful attempts to monitor Fp-AF in vivo led to signals with unexpected large amplitude [8,12]. For instance electrical stimulation of cerebellar cortex elicited Fp-AF signals with much higher amplitude than NADH-AF signals . These reports left open the problem of the best molecular candidate for functional mapping. Importantly, amplitude of AF signals are not only determined by local concentrations of NADH and Fp but also by hemodynamic changes as pointed out by early experiments in perfused and blood free organs [38,39]. Therefore, in some studies, late in vivo AF signals (>2s after stimulation onset) were considered to be affected by increase in blood flow following stimulation [8,13]. Contrarily, on the basis of pharmacological experiments, Reinert et al.  supported the idea that “hemoglobin absorption and changes in blood flow do not contribute significantly to the AF signal” . Thus, another key issue dealing with the interference of hemodynamics with AF signals time course has to be quantitatively discussed.
4.2 Which is the best candidate for AF functional mapping : NADH or Fp?
The yield of the complete imaging process (i.e. fluorescence excitation and detection) depends on several anatomical and physiological parameters, which have opposite effects on AF signals. Indeed, optical absorption and scattering properties are directly related to the vascularization of the tissues, the density of highly scattering structures, and the dimension and depth of the functional area. Regarding the distribution of the fluorescence sources within the layers, it is related to the distribution of mitochondria where most of the NADH fluorescence occurs and where the Fp are located. Although literature on the mitochondria density in brain tissues is sparse, it is well known that the mitochondria concentration is high at the pre-synaptic and post-synaptic locations where the energy demand is high. The depth and extension of the locations with high synaptic density, hence high NADH and Fp concentrations, vary from one brain structure to another. Considering all these anatomofunctionnal parameters for the OB where AF recordings have not been reported so far, it is not straightforward to state whether such recording will be feasible in vivo. In the OB, the dense micro-vascularization of functional units results in higher absorption properties than in surrounding tissues. In addition, the superficial location of the functional units in the OB a priori favorable to optical recordings is counterbalanced by the high scattering properties of the first layer (ONL) which lead to a large amount of backscattered light and consequently limit photons penetration and fluorescence excitation in the OB. Finally, the small dimensions of the functional units in the OB leads to a lower absolute number of emitted fluorescence photons compared to those emitted from the very large volume of the neocortical column in the SsC. However, considering an optical apparatus with a thin optical section (high NA objective with wide-field epi-fluorescence, or two-photon excitation) one might expect very similar detection abilities from regions with similar properties in terms of optical properties and mitochondrial density. The MC simulations allow integrating all these effects for AF recordings in the SsC and OB. Regarding the location of the detected AF signals, the results for the SsC are in agreement with previous findings in SsC coronal slices showing that the maximum intensity of fluorescence was located in L2 around a cortical depth of 300µm . In the OB, MC simulations show that the maximum fluorescence intensity is found in the GL around a 190µm depth. This is in good agreement with functional MRI studies  and 14C-2-DG autoradiography studies , which have shown that most of the energy-related signals detected in the OB are located in the GL.
Further investigations of the MC results allow quantifying the expected relative intensities of NADH-AF and Fp–AF in both structures. Considering simulated intensities ratios, the absorption properties of blood in brain tissues (table 1 and 2) are more favorable to NADH excitation compared to Fp-AF but more favorable to Fp-AF emission compared to NADH. To evaluate the influence of the absorption from blood we have considered the overall process with its 4 successive steps i) propagation of excitation photons to the endogenous fluorophores, ii) fluorescence with a quantum yield η, iii) propagation of fluorescence photons and iv) detection of exiting photons. Our simulations confirm that NADH excitation is favored compared to Fp excitation in the SsC (Fig. 2(c)) and OB (Fig. 3(c)) with ratios of 3 and 4 between the cellular absorption at 350 and 440 nm respectively. Although quantum yields are likely to be affected by physiological conditions (pH, ions, molecular conformation), to our knowledge only in vitro values are available for NADH and Fp and are respectively 0.019 and 0.025 . The third and last steps of the process were combined in the second series of MC simulations. The results (Fig. 4(a)) showed a pronounced advantage for Fp-AF in SsC (ratio of 7) and less pronounced but yet significant advantage for Fp-AF in the OB (ratio of 2). Considering the whole process including the quantum yield, simulations show a ratio of 3.4 in favor of the Fp-AF in the SsC and a ratio of 1.6 in favor of NADH in the OB. It is noticeable that for the SsC the ratio of Fp-AF to NADH-AF intensities derived from our simulations is comparable to that observed experimentally in vivo . In addition, as already observed with other intrinsic optical signals  our results point out that AF intensity varies from one brain region to another. Considering the basal hemodynamics of these structures, our simulations show that in terms of intensities, the Fp signal is the best choice for in vivo functional imaging in the SsC whereas NADH-AF and Fp-AF are both good candidates for AF recordings in the OB.
4.3 Do hemodynamics interfere with AF signals time course?
The time course of AF signals was studied in brain slices 2 [8,16] as well as in vivo [8,12] and was shown to be related to the duration and intensity of the stimulation. All studies have shown biphasic profiles for AF signals, which are attributed to changes in the redox states of NADH and Fp. After stimulation onset, Fp-AF and NADH-AF present inversed temporal profiles with an initial rapid decrease of NADH-AF (corresponding to a fast Fp-AF increase) followed by a slow increase of NADH-AF intensity (corresponding to a slow Fp-AF decrease). To study the influence of hemodynamics on the time course of AF signals, we have simulated the activation-induced transient increase in blood flow , which affects the optical properties of tissues. It is important to note that the time course of hemodynamic changes is much slower compared to that of AF signals, which are more closely related to electrical transmission [8,12]. In the SsC, the maximum Δ[Hb]t was reported to occur 2 to 3 seconds following the end of a 1 second electrical stimulation [26-29,44,45]. In the OB, an increase in blood flow was observed about 2 seconds after olfactory stimulation onset [15,21]. Considering our simulations, we can assume that the AF-signals are almost left unaffected by hemodynamic changes during the first 1-2 seconds following activation (corresponding to Δ[Hb]t of about 10-20%) except for NADH–AF in the SsC. Considering longer delays after stimulation onset, an important proportion of fluorescent photons is absorbed due to increase in optical absorption of the tissues. As expected from the absorption properties of blood, the effect is much stronger for NADH than for Fp. It is also much stronger in the SsC than in the OB since fluorescent photons have to travel for a longer path in the SsC compared to OB. In vivo experiments where hemodynamic changes were blocked by nitric oxide synthase inhibitors have been carried out to separate AF from subsequent hemodynamic responses, but have led to different conclusions. On the one hand, Reinert et al. (2004) showed that in the cerebellar cortex L-nitroarginine-methyl ester does not affect Fp-AF intensities at different stimulation conditions (n=3, Fig. 9 and Fig. 10 within ). On the other hand, Shibuki et al. showed that Fp-AF signals recorded under NG-nitro-L-arginine challenge are about twice more intense compared to the control experiment. In the same study the second phase of the AF signals is temporally delayed by one second (Fig. 7(a) and Fig. 7(d) in ). In the SsC, our results lean towards the findings of Shibuki et al. since they show that for Δ[Hb]t=30% about 10% of the signal is absorbed for Fp-AF and about 50% for NADH–AF. Furthermore, in some experiments in the SsC, the Fp-AF signals reach their maximum and start to decrease before the end of the stimulation , which implies that Fp-AF signals are affected by hemodynamic changes within seconds following stimulation onset. Our results support and quantify these findings and the methodological choice made by Weber et al  to exclude the late phase (>2s after stimulation onset) from the Fp-AF recordings to map brain activation accurately.
Using MC simulations of AF signals in realistic models of the rat SsC and OB, we found that the feasibility of AF recordings in different brain areas is highly dependent on local and specific anatomofunctional features. For the SsC, we showed that Fp-AF signal is significantly more intense compared to NADH-AF in agreement with previous studies carried out in vivo. Thus, the Fp signal is the best choice for in vivo functional imaging in the SsC whereas NADH-AF and Fp-AF are both good candidates for AF recordings in the OB. In addition, we demonstrated that Fp-AF recordings in the SsC reflect the propagation of activation to the superficial layers rather than its onset in deeper layers. Finally, our simulations allowed quantifying the influence of hemodynamics on AF signals intensities and time course. Consequently, to obtain accurate activation maps using AF signals it is important to carefully consider the contamination of AF signals by subsequent hemodynamic changes especially for AF-Fp imaging in deep structures.
This work was supported by ANR grant ANR-06-NEURO-004-01/Astroglo and the BQR program from Université Paris Diderot - Paris VII. Ph-D scholarship of B. L’Heureux is funded by CNRS-IN2P3. We thank Dr C. Martin for helpful comments on the manuscript.
References and links
1. T. Bonhoeffer and A Grinvald “Optical Imaging based on intrinsic signals: the methodology” in Brain mapping; the methods”. A.W. Toga and J.C. Mazziotta, Eds. (Academic Press, Los Angeles, CA, 1996).
4. M. Hashimoto, Y. Takeda, T. Sato, H. Kawahara, O. Nagano, and M. Hirakawa, “Dynamic changes of NADH fluorescence images and NADH content during spreading depression in the cerebral cortex of gerbils,” Brain. Res. 872, 294–300 (2000). [CrossRef] [PubMed]
6. A. Mayevsky and G.G. Rogatsky, “Mitochondrial function in vivo evaluated by NADH fluorescence: from animal models to human studies,” Am. J. Physiol. Cell. Physiol. 292, C615–40 (2007). [CrossRef]
7. K.C. Reinert, R.L. Dunbar, W. Gao, G. Chen, and T.J. Ebner, “Flavoprotein autofluorescence imaging of neuronal activation in the cerebellar cortex in vivo,” J. Neurophysiol. 92,199–211 (2004). [CrossRef] [PubMed]
8. K. Shibuki, R. Hishida, H. Murakami, M. Kudoh, T. Kawaguchi, M. Watanabe, S. Watanabe, T. Kouuchi, and R. Tanaka., “Dynamic imaging of somatosensory cortical activity in the rat visualized by flavoprotein autofluorescence,” J. Physiol. 549, 919–27 (2003). [CrossRef] [PubMed]
9. H. Murakami, D. Kamatani, R. Hishida, T. Takao, M. Kudoh, T. Kawaguchi, R. Tanaka, and K. Shibuki, “Short-term plasticity visualized with flavoprotein autofluorescence in the somatosensory cortex of anaesthetized rats,” Eur. J. Neurosci. 19, 1352–60 (2004). [CrossRef] [PubMed]
10. M. Tohmi, H. Kitaura, S. Komagata, M. Kudoh, and K. Shibuki, “Enduring critical period plasticity visualized by transcranial flavoprotein imaging in mouse primary visual cortex,” J. Neurosci. 26, 11775–85 (2006). [CrossRef] [PubMed]
11. Y. Kubota, D. Kamatani, H. Tsukano, S. Ohshima, K. Takahashi, R. Hishida, M. Kudoh, S. Takahashi, and K. Shibuki, “Transcranial photo-inactivation of neural activities in the mouse auditory cortex,” Neurosci. Res. 60, 422–30 (2008). [CrossRef] [PubMed]
13. B. Weber, C. Burger, M.T. Wyss, G.K. von Schulthess, F. Scheffold, and A. Buck, “Optical imaging of the spatiotemporal dynamics of cerebral blood flow and oxidative metabolism in the rat barrel cortex,” Eur. J. Neurosci. 20, 2664–70 (2004). [CrossRef] [PubMed]
15. G.C. Petzold, D.F. Albeanu, T.F. Sato, and V.N. Murthy, “Coupling of neural activity to blood flow in olfactory glomeruli is mediated by astrocytic pathways,” Neuron 58, 897–910 (2008). [CrossRef] [PubMed]
16. C.W. Shuttleworth, A.M. Brennan, and J.A. Connor, “NAD(P)H fluorescence imaging of postsynaptic neuronal activation in murine hippocampal slices,” J Neurosci. 23, 3196–208 (2003). [PubMed]
17. S. A. Prahl, “Optical Absorption of Hemoglobin,” http://omlc.ogi.edu/spectra/hemoglobin/index.html
19. T.A. Woolsey, C.M. Rovainen, S.B. Cox, M.H. Henegar, G.E. Liang, D. Liu, Y.E. Moskalenko, J. Sui, and L. Wei, “Neuronal units linked to microvascular modules in cerebral cortex: response elements for imaging the brain,” Cereb. Cortex. 6, 647–60 (1991). [CrossRef]
20. G.M. Shepherd, “Synaptic organization of the mammalian olfactory bulb,” Physiol. Rev. 52, 864–917 (1972). [PubMed]
21. E. Chaigneau, M. Oheim, E. Audinat, and S. Charpak, “Two-photon imaging of capillary blood flow in olfactory bulb glomeruli,” Proc. Natl. Acad. Sci. U. S. A. 100, 13081–6 (2003). [CrossRef] [PubMed]
22. M. Kohl, U. Lindauer, G. Royl, M. Kuhl, L. Gold, A. Villringer, and U. Dirnagl., “Physical model for the spectroscopic analysis of cortical intrinsic optical signals,” Phys. Med. Biol. 45, 3749–64 (2000). [CrossRef] [PubMed]
23. N. Plesnila, C. Putz, M. Rinecker, J. Wiezorrek, L. Schleinkofer, A.E. Goetz, and W.M. Kuebler, “Measurement of absolute values of hemoglobin oxygenation in the brain of small rodents by near infrared reflection spectrophotometry,” J. Neurosci. Methods 114, 107–17 (2002). [CrossRef] [PubMed]
25. A.N. Yaroslavsky, P.C. Schulze, I.V. Yaroslavsky, R. Schober, F. Ulrich, and H.J. Schwarzmaier, “Optical properties of selected native and coagulated human brain tissues in vitro in the visible and near infrared spectral range,” Phys. Med. Biol. 47, 2059–73 (2002). [CrossRef] [PubMed]
26. E.M. Hillman, A. Devor, M.B. Bouchard, A.K. Dunn, G.W. Krauss, J. Skoch, B.J. Bacskai, A.M. Dale, and D.A. Boas, “Depth-resolved optical imaging and microscopy of vascular compartment dynamics during somatosensory stimulation,” Neuroimage 35, 89–104 (2007). [CrossRef] [PubMed]
27. M. Jones, J. Berwick, and J. Mayhew, “Changes in blood flow, oxygenation, and volume following extended stimulation of rodent barrel cortex,” Neuroimage 15,474–87 (2002). [CrossRef]
28. J. Mayhew, Y. Zheng, Y. Hou, B. Vuksanovic, J. Berwick, S. Askew, and P. Coffey, “Spectroscopic analysis of changes in remitted illumination: the response to increased neural activity in brain,” Neuroimage 10, 304–26 (1999). [CrossRef] [PubMed]
29. A. Devor, A. K. Dunn, M. L. Andermann, I. Ulbert, D. A. Boas, and A. M. Dale1 “Coupling of total hemoglobin concentration, oxygenation, and neural activity in rat somatosensory cortex,” Neuron 39, 353–9 (2003). [CrossRef] [PubMed]
30. S. A. Prahl, M. Keijzer, S. L. Jacques, and A. J. Welch, “A Monte Carlo Model of Light Propagation in Tissue,” Proc. SPIE IS 5, 102–11(1989).
32. F. Agner, “Pseudo random number generator;” http://www.agner.org/random/mother.
33. B. Chance, B. Schoener, R. Oshino, F. Itshak, and Y. Nakase, “Oxidation-reduction ratio studies of mitochondria in freeze-trapped samples. NADH and flavoprotein fluorescence signals,” J. Biol. Chem. 254, 4764–71 (1979). [PubMed]
35. A.M. Brennan, J.A. Connor, and C.W. Shuttleworth, “Modulation of the amplitude of NAD(P)H fluorescence transients after synaptic stimulation,” J. Neurosci. Res. 85, 3233–43 (2007). [CrossRef] [PubMed]
36. A.M. Brennan, J.A. Connor, and C.W. Shuttleworth, “NAD(P)H fluorescence transients after synaptic activity in brain slices: predominant role of mitochondrial function,” J. Cereb. Blood Flow. Metab. 26, 1389–406 (2006). [CrossRef] [PubMed]
38. R. Scholz, R.G. Thurman, J.R. Williamson, B. Chance, and T. Bücher., “Flavin and pyridine nucleotide oxidation-reduction changes in perfused rat liver. I. Anoxia and subcellular localization of fluorescent flavoproteins,” J. Biol. Chem. 244, 2317–24 (1969). [PubMed]
39. F.F. Jöbsis, M. O’Connor, A. Vitale, and H. Vreman, “Intracellular redox changes in functioning cerebral cortex. I. Metabolic effects of epileptiform activity,” J. Neurophysiol. 34, 735–49 (1971). [PubMed]
40. F. Xu, I. Kida, F. Hyder, and R.G. Shulman, “Assessment and discrimination of odor stimuli in rat olfactory bulb by dynamic functional MRI,” Proc Natl Acad Sci U S A. 97, 10601–6. (2000). [CrossRef] [PubMed]
41. B.A. Johnson and M. Leon, “Spatial distribution of [14C]2-deoxyglucose uptake in the glomerular layer of the rat olfactory bulb following early odor reference learning,” J Comp Neurol. 37, 6557–66. (1996)
42. O. Wolfbeis “Fluorescence of organic natural products,” in Molecular Luminescence Spectroscopy. Part 1: Methods and Applications, John Wiley and Sons, ed. (S. G. Schulman, 1985), pp. 167–370.
43. A. Grinvald, E. Lieke, R.D. Frostig, C.D. Gilbert, and T.N. Wiesel, “Functional architecture of cortex revealed by optical imaging of intrinsic signals,” Nature 324, 361–4 (1986). [CrossRef] [PubMed]
45. Prakash, J.D. Biag, S.A. Sheth, S. Mitsuyama, J. Theriot, C. Ramachandra, and A.W. Toga, “Temporal profiles and 2-dimensional oxy-, deoxy-, and total-hemoglobin somatosensory maps in rat versus mouse cortex,” Neuroimage 37 Suppl 1, S27–36 (2007). [CrossRef] [PubMed]