During experiments, the region of the left parietal somatosensory barrel cortex centered at the whisker C2 functional representation location was monitored pre- and post-MCA occlusion. A rat brain illustration together with the cranial window for optical imaging and the approximate position of the left MCAo are presented in . The enlarged region photograph of the cranial window is shown in . The size of this ROI is about 4(V)×4.9(H) mm2. The relationship between the measured diffuse reflectance from brain tissue and the spatial modulation frequency is shown in for 680-nm light. As demonstrated by this plot, the brain tissue acts like a low-pass spatial filter, attenuating the reflectance more strongly as the spatial frequency increases. The data was fit to the MC model using a nonlinear least-squares optimization routine. As can be seen, the numerical calculations are in quantitative agreement with experimental data with low fitting error. These errors are believed to be mainly due to the phantom calibration process and the assumption of homogeneity in the computational model.
Once the information of is obtained, optical coefficients are retrieved by using an MC model, as discussed previously in Sec. 2.5. As mentioned in the introduction, mapping the absorption coefficient at each wavelength with the knowledge of the wavelength-dependent molar extinction (ε
) values of the chromophores15-17
enables us to calculate tissue chromophore concentration (C
) separately at each pixel using the following equation18
where the brackets [ ] represent a matrix, the bold type indicates a vector, and λ
is the wavelength. We assume that the measured absorption spectra are linear combinations of the component chromophore spectra, and we solve the equation by adjusting μa
into the appropriate vectors, ε
into a matrix, and solving the matrix equation.
An example of the changes in the ctO2Hb, ctHHb, and stO2 maps before and after the occlusion is shown in . The grayscale to the right of each map represents the absolute chromophore value of each pixel in the map. To better visualize changes, mean color scales were set at 50% for all of the preocclusion data. No a priori knowledge about tissue chromophore content has been assumed (other then standard nonnegative constrains). Averaging over each of these maps we found a 22±10.2% increase in deoxyhemoglobin concentration and a 43±6.6% and 32±4.1% decrease in oxyhemoglobin and oxygen saturation, respectively, from baseline values following MCA occlusion. These changes reflect the pathophysiologic state of the brain and the ability of spatially modulated light to quantify changes in chromophore concentrations with time.
Fig. 3 Results from a representative rat (out of seven). (a) Oxy-, deoxyhemoglobin (μM) and oxygen saturation (%) maps pre-and post-MCAo. In all panels, higher concentration values correspond to brighter pixels. Scale values are represented by the scale (more ...)
A representative time course of the absolute changes in chromophore concentration within the entire cortex of pre-MCAo (baseline) and post-MCAo are presented in . A decrease in the ctTHb from baseline in the amount of 19±7.3% is observed. In the intact brain, ctTHb tracks blood flow as long as the hematocrit is constant, but in thrombosed vessels, this relationship may change. For example, if blood flow decreases or stops but the vessels do not collapse, ctTHB can remain high in the occluded region.7,14
However, our signals are derived primarily from superficial microvasculature a few millimeters in depth beneath the skull. Because these vessels are significantly downstream from the MCA occlusion site, we expect that the observed reduction in ctTHB is a consequence of stroke-induced termination of flow.
Fluctuations in chromophore concentrations pre- and post-MCAo can also be seen in . Mean values do not exceed more then ±3 μM for hemoglobin and ± 3% for stO2. As previously stated in Sec. 2.4, measurement of brain optical properties and generation of an image (a unit measurement) required 1.5 min. Thus, each unit measurement is an average of the dynamic optical properties over 1.5 min. The nature of these fluctuations may change after the stroke and thus account for additional measurement errors. With further advances in technology and faster image acquisition, this potential source of error is likely to be reduced.
While light absorption is related to the concentrations of individual chromophores, light scattering is related to tissue structural properties such as density, size, and distribution of scattering. Tissue scattering properties are believed to depend on intracellular (nuclei, mitochondria) and extracellular (protein) structures.19
Therefore, examination of the scattering behavior can provide fundamental insight into microstructural physiology. Extracting the mean particle size and number density of scattering particles from the reduced scattering co-efficient spectrum in breast tissue was discussed recently by Wang et al.20
In the following, we report our observation on the behavior of scattering properties during brain ischemia. Several studies have shown that the wavelength dependence of scattering in tissue in the NIR follows a power-law dependence of the form21,22
is the scattering amplitude, and sp (sometimes indicted by b
) is the scattering power. It has been shown that A
and sp are related to geometrical properties; a decrease in sp reflects an increase in scatterer size and vice versa.23
Decreases in A
occur with changes in density (number density of the scatterer), distribution (histogram of particle number density) and refractive index.20
Changes in the reduced scattering coefficient, scattering amplitude, and scattering power over time for the entire ROI (single animal) shown in are plotted in , respectively. The mean scattering values pre-MCA occlusion are ~4% higher than postocclusion. The decrease in the scattering power with ischemia can be explained by an increase in the average size of the scatterers. This could be the result of cell swelling in response to acute injury.24
Changes in scattering amplitude (A
) are influenced by (1) number of particles, (2) the ratio of the refractive index outside (extra-cellular) to inside (intracellular) the scattering particle, and (3) particle distribution. Since 1 and 3 are assumed not to change in the first hour of ischemic injury, the measured drop in scattering amplitude most likely reflects a homogenization of refractive index due to injury-induced alterations in cell and organelle permeability. This reduction in spatial variation of refractive index causes more light to passes through the cortex, and less light to be scattered. Hence, by measuring scattering amplitude and power one can get an impression of the degree of cellular injury in terms of cell swelling (edema) and membrane damage. A summary of scattering changes for all animals is presented in .
Time course of the quantitative values of scattering properties (μ’s, sp, and A) before and after MCA occlusion obtained from the entire ROI of .
Statistics comparing brain properties pre-MCAo and 45 min post-MCAo over the entire cortex for seven rats.
Regional changes in cortical properties can also observed by converting absorption and scattering maps into histogram distributions before and after the MCAo, as shown in and , respectively. illustrates that when a wide ROI is selected, overlap between histograms diminishes contrast between pre- and post-MCAo optical properties. This is a result of partial-volume effects (in all three spatial dimensions) of sampling heterogeneous tissue. When a more homogeneous ROI is selected (), pre- and post-MCAo histograms are clearly separated. This is an expected result as we sample a more homogenous region of the cortex. Note that mean values of each property in are almost identical to those in . However, values for the variance are reduced, for example, by 67, 63, 25, and 38% for absorption, scattering, scatter amplitude, and scatter power, respectively, in the homogenous region ().
(a) and (b) Absorption and (c) and (d) scattering property histograms before (solid lines) and after (dashed lines) MCAo for the entire cortex.
(a) and (b) Absorption and (c) and (d) scattering property histograms before (solid lines) and after (dashed lines) MCAo for the parenchyma. The small box denotes the ROI used for analysis.
summarizes the results of statistical comparisons before and 45 min after MCAo for ctO2
Hb, ctHHb, ctTHb, stO2
, and the three scattering parameters for the seven rats used in this study. For each parameter, we calculated both the mean and the median for the entire cortical ROI. In all analyses, a difference of p
<0.05 was regarded as statistically significant. The results of the analysis show that all chromophores display statistically highly significant differences between pre- and post-MCAo. The percent changes apparent from the table show a 17±4.7% increase in tissue concentration of deoxyhemoglobin and 45±3.1, 23±5.4, and 21±2.2% decreases in oxyhemoglobin, total hemoglobin concentration, and cerebral tissue oxygen saturation levels, respectively, following induction of cerebral ischemia. Our findings are consistent with previous observations, considering the variance in measurements, and agree well with known directions of change with cerebral ischemia. For comparison, Culver et al.7
reported chromophore baselines for ctO2
Hb (55 to 85 μ
M), ctHHb (23 to 41 μ
M), ctTHb (81 to 125 μ
M), and stO2
(65 to 73%) in a rat model. The authors also observed a 15% maximum decrease of stO2
after MCAo, while we observe a maximum decrease of 21% in stO2
following MCAo. In a human subject, an increase of 20% from baseline in deoxyhemoglobin and 42, 18, and 25% decreases in oxyhemoglobin, total hemoglobin, and brain tissue oxygen saturation concentrations, respectively, were reported.10
Ischemia results in higher concentrations of ctHHb due to ongoing metabolic activity, and lower ctO2 Hb with reduced hemoglobin delivery. These physiological parameters can be grouped into a simple ischemic index, Iisch=ctHHb/ctO2Hb.Average values of Iisch for seven rats increased more then twofold after MCAo from 0.66±0.08 to 1.52±0.28 (). To visualize ischemic spatial variations before and after MCAo, we show Iisch images for a representative rat in . A cross-sectional profile from the left vein towards the right vein was applied as shown in . The graphs in represent the gradient profile of Iisch before and after MCAo, respectively. Overall, our results show approximately threefold increase in Iisch levels after MCAo. Prior to occlusion, we observe that Iisch varies across the cortex, with significantly higher levels in the cortical veins (0.83 to 1) compared to the central parenchymal region of mixed microvasculature (0.65 to 0.75). As expected the Iisch spatial profile after MCAo remained approximately the same, but overall levels increased (~2.6 to 3 in veins versus ~1.7 to 2.2 in parenchyma).
Ischemic index (Iisch=ctHHB/ctO2Hb) gradient profile measured before (left) and after (right) MCAo. The dotted lines represents the cross-sectional view along the two cortical veins. Field of view ~6×6 mm.