Multispectral optical measurements have been widely used to assess changes in cortical concentrations of oxy- and deoxyhemoglobin from changes in light reflectivity as predicted by the modified Beer-Lambert law. The work of Kohl34
explored the use of the modified Beer-Lambert approximation in intrinsic optical imaging. Up to now, the changes measured have been due to functional activation16,32,40
and as such have been smaller perturbations than observed here. As a result, linearization has been considered adequate to calculate changes in HbR and HbO concentration. In focal cerebral ischemia, the changes measured are much larger; therefore, we did not assume that linearization was sufficient. The severe changes in HbO and HbR during ischemia lead to significant changes in the path lengths of the detected photons. Based on Monte Carlo simulations, we estimated that the photon path length in the ischemic core increases by more than 60% compared with baseline conditions (~0.6 mm versus ~0.95 mm). While the change in path length might in principle affect the laser speckle measurement, in practice, the wavelength used is in a region where little absorption occurs and the path length is short; therefore, path length changes are unlikely to have a significant effect.
Numerical simulations and comparisons of results show that the use of the Beer-Lambert approximation in the analysis of ischemia data leads to serious errors, most seriously a large overestimation of the reduction in HbO, leading to calculated values of HbO of less than zero. For this reason, we solved for changes in HbO and HbR from reflectance data using a nonlinear minimization method in a manner similar to that used in other measurements of tissue optical properties36,42–44
. While this method has the disadvantage of requiring a large increase in computational time, it is practical to extract multiple time courses for relevant regions of interest. In this work we have used the Beer-Lambert approximation to extract qualitative spatial maps of parameter changes that can be useful for observations of hemodynamic behavior (see .) With additional optimization, the nonlinear fitting procedure should be capable of producing quantitative spatiotemporal maps of hemodynamic changes.
There has been much interest in the measurement of CMRO2
in the ischemic and injured brain in both experimental animals45–47
Metabolic measures are thought to be strong indicators of tissue outcome, with oxygen and glucose metabolism being closely linked.50
LSF imaging has previously been applied to cerebral ischemia,5,22
and its advantages over laser Doppler flowmetry have been previously established in terms of the free placement of ROIs and increased spatial and temporal resolution.19
Other groups have reported similar decreases in CMRO2
at isolated time points post-ischemia. For example, Singh et al.45
found a 60% decrease, 20 min post global brain ischemia (neck tourniquet), that lasted to 6 h post-ischemia. Temma et al.51
also showed approximately 50% reduction in CMRO2
1 h after transient MCA occlusion in rats. There have been positron emission tomography (PET) studies in primates and cats52,53
that have shown reductions in CBF of ≈35 and 60%, respectively, and reductions in CMRO2
levels of ≈35%. In addition, they report an increase in oxygen extraction fraction,19
which we do not observe. A study with near infrared spectroscopy (NIRS) and Doppler flowmetry in rats has shown a less severe reduction in CBF (≈31%), although encouragingly, similar time courses during PIDs were observed.54
Also in a gerbil model, Li et al.55
reported a drop in HbO of ≈50% and an increase in HbR of ≈40%. While less severe that our core results, these data are qualitatively similar; the discrepency may simply be a function of our improved spatial resolution. Taken togther, our data are consistant with the wider stroke literature. It is important to note that while the various reports vary slightly, they are conducted in different animal models; we are specifically looking at the cortex, and our technique has higher resolution. In addition, it is difficult to compare across modalities, as simultaneous reporting of all hemodynamic parameters is technically challenging.
Here, we demonstrate measurement of CMRO2 changes during AD and subsequent PIDs over regions of interest in three different areas of the cortex in relation to arterial occlusion. We found that immediately after MCA occlusion, CBF dramatically decreases with relatively little change in HbT, even in the core, suggesting that vascular caliber is not altered during this initial phase. An increase in HbR and a decrease in HbO are observed, as the sluggish blood spends more time in the capillary bed and is deoxygenated. In the nonischemic cortex, there was little effect on any hemodynamic parameter during this early phase (i.e., prior to the spread of the first PID). After approximately one minute, HbT decreases in the core, suggesting either passive vascular collapse or active vasoconstriction. This event is accompanied by a further reduction in CBF and oxygenation. As the PID triggered by AD spreads into the nonischemic cortex, secondary reductions in HbT and CMRO2 are observed there as well.
During PIDs, CMRO2
is reduced in each of the three regions of interest. Following this initial drop, there is a CMRO2
overshoot in the nonischemic cortex that does not occur in the ischemic core. This overshoot may be a useful measure of integrity of neurovascular coupling. The drop in CMRO2
in the core and the lack of a CMRO2
overshoot suggest that oxygen consumption in the core is delivery limited. Further, it may be possible that the maintenance of neurovascular coupling is a measure of tissue viability. The ischemic core invariably progresses to infarct, whereas the penumbra can either recover or proceed to infarct as verified by histological studies.56
More work is needed to determine whether the neurovascular coupling response to PID is correlated with the viability thresholds of tissue. It is important to note that intense neuroglial depolarizations may impact infarct outcome differently, compared to more physiological modes of depolarization (e.g., somatosensory activation via electrical forepaw stimulation), which have recently been shown to reduce cortical and striatal infarct volumes in rats undergoing MCA occlusion.57
The difference here could be that PIDs are likely to cause higher metabolic demand than forepaw stimulation and are associated with vasoconstrictive coupling rather than vasodilation.
Our technique generates spatiotemporal qualitative maps of hemodynamic parameters, allowing the observation of interesting hemodynamic events. See for a typical dMCAO experiment. The video shows AD; multiple PIDs, which can be seen moving around the nonischemic cortex; reperfusion; and ultimately, the death of the animal by induced anoxia. Advancing the first few frames of the video frame by frame shows in detail how the initial change in HbO and HbR precedes the change in HbT and leads to the PID that spreads quickly over the cortex.