Large field-of-view movement-compensated intrinsic optical signal imaging for the characterization of the haemodynamic response to spreading depolarizations in large gyrencephalic brains

Abstract

Haemodynamic responses to spreading depolarizations (SDs) have an important role during the development of secondary brain damage. Characterization of the haemodynamic responses in larger brains, however, is difficult due to movement artefacts. Intrinsic optical signal (IOS) imaging, laser speckle flowmetry (LSF) and electrocorticography were performed in different configurations in three groups of in total 18 swine. SDs were elicited by topical application of KCl or occurred spontaneously after middle cerebral artery occlusion. Movement artefacts in IOS were compensated by an elastic registration algorithm during post-processing. Using movement-compensated IOS, we were able to differentiate between four components of optical changes, corresponding closely with haemodynamic variations measured by LSF. Compared with ECoG and LSF, our setup provides higher spatial and temporal resolution, as well as a better signal-to-noise ratio. Using IOS alone, we could identify the different zones of infarction in a large gyrencephalic middle cerebral artery occlusion pig model. We strongly suggest movement-compensated IOS for the investigation of the role of haemodynamic responses to SDs during the development of secondary brain damage and in particular to examine the effect of potential therapeutic interventions in gyrencephalic brains.

Keywords: Brain imaging; acute stroke; animal models; intrinsic optical imaging; spreading depression.

Figure 1.

Illustration of the process to warp an image for an optimal fit to the reference image. In the right column, the difference images of the left and middle images are shown. Note, how the difference decreases successively until Step 3. Most of the remaining difference in Step 3 can be explained by changed vessel diameters that are not compensated. In Step 4, the difference increases again due to low frequency portions, which have been eliminated in Step 2 and are reintroduced in this step. The entire process is described in detail in the text. AR: affine registration; HP: high-pass filter; ER: elastic registration.

Figure 2.

Illustration of movement-compensation quality. (a) (I): Part of one hemisphere after MCAO. The white line shows the line of interest for (II) and (III). The arrow points to the point of interest shown in (IV). (II, III): Intensity along the line of interest (along the horizontal axis) over a timespan of 1 h (vertical axis). The image in (II) is taken from the unprocessed data and exhibits strong movement artefacts. The image in (III) is extracted from the movement-compensated images series and shows the successful compensation of the movement artefacts. Three SDs are visible as nearly vertical stripes of first brighter and later slightly darker intensity. Horizontal dark stripes are caused by blood vessels. The arrows on the left point to the line for which the time course of inverted intensities is shown in (IV). (IV) Time courses of IOS_inv at a selected POI in unprocessed (upper trace) and registered data (lower trace). The vertical and horizontal scales are the same in both traces. (b) Exemplary IOS_inv profiles of relative intensity changes of paper (POIs 1 and 2) and normal perfused tissue (POIs 3 to 6). Changes of intensity in POIs 3 to 6 are mainly caused by low frequency vascular fluctuations and a propagating SD, which was elicited by KCl-application at the location marked with the cross. Vertical scaling is the same for all POIs.

Figure 3.

Comparison of IOS and LSF in an MCAO model. (a) Movement-related artefacts in IOS_inv (top) and LSF_inv (bottom). Dashed boxes mark the occurrence of (unusually) strong general movements, which cause considerable artefacts in LSF, whereas the artefact is hardly visible in movement-compensated IOS. However, even outside of episodes of general movement, mechanical ventilation and heart beat lead to more noise in LSF than in IOS. (b) IOS_inv and LSF_inv of 14 consecutive SDs at two different POIs. Upper traces are IOS_inv and lower traces are LSF_inv. Horizontal and vertical scalings are the same for all charts. The signals were divided by the mean intensity of the shown period to normalize both modalities and diminish illumination differences. The black, thicker line represents the mean of all 14 SDs of the corresponding POI. To improve signal-to-noise ratio, only the LSF signal was smoothed by a median filter (with a width of 10 samples at a sampling rate of one image per second). Left side: Mainly normoxic response from tissue outside of the affected area. Right side: Mainly inverse responses from tissue of a hypoperfused area.

Figure 4.

Spatial distribution of various haemodynamic parameters from a single exemplary SD, which propagated over the whole visible part of the cortex about 1 h after MCAO. (a) Reference image with the ROIs shown in (c). The dotted region is the most hypoperfused visible area, presumably the core at the moment of the SD. (b) Effect of the clipping as maximum IOS_inv decrease during the first seconds after clipping (upper image) and the subsequent IOS_inv increase at 60 s after clipping, caused by lateral perfusion (lower image). (c) Parameters and IOS from the POIs shown in A with increasing distance from the supposed core. The first column denotes the POI number, the middle two columns quantify the effect of the clipping and subsequent lateral perfusion. The last column states the LSF_inv baseline reduction just before the SD as compared with just before clipping. Example: In POI 3, the clipping lead to a 9% IOS_inv decrease with a subsequent increase to a level of 18% above baseline. The LSF_inv just before the shown SD was 47% lower than just before the clipping. We can observe an inverse response in both modalities. (d–f) Amplitudes (IOS and LSF) and durations (only IOS) of initial hypoperfusion, hyperemia and oligemia, respectively. Duration of oligemia could not be determined, because of subsequent SDs.

Figure 5.

Patterns of IOS_inv time courses. (a) Exemplary profiles of IOS_inv of the most frequently observed patterns. From left to right: Hypoperfusion with following hyperemia; hypoperfusion with following hyperemia and oligemia; hyperemia with following oligemia. Below: Bar graph of the fraction of POIs with these three patterns for KCl (upper bar) and MCAO (lower bar). Of importance in the MCAO model is the larger proportion of the first two patterns, which include a considerable initial hypoperfusion, resulting from the region affected by the clipping. (b, c) Two clusters of SDs with different blood flow responses and subsequently following prolonged oligemia after termination of the cluster.