Spatial frequency-based analysis of mean red blood cell speed in single microvessels: investigation of microvascular perfusion in rat cerebral cortex

Abstract

Background: Our previous study has shown that prenatal exposure to X-ray irradiation causes cerebral hypo-perfusion during the postnatal development of central nervous system (CNS). However, the source of the hypo-perfusion and its impact on the CNS development remains unclear. The present study developed an automatic analysis method to determine the mean red blood cell (RBC) speed through single microvessels imaged with two-photon microscopy in the cerebral cortex of rats prenatally exposed to X-ray irradiation (1.5 Gy).

Methodology/principal findings: We obtained a mean RBC speed (0.9±0.6 mm/sec) that ranged from 0.2 to 4.4 mm/sec from 121 vessels in the radiation-exposed rats, which was about 40% lower than that of normal rats that were not exposed. These results were then compared with the conventional method for monitoring microvascular perfusion using the arteriovenous transit time (AVTT) determined by tracking fluorescent markers. A significant increase in the AVTT was observed in the exposed rats (1.9±0.6 sec) as compared to the age-matched non-exposed rats (1.2±0.3 sec). The results indicate that parenchyma capillary blood velocity in the exposed rats was approximately 37% lower than in non-exposed rats.

Conclusions/significance: The algorithm presented is simple and robust relative to monitoring individual RBC speeds, which is superior in terms of noise tolerance and computation time. The demonstrative results show that the method developed in this study for determining the mean RBC speed in the spatial frequency domain was consistent with the conventional transit time method.

Figure 1. Representative image and measurement of RBC speed in single microvessels of the exposed animals.

(A) A raw image of RBC moving through a microvessel was obtained using a line-scan mode with two-photon microscopy. The representative image showed 512 lines captured at the center of the target microvessel in parallel to the longitudinal direction. The x-axis represents Δt of 1 msec/pixel (time domain), and the y-axis is Δx of 0.20 µm/pixel (spatial domain). The green color represents the measured intensity of the fluorescent signal that originated from the injected plasma marker. Dark streaks observed around the center of the image were mainly caused by unlabelled RBC motions, and their slopes reflect a speed of the RBC motions that were parallel to the vessel length (a scan direction). (B) Power spectrum image. The image represents a FFT image constructed from the original 512 by 512 pixel image (A). The power spectrum was used to characterize a periodic pattern of the pixel intensity distribution represented in the raw image (A). A slanted line reflects a direction preference, perpendicular to the RBC traces (i.e., a slope of dark streaks), which appeared in the original image. The angle between this slanted line and the temporal axis was used to calculate the mean RBC speed (see text). (C) The spatial pattern of the pixel intensity distribution converted by the FFT analysis. Summation of the power spectrum at each angle was calculated for all directions (-0.5 to 0.5 π), and the angle that had the maximum power (arrow head) was used to calculate mean RBC speed (1.1 mm/sec in this representative image) (see Eq. 1).

Figure 2. Mean RBC speed in single microvessels.

The histogram shows the frequency distribution of the mean RBC speed obtained from all 121 vessels in the exposed rats (N = 4). The mean RBC speed ranged from 0.2 to 4.4 mm/sec (minimum to maximum), and the mean of all measurements was 0.9±0.6 mm/sec.

Figure 3. Representative image and measurement of AVTT.

(A) A representative image of the appearance time obtained from one animal. Time-lapse imaging was performed at the cortical surface with a frame-capture rate of 14.2 frame/sec. The field of view was 512 by 512 pixels, and in-plane resolution was 3.6 µm/pixel. The pixel-basis analysis of appearance time was performed (see text). The image showed a continuity of appearance time in each vascular segment. An early appearance time (red) mostly represented an arterial flow, whereas a late appearance time (yellow to blue) resulted from venous flow. The color bar indicates the image acquisition time. Scale bar: 0.5 mm. (B) Segmentation of arterial and venous compartments. The arterial (red) and venous (blue) blood vessel areas were determined based on the spatial continuity of the appearance time observed along a longitudinal direction of the vessels (see text). Scale bar: 0.5 mm. (C) AVTT. The histogram represents the frequency distribution of appearance time observed in respective arterial (red) and venous (blue) areas. A total of 37,638 and 97,272 pixels were counted for arterial and venous segments, respectively, in this representative animal. Mean appearance time was observed as 0.29±0.28 and 1.43±0.48 sec in the arterial and venous segments, respectively, and thus, AVTT was 1.14 sec (a width between dashed vertical lines).

Figure 4. A comparison of mean AVTT.

(A) The normalized histogram of the appearance time in all exposed rats (N = 4). The consistent distribution pattern of the appearance time was observed for both the arterial (red) and venous (blue) segments. (B) Population data on mean AVTT. Mean of AVTT was 1.2±0.3 and 1.9±0.6 sec in non-exposed (N = 9) and radiation exposed animals (N = 4), respectively. There were statistically significant differences between the two groups (p<0.05).

Figure 5. A comparison of capillary blood speeds.

(A) Mean capillary RBC speed. The mean capillary RBC speed measured in the present study for the exposed animals (N = 4) was 40% lower than that for the non-exposed animals measured in a previous study (modified from [30]). (B) Inverse of the mean AVTT. The inverse of the mean AVTT of the exposed animals was 34% lower than that of the non-exposed animals. The ratio of exposed to non-exposed animal results was similar for both data sets.