Correlation of Visually Evoked Functional and Blood Flow Changes in the Rat Retina Measured With a Combined OCT+ERG System

Abstract

Purpose: To correlate visually evoked functional and blood flow changes in the rat retina measured simultaneously with a combined optical coherence tomography and electroretinography system (OCT+ERG).

Methods: Male Brown Norway (n = 6) rats were dark adapted and anesthetized with ketamine/xylazine. Visually evoked changes in the retinal blood flow (RBF) and functional response were measured simultaneously with an OCT+ERG system with 3-μm axial resolution in retinal tissue and 47-kHz image acquisition rate. Both single flash (10 and 200 ms) and flicker (10 Hz, 20% duty cycle, 1- and 2-second duration) stimuli were projected onto the retina with a custom visual stimulator, integrated into the OCT imaging probe. Total axial RBF was calculated from circular Doppler OCT scans by integrating over the arterial and venal flow.

Results: Temporary increase in the RBF was observed with the 10- and 200-ms continuous stimuli (∼1% and ∼4% maximum RBF change, respectively) and the 10-Hz flicker stimuli (∼8% for 1-second duration and ∼10% for 2-second duration). Doubling the flicker stimulus duration resulted in ∼25% increase in the RBF peak magnitude with no significant change in the peak latency. Single flash (200 ms) and flicker (10 Hz, 1 second) stimuli of the same illumination intensity and photon flux resulted in ∼2× larger peak RBF magnitude and ∼25% larger RBF peak latency for the flicker stimulus.

Conclusions: Short, single flash and flicker stimuli evoked measureable RBF changes with larger RBF magnitude and peak latency observed for the flicker stimuli.

Visual stimulation of the retina evokes neurovascular activity such as neuronal activation, which results in higher cellular metabolic demand and subsequent temporary vasodilation and blood flow increase, termed functional hyperemia. Potentially blinding ocular diseases, such as glaucoma and diabetic retinopathy can cause both temporary and permanent changes in the structure, blood perfusion, and functional response of the retina.^1–3 Retinal blood flow (RBF) changes in response to visible light stimulation have been measured and studied in the past in both healthy and diseased retinas with a variety of optical methods such as the blue field entoptic method,⁴ scanning laser Doppler flowmetry,⁵ fluorescence-based angiography,⁶ and fluorescence microspheres.⁷ Because these methods have limited temporal resolution, they are not able to provide information about the rapid communication between visually stimulated retinal neurons and neighboring blood vessels. Two-photo microscopy⁸ and functional magnetic resonance (fMRI-BOLD)⁹ offer both high temporal and high spatial resolution and have been used in the past to image and quantify RBF changes in response functional stimuli. Because both methods rely on tracking the motion of red blood cells in individual blood vessels, measurements are limited to very small regions of the examined retinal tissue. 

Recently, optical coherence tomography based angiography (OCTA) was used to investigate detectable changes in the retinal microvasculature density in response to visual stimulation in rodents and human subjects.^10–14 Doppler OCT utilizes phase information to detect and quantify blood flow in retinal blood vessels and provides an alternative method for investigation of neurovascular coupling in vivo. In general, Doppler OCT averages multiple cross-sectional images (B-scans), acquired from the same location to measure blood flow rates in biological tissue. Therefore, by utilizing broad-bandwidth light sources and high speed cameras or tunable lasers, both high spatial and temporal resolution Doppler OCT imaging data can be acquired, which enables the investigation of the rapid vascular response of the retina to visual stimulation. Total axial RBF in the retina can be calculated from circular Doppler OCT scans centered at the optic nerve head (ONH) by averaging over the blood flow in all retinal blood vessels in the B-scan. All of the previous Doppler OCT or OCTA studies mentioned above used very long duration (10 seconds to 2.5 minutes) of either single flash or flicker stimuli. It would be of interest to determine what is the shortest possible stimulus duration that can evoke changes in the RBF that are measurable with Doppler OCT, to correlate the RBF changes with changes in the retinal neuronal activity, as well as to investigate any difference in the retinal response to flicker and single flash stimuli of the same intensity and color. 

Electroretinography (ERG) records the electrical activity of visually stimulated retinas and therefore provides a way to measure the visually evoked cellular response from different types of retinal cells with millisecond scale time resolution. By combining information obtained from ERG recordings and Doppler OCT measurements, a more complete model of the neurovascular coupling in the visually stimulated retina can be generated. Previous studies have shown that flicker ERG traces and the first two harmonic components of ERG recordings are indirectly correlated to changes in the RBF in response to visual stimuli.^15,16 However, the ERG and Doppler OCT data in those studies were not recorded simultaneously and the stimuli durations ranged from 10 seconds to 2.5 minutes. 

In this study, we used a combined Doppler OCT and ERG system (DOCT+ERG) to measure simultaneously with high spatial and temporal resolution visually evoked changes in the retinal neuronal activity and RBF, as well as to determine the shortest single flash stimuli that would induce a measurable change in the RBF. Furthermore, in this study we investigate the magnitude, latency, and recovery rate of the RBF as function of the stimulus type, intensity, and duration. 

Materials and Methods

Animal Protocol

Eight-week-old, male, Brown Norway rats (n = 6, Harlan Laboratories, Inc., Indianapolis, IN, USA), weighing ∼250 g were used for this study. The animals were kept at a 12-hour light/dark cycle and dark adapted for at least 12 hours prior to the experimental sessions. The rats were anesthetized with ketamine/xylazine cocktail (0.2 mL/100 g body weight) that was delivered intraperitoneally. Subcutaneous injections of 5 mL sterile saline were administered immediately after the ketamine injection and approximately every 1 hour afterwards to keep the animal well hydrated. The rats were placed on a custom stereotactic stage to reduce head motion artefacts and allow for translational and rotational alignment of the imaged eye with respect to the DOCT+ERG system's imaging probe. During the experimental procedures, the animals were kept at 38°C with a thermal pad placed under belly (Kent Scientific, Torrington, CT, USA). One drop 0.5% proparacaine hydrochloride (topical anaesthetic; Alcaine, Alcon, Mississauga, ON, Canada) was applied to the imaged eye, followed by one drop of 0.5% tropicamide (pupillary dilator; Alcon). Artificial tears were applied every 5 minutes to keep the cornea well hydrated and optimize the impedance match between the cornea and the ERG corneal loop electrode. Metacam (2 mL/100 body weight diluted in sterile water) was administered for pain relief after completion of the experiments to help with the animal recovery. All experiments described here were approved by the University of Waterloo Animal Research Ethics Committee and adhered to the ARVO Statement for Use of Animals in Ophthalmic and Vision Research. 

Doppler OCT+ERG System

A research-grade, spectral domain OCT system, designed and built by our group for various imaging studies of the rodent retina,^17–19 was modified for use in this study (Fig. 1). Briefly, a broad bandwidth superluminescent diode (λ_c = 1020 nm, Δλ = 110 nm, Superlum, Carrigtohill, Co. Cork, Ireland) was used to achieve 3-μm axial resolution in retinal tissue and ensure that the Doppler OCT (DOCT) imaging beam does not visually stimulate the retina. The Doppler OCT retinal imaging probe, comprising three broadband NIR achromat doublet lenses (f₁ = 10 mm, f₂ = 60 mm, and f₃ = 30 mm; Edmund Optics, Barrington, NJ, USA) and a pair of galvanometric scanners (Cambridge Technologies, Bedford, MA, USA), was designed to deliver a collimated imaging beam of 1.5-mm diameter and 1.7-mW optical power to the rat cornea, thus achieving ∼5 μm lateral resolution in retinal tissue. A high-resolution spectrometer (P&P Optica, Waterloo, Canada) and a NIR line scan camera (1024-LDH2 92 KHz, Sensors Unlimited, Inc., Princeton, NJ, USA) were used at the detection end of the DOCT system. A commercial ERG system (Diagnosys LLC, Lowell, MA, USA) was interfaced with the DOCT system, and the data acquisition was synchronized to allow for simultaneous DOCT and ERG recordings. A new, custom-built visual stimulator that utilizes a white light LED was integrated into the DOCT retinal imaging probe. Light from this LED was focused at the pupil plane (Fig. 1, green line) of the rat eye to generate almost uniform, Maxwellian illumination of the retinal surface. The illumination intensity and the temporal pattern of the LED were controlled from the ERG system's console. 

Figure 1

 

Schematic of the combined OCT+ERG system. L1-L6, achromatic doublet lenses; TS, translation stage; M, reference mirror; FC, fiber coupler; DCP, dispersion compensation prisms; PC, polarization controller; CL, collimator lens; DG, dispersion grating; FFT, Fast Fourier transform.

DOCT and ERG Measurement Protocols

A volumetric (1000 × 1000 × 512) morphological image of the rat retina was acquired from the region around the ONH in each animal prior to conducting the visual stimulus tests (representative image shown in Fig. 2A). Subsequently, multiple DOCT cross-sectional images (4000 × 512) were acquired continuously from a circular pattern centered at the retinal ONH at the rate of 12 fps. The diameter of the circular DOCT scan was set to ∼0.8 mm (Fig. 2B) to allow for visualization of cross-sections of the retinal arteries and veins in the vicinity of the ONH and to avoid the necessity of phase unwrapping in the calculation of the axial RBF. The axial RBF velocity was calculated from the phase difference between adjacent A-scans in the circular cross-sectional DOCT images. A representative cross-sectional circular OCT scan with color-coded blood vessels is shown in Figure 2C. One hundred repeated OCT frames (total acquisition time of ∼8.5 seconds) were acquired to track the axial blood flow velocity over time in response to the stimuli. The DOCT data was acquired simultaneously with the ERG recordings. 

Figure 2

 

(A) Volumetric OCT image of the rat retina centered at the ONH. (B) Enface projection of the retinal OCT image. The white circle marks the DOCT scanning pattern with diameter ϕ ∼0.8 mm. The major retinal blood vessels are identified (V – vein and A – artery). (C) Circular cross-sectional OCT image of the retinal structure and blood vasculature.

The positive ERG electrode (a 4.5-mm diameter silver wire loop) was placed gently on the rat cornea to ensure clear aperture for the DOCT imaging and visual stimulus beams. The negative and reference needle ERG electrodes were placed under the skin behind the rat's ears. Single flash (10- and 200-ms duration, 1.14 log scotopic cd.s/m²) and flicker (10 Hz, 20% duty cycle, 1- and 2-second duration, 1.14 and 0.80 log scotopic cd.s/m²) visual stimuli were projected onto the retinal surface. The duration of each ERG recording was 8.5 seconds with 1-second prestimulus baseline. For each type of the visual stimulus (single flash or flicker) and stimulus settings (intensity and duration), five ERG recordings were acquired with 5-min dark adaptation period in between. At least 15-min dark adaptation was used between consecutive sets of recordings, acquired with different settings of the stimulus. 

Doppler OCT Data Analysis

A subpixel registration algorithm²⁰ was used to correct for any bulk motion between neighboring OCT B-scans prior to segmentation of the retinal blood vessels. The retinal blood vessels were manually selected, and the axial blood flow was calculated by integrating the blood velocity over the selected blood vessel area. Arterial and venal blood flow was calculated separately due to the different polarity of the respective phase changes. Total axial RBF was determined as an average of the magnitudes of the arterial and venal RBF. A moving window smoothing algorithm (Savitzky-Golay) was used to filter out oscillations in the temporal DOCT recordings due to pulsatile blood flow. Figure 3A shows a representative recording of the RBF measured from one retinal blood vessel over time (black line). The red line shows the filtered RBF recording after removal of the pulsatile oscillations. Figures 3B and 3C show the spatial distribution of the measured phase changes within the blood vessel's cross-section before (t = 0.8 seconds) and after (t = 3 seconds) application of the visual stimulus, respectively. Fractional changes in the RBF induced by the visual stimuli were calculated relative to the prestimulus (baseline) part of the recording. A Student's t-test was used to determine the significant changes in the RBF peak amplitude and latency for the different settings of the visual stimuli. 

Figure 3

 

(A) Total axial RBF change as a function of time for a 2-second flicker stimulus (gray area). The original RBF data (black line) shows pulsatile oscillations due to the animal heart rate. Filtered RBF data are shown in red. Spatial distribution of the DOCT signal within the cross-section of a retinal blood vessel at prior to (B) and post visual stimulation (C).

ERG Data Analysis

Analysis of the single flash and flicker ERG recordings followed the International Society for Clinical Electrophysiology of Vision (ISCEV) standards.²¹ For single flash ERG recordings, the amplitude and latency of the a-wave and b-wave were determined. For flicker ERG, the amplitude was calculated as averaged voltage differences between peaks to troughs, excluding first two peaks. 

Results

Figure 4 summarizes results from measurements conducted with the 10- and 200-ms single flash stimuli. Representative ERG traces are shown in Figure 4A with the gray and yellow marked areas corresponding to the duration of the visual stimuli. Figure 4B shows the total axial RBF, averaged over all recordings from all animals, as a function of time. For the same stimulus intensity, the 200-ms flash resulted in 5× larger RBF peak magnitude compared to the 10-ms flash, though no significant change in the latency of the RBF peak between the two stimulus durations was observed. Although the 10-ms single flash stimulus generated measureable changes in the RBF, no reproducible changes in the RBF were measured with 10-ms single flash stimuli with intensity <1.14 log scotopic cd.s/m², or with stimuli with 1.14 log scotopic cd.s/m² and duration shorter than 10 ms. Figure 4C presents normalized averaged total axial RBF for the two single flash stimuli, which shows that there are no significant differences in the stimulation and recovery rates of the RBF (positive and negative slopes of the RBF peak). Figure 4D shows statistical correlation between the changes in the RBF and the ERG b-wave magnitude for the 10- and 200-ms single flash stimulus durations. Although the RBF peak magnitude showed significant differences for the two flash durations (P = 0.023), the ERG a-wave and b-wave magnitudes showed no significant differences between the two flash durations (P = 0.256 and P = 0.056, respectively). 

Figure 4

 

Effect of single flash stimulus duration. (A) Representative ERG traces acquired with 10- and 200-ms single flash stimulus duration. (B) Total axial RBF as a function of time for the 10- and 200-ms single flash stimuli, averaged over all recordings from all animals. (C) Normalized RBF data showing differences in the recovery rate for the two stimulus durations. (D) Comparative statistics for the maximum RBF change and the ERG b-wave magnitude data for the two stimulus durations. The “*” denotes significant difference between the two groups of data and the data are presented as mean ± SD.

Figure 5 summarizes results from the tests with flicker stimuli of different duration. The gray and pink shaded areas correspond to the duration of the visual stimuli. Figure 5A shows representative ERG traces for 1- and 2-second long flicker stimuli. Figure 5B shows the total axial RBF, averaged over all recordings from all animals, as a function of time. Although the 1-second flicker stimulus results in ∼8% RBF peak change compared to ∼10% for the 2-second flicker stimulus, the difference in the peak RBF magnitudes are not statistically significant (P = 0.169, Fig. 5D). Statistical results also showed no significant differences in the ERG b-wave magnitude (P = 0.086). Also, there is no significant difference in the latency of the RBF peak for the 1- and 2-second flicker stimuli. Normalized RBF recordings for the 1- and 2-second flicker stimuli are shown in Figure 5C. Although there is no significant difference in the RBF rate of increase with the application of the visual stimulus, the RBF recovery rate is ∼3× faster for the 1-second stimulus compared to the 2-second flicker stimulus of the same intensity. 

Figure 5

 

Effect of flicker stimulus duration. (A) Representative ERG traces acquired with 1- and 2-second flicker stimulus duration. The pink and gray shaded areas mark the duration of the visual stimuli. (B) Total axial RBF as a function of time for the 1- and 2-second flicker stimuli, averaged over all recordings from all animals. (C) Normalized RBF data showing differences in the recovery rate for the two stimulus durations. (D) Comparative statistics for the maximum RBF change and the ERG b-wave magnitude data for the two flicker stimulus durations. The data are presented as mean ± SD.

Figure 6 summarizes results from the tests with 2-second long flicker stimuli of different intensities (1.14 and 0.80 log scotopic cd.s/m²). Figure 6A shows representative ERG traces for the flicker stimuli of different intensities. The total axial RBF, averaged over all recordings from all animals, as a function of time is shown in Figure 6B. The brighter flicker stimulus resulted in significantly higher RBF peak magnitude compared to the stimulus of lower intensity (P = 0.0471, Fig. 6D); however, there was no significant difference in the RBF peak latencies for the two stimuli. Statistics of the ERG data (Fig. 6D) shows that the flicker ERG magnitude is larger with higher flicker stimuli intensity (P = 0.020). Normalized RBF recordings for the 1- and 2-second flicker stimuli are shown in Figure 6C. Although there is no significant difference in the RBF rate of increase with the application of the visual stimulus, the RBF recovery rate is ∼50% faster for the low intensity stimulus. 

Figure 6

 

Effect of flicker stimulus intensity. (A) Representative ERG traces for the flicker stimuli with different intensities. (B) Total axial RBF change as a function of time, for the two stimulus intensities, averaged over all recordings from all animals. (C) Normalized RBF data showing differences in the recovery rate for the two stimulus intensities. (D) Comparative statistics for the maximum RBF change and the ERG b-wave magnitude data, presented as mean ± SD for the two flicker stimulus intensities. The “*” denotes significant difference between the two groups of data.

Figure 7 shows results from a comparative analysis between the 200-ms single flash and the 1-second flicker stimuli. Since in our studies we used 10-Hz flicker with 20% duty cycle, each cycle of the flicker stimulus corresponds to 20-ms long continuous flash. Therefore, the 1-second flicker has the same average photon energy as the 200-ms single flash. Figure 7B shows representative ERG traces for the 200-ms single flash and 1-second flicker stimuli. The total axial RBF, averaged over all recordings from all animals, as a function of time is shown in Figure 7B. The 200-ms single flash stimuli induced ∼2× smaller RBF peak magnitude change compared to the 1-second flicker stimulus (4.1 ± 2.4% vs. 8.0 ± 1.7%, P = 0.034). Furthermore, the single flash stimulus had ∼25% smaller latency (1.7 ± 0.2 seconds vs. 2.1 ± 0.1 seconds, P < 0.001). Figure 7C presents normalized RBF traces for the single flash and flicker stimuli. Although the single flash stimulus has ∼2× faster rate of the RBF increase, it also shows ∼2× lower recovery rate compared to the flicker stimulus. Statistical results for the RBF peak magnitude and latency for the two types of visual stimuli are presented in Figure 7D. These results show that regardless of the fact that the overall photon energy delivered to the retina by the 200-ms single flash and 1-second flicker stimuli is the same, differences in the RBF peak magnitude and latency for the different stimuli types are significant (marked with “*” and “†” in Fig. 7D, respectively). 

Figure 7

 

Comparison of the changes in the total axial RBF and the ERG b-wave magnitude resulting from 200-ms single flash and 1-second, 10 Hz, 20% duty cycle flicker stimuli of the same illumination intensity. (A) Representative ERG traces. The pink and gray shaded areas mark the duration of the visual stimuli. Original (B) and normalized (C) time recordings of the total axial RBF in response to the single flash and flicker stimuli. (D) Peak latency and amplitude statistics for the RBF data, where “*” and “†” denote significant differences in the RBF peak amplitude and latency, respectively, between data acquired with the single flash and flicker stimuli. The data are presented as mean ± SD.

Discussion

Results from our study on the flicker-induced RBF changes agree in general with results from similar studies conducted by other research groups with different imaging methods. Specifically, Kornfield et al.²² observed 11% increase in the superficial RBF following 2-second flicker stimulation by using functional MRI (BOLD), which compares well with the ∼10% increase we measured with DOCT for flicker stimulus of the same duration. Radhakrishnan et al.²³ reported ∼12% blood flow increase for a 10-second flicker stimulus using en-face Doppler OCT, while Werkmeister et al.²⁴ detected over 30% increase of blood flow with 60-second flicker stimulus using DOCT. Results from our study agree with the general trend established by all of these studies that longer duration of the flicker stimulus contributes to a larger peak magnitude of the RBF and that the time of RBF return to baseline is directly proportional to the flicker stimulus duration. One new and significant result from our study is that DOCT was proven able to measure reproducibly RBF changes in response to much shorter visual stimuli (10 ms for a single flash and 1 second for flicker), compared to all other previous studies that utilized continuous and flicker stimuli with durations ranging from 10 seconds to 2.5 minutes.^{10–14,22–24} 

Results from our study also showed that DOCT is able to measure reproducibly changes in the RBF in response to continuous visual stimuli with duration as short as 10 ms. However, as demonstrated by the results in Figure 7, single flash and flicker stimuli of the same intensity generate different responses of the RBF. Almost 2× larger RBF peak magnitude was measured with the flicker stimuli, which indicates that retinal neurons respond differently to the frequency content of the visual stimulus. Our results correlate well with results from other studies conducted with different imaging modalities, which also indicate that flicker stimuli have stronger effect on the neurovascular coupling and vasodilation^5,25 compared to continuous visual stimuli. 

The DOCT imaging protocol used in our study was designed to emphasize axial blood flow measurement from retinal blood vessels located at the retinal nerve fiber layer. By changing the image acquisition protocol, it is possible to measure changes in the total RBF, as well as the capillary flux in the inner retina in response to visual stimulation by counting the number of red blood cells passing through the repeated cross-section area.^26,27 

In our study, we utilized ERG to investigate physiological responses of the retina to visual stimuli and to correlate those changes to the RBF changes. Since the rat retina is rod dominated and in our study we used scotopic ERG recordings, the visually evoked changes in the RBF are most likely associated with metabolic changes in the rods. Future studies that utilize different types of visual stimuli are necessary in order to identify the contribution of the cones and rods to the visually evoked RBF changes. Functional OCT has been proven able to image intrinsic optical changes (IOS) in the retina in response to visual stimuli.^28–32 Since the fast IOS changes occur on a millisecond scale, while changes in the RBF occur on the scale of seconds, the DOCT protocol we utilized for our current study was not suitable for simultaneous recording of both IOS and RBF changes with the same OCT system. Future development of the OCT technology and image acquisition protocols could allow for simultaneous recording of stimulus-induced IOS and RBF changes in the living retina. 

In conclusion, we have developed a combined OCT+ERG system to allow for simultaneous measurement of the physiological and blood flow changes in the rat retina induced by visual stimuli. We showed that both single flash and flicker stimuli of short duration induce measurable changes in the RBF and demonstrated that DOCT is capable of measuring reproducibly RBF changes from continuous single flash stimuli as short as 10 ms. We also showed a correlation between changes in the RBF and the retina functional response to visual stimuli of different type, intensity, and duration. The combined OCT+ERG system, the Doppler OCT scanning protocol, and data analysis described here can find numerous applications in studies of animal models of retinal diseases such as glaucoma and diabetic retinopathy, where the visually evoked changes in the RBF maybe affected by the disease. 

Acknowledgments

The authors thank Nancy Gibson for assistance with the animal care and related procedures, and H. van der Heide from the UW Science shop for assistance with the design of the custom stereotactic holder. 

Supported in part by the Natural Sciences and Engineering Research Council (NSERC) of Canada Discovery Grant 312037 and the Canadian Institutes for Health Research (CHRP) Grant 47219. 

Disclosure: B. Tan, None; E. Mason, None; B. MacLellan, None; K.K. Bizheva, None