Effect of A-scan rate and interscan interval on optical coherence angiography

Ting-Hao Chen; Yi-Chun Wu; Ting-Yen Tsai; Chuan-Bor Chueh; Bo-Huei Huang; Yin-Peng Huang; Meng-Tsan Tsai; Meng-Tsan Tsai; Yoshiaki Yasuno; Hsiang-Chieh Lee; Hsiang-Chieh Lee; Hsiang-Chieh Lee

doi:10.1364/BOE.409636

1. Introduction

Optical coherence tomography (OCT) is a real-time and noninvasive optical imaging technique that involves using low-coherence interferometry to provide volumetric imaging of architectural tissue information [1]. OCT has feasibility as a high-resolution imaging modality in various clinical applications such as those related to cardiology, dermatology, ophthalmology, and gastroenterology diseases [2]. OCT angiography (OCTA) is a functional extension of this technique that allows volumetric imaging of the subsurface microvasculature without the requirement of exogenous contrast agents. OCTA can be used for diagnostic assessments and the evaluation of new treatments along with the ongoing development of high-speed OCT imaging technology. To provide microvascular information of biological tissue, scientists have measured changes in the phase [3–7], amplitude (or intensity) [8–12], or complex values [13–17] of OCT signals resulting from the dynamic scattering of red blood cells in functional vessels relative to the static scattering of surrounding tissues. This approach can provide depth-resolved microangiograms with a capillary-level resolution, delineating three-dimensional (3D) microvascular networks in biological tissue.

Recently, OCTA has been implemented in human and animal models. In clinical medicine, using either commercial or in-house-developed OCT imaging systems, studies have demonstrated the applicability of OCTA in ophthalmology and dermatology for providing new insight into the pathophysiology, including the deceases of human eyes and skin [18–23]. In particular, OCTA can be performed in combination with standard OCT examination of the human retina, eliminating the need for a separate procedure. Repetitive OCTA can be performed as well, unlike conventional fluorescence angiography with ICG. However, studies have demonstrated the applicability of OCTA in various animal studies, either focusing on skin imaging or brain microvascular imaging. For example, Tsai et al. demonstrated the use of OCTA as a monitoring tool in transdermal drug-delivery systems with a mouse ear skin model [24]. Li et al. also employed an electrically tunable lens in a 1.3-µm spectral-domain OCT (SD-OCT) system to perform multi-focal-plane cerebral blood flow and optical microangiography imaging in mouse cortexes [25].

It is challenging to provide quantitative blood flow information with OCTA, even though OCTA can provide better visualization of the finer vascular network than Doppler OCT can. Studies have demonstrated the results of attempting to provide relative blood flow information with the variable interscan time analysis (VISTA) algorithm [26]. In this method, the dynamic range of OCTA can be adjusted with different interscan times of the repeated B-scans acquired at the same location. For example, OCTA with longer interscan time is more sensitive in detecting microvasculature with a lower blood flow rate relative to OCTA, with a shorter interscan time. Subsequently, by taking advantage of the aforementioned VISTA algorithm, Ploner et al. [27] provided OCTA image encoding relative blood flow information by employing OCTA images computed with two interscan times. The VISTA algorithm has successfully enhanced the dynamic range of OCTA for imaging the retinal vasculature of human eyes.

However, in these OCTA studies using the VISTA algorithm, researchers did not examine whether the effect of OCTA could be altered as a function of the light source A-scan rate and the interscan time for OCTA image reconstruction. In this study, therefore, we used a self-developed 1-µm OCT system to investigate the changes of OCTA contrast with different A-scan rates (100 kHz and 200 kHz) and interscan times between the repetitive OCT images used to reconstruct OCTA images. This study focused on comparing the OCTA images acquired using different A-scan rates but with the same interscan time. To avoid introducing bias into the contrast analysis of the OCTA results acquired with different A-scan rates, we adjusted the reference arm power to achieve almost the same detection sensitivity between the 100-kHz and 200-kHz OCT imaging modalities. In addition, parameters including the vessel area, total vessel length, and junction density were computed from the OCTA images acquired from five mouse ear skin models based on AngioTool (version 0.6a, National Institute of Health) [28].

2. Method

2.1 Swept-source OCT imaging system

Figure 1 presents a schematic of the 1-µm swept-source OCT (SS-OCT) system used in this study. Two separate light sources with a central wavelength of 1060 nm and a sweep rate of 100 kHz (Axsun Technologies) and 200 kHz (Axsun Technologies) were used alternatively in the SS-OCT system to provide OCT imaging with different A-scan rates. Ninety percent of the laser output was connected to the OCT interferometer; the remaining 10% of power was connected to a fiber Bragg grating (O\E Land Inc.) to provide a stable A-scan trigger signal. In the sample arm, the light was collimated with an achromatic lens (AC080-016-C, Thorlabs) and directed to a pair of galvanometer scanning mirrors (GVS102, Thorlabs) to provide beam scanning over the sample surface. A single achromatic lens (AC254-060, Thorlabs) with a focal length of 60 mm was used in the sample arm, providing a lateral resolution of approximately 13.5 µm (full-width at half-maximum, FWHM) based on knife-edge measurements. The same optical components were used in the reference arm to minimize the dispersion mismatch between the sample arm and reference arm. The backscattered light from the sample arm and the reference arm interfered at the 50:50 coupler and was subsequently detected by a dual-balanced photodetector (PDB435C, Thorlabs) with a bandwidth of 350 MHz. The detected signals were filtered by the low-pass filter (BLP150+, Mini-circuits) with a 155-MHz 3-dB cutoff frequency to reduce high-frequency noise. A high-speed digitizer (ATS9373, AlazarTech) with a 12-bit resolution was used to convert the detected interference signal into a digitized signal using the sampling clock provided from the swept source. A custom-developed graphic user interface written in C++ was used to integrate data acquisition and optomechanical control, including galvo-scanning, real-time image preview, and data saving, in a synchronized manner.

Fig. 1. Schematic of the SS-OCT system. For the OCT imaging, the swept source with a central wavelength of 1.06 µm and an A-scan rate of either 100 kHz or 200 kHz was used. Col: collimator; RM: reference mirror; DBPD: dual-balanced photodiode; PD: photodetector; FBG: fiberoptic Bragg grating; Galvo: galvanometer scanner; DAQ: data acquisition card.

Download Full Size | PDF

The specifications of the 100-kHz and 200-kHz OCT imaging configurations are shown in Table 1. The axial resolution (FWHM) values were 10.4 µm and 11.4 µm for the 100-kHz and 200-kHz OCT imaging modalities, respectively. It could be expected, based on the A-scan rate and the illumination power over the sample surface, that the theoretical detection sensitivity for 100-kHz OCT imaging would be higher than the 200-kHz OCT imaging. Thus, we lowered the reference arm power when performing the 100-kHz OCT imaging, leading to a comparable detection sensitivity between the 100-kHz and 200-kHz OCT imaging processes. This allowed us to accurately evaluate how the A-scan rate and interscan time affects the OCTA contrast between different settings. As listed in Table 1, the measured detection sensitivity was 94 dB at a depth of 490 µm and 93.5 dB at a depth of 627 µm for the 100-kHz and 200-kHz OCT imaging modalities, respectively. Figure 2 presents the measured roll-off of the detection sensitivity for the developed SS-OCT system with A-scan rates of 100 kHz or 200 kHz; the figure shows roll-off slopes of 2.78 dB/mm and 1.58 dB/mm in air, respectively. Also, the depth of interest ranging from 500 µm to 1100 µm is still in the region of interest (ROI), exhibiting negligible roll-off for both cases in Fig. 2.

Fig. 2. Roll-off of the detection sensitivity for the SS-OCT system using the swept source with A-scan rates of (a) 100 kHz and (b) 200 kHz, respectively.

Download Full Size | PDF

Table 1. Specifications of the 1.06-µm SS-OCT system employed in this study.

View Table | View all tables in this article

2.2 OCTA image processing

Figure 3 displays the flowchart for the OCTA image processing with the correlation mapping algorithm used in this study. To perform OCTA, five repetitive B-scan frames were acquired at each location. Thus, one volumetric or 3D OCT raw data set contains 1000 × 500 × 5 A-scans, where each B-scan frame consists of 1000 A-scans. After OCT images had been reconstructed from the raw data set, the OCT images were used to compute the OCTA images following the algorithm presented in Fig. 3. First, the correlation coefficients for each B-scan pixel were calculated based on Eq. (1) [11] to generate one correlation coefficient matrix, where I_A and I_B are the OCT signal intensity values in a linear scale format for the two adjacent frames selected from the five repeated frames.

(1)$${I_{cm}}(x,z) = \frac{{\sum\nolimits_{j = 1}^M {\sum\nolimits_{k = 1}^N {\{{[{{I_A}(x + k,z + j) - \overline {{I_A}(x,z)} } ][{{I_B}(x + k,z + j) - \overline {{I_B}(x,z)} } ]} \}} } }}{{\sqrt {\sum\nolimits_{j = 1}^M {\sum\nolimits_{k = 1}^N {{{[{{I_A}(x + k,z + j) - \overline {{I_A}(x,z)} } ]}^2}} } } \sqrt {\sum\nolimits_{j = 1}^M {\sum\nolimits_{k = 1}^N {{{[{{I_B}(x + k,z + j) - \overline {{I_B}(x,z)} } ]}^2}} } } }}$$

In Eq. (1), M and N represent the grid size or window size, and $\bar{I}$ is the mean intensity for each grid. In our study, M × N was chosen to be 7 × 7 pixels corresponding to a matrix size of approximately 35 µm × 27 µm and 37.8 µm × 27 µm for the 100-kHz and 200-kHz OCT imaging, respectively, which achieved a reasonable balance between the imaging resolution and the OCTA image noise. The interval between the two frames used for the correlation coefficient calculation was not limited to 1 (i.e., adjacent frames), yielding a different interscan time for the OCTA contrast. For example, an interval of 1 frame (i.e., frame 1↔frame 2, frame 2↔ frame 3, etc.) used to compute the OCTA images from the five repetitive B-scan frames corresponds to OCTA interscan times of 12.5 ms and 6.25 ms for the data sets acquired at rates of 100 kHz and 200 kHz, respectively. However, an interval of two frames allows an OCTA interscan time of 12.5 ms for 200-kHz OCT imaging.

Fig. 3. Flowchart of the OCTA processing with the correlation mapping algorithm. By selecting the designated interval between the OCT images of the repetitive frames per location acquired using either a 100-kHz or a 200-kHz swept source, OCTA information with different interscan times can be computed afterward.

Download Full Size | PDF

The calculated correlation coefficient matrices from the five repetitive frames based on the designated frame interval were averaged to improve the signal-to-noise ratio of the cross-sectional OCTA image. For example, for a frame interval of 1, the system has four correlation coefficient matrices to be averaged as one matrix. The averaged correlation coefficient matrix was thresholded based on the steps described in later passages. First, pixels with absolute correlation coefficients ranging from 0 to 0.6 were regarded to denote a potential location of blood flow. By contrast, any value outside this range was considered as the static background. Also, the noise floor of the averaged OCT images from the five repetitive frames was used to generate a mask. Finally, we applied this noise threshold mask to an averaged correlation coefficient matrix with consideration of the aforementioned threshold value to generate the final OCTA images.

Because five repetitive frames were acquired per location, the OCTA interscan time could be adjusted or varied by changing the frame interval used to compute the OCTA images, enabling the differentiation of microvasculature with different blood flow rates, similar to what can be achieved using the VISTA algorithm employed by Choi and Ploner et al. [26,27]. In this study, the duty cycle for scanning waveform along the fast axis for the galvanometer was 80%, yielding an OCTA interscan time of 12.5 ms for 200-kHz OCT imaging. All data processing, including the OCT and OCTA image reconstruction, was performed using MATLAB 2019b (MathWorks).

2.3 Animal imaging procedures

In this preliminary study, five small white ICR (Institute of Cancer Research) mice (BioLASCO, Taiwan) aged 5 weeks were used. The experimental procedure was approved (Protocol number: CGU105-062) by the Institutional Animal Care and Use Committee at Chang Gung University. Before the image session, each mouse was anesthetized with isoflurane, and the hair over the ear skin of mice was removed using hair removal cream. Mouse ears were temporarily fixed on a customized imaging holder using 3D printing technology during the OCT/OCTA session. For each mouse, 200-kHz OCT imaging was performed immediately after the 100-kHz OCT imaging by switching the 100-kHz swept source with the 200-kHz swept source after data saving had been completed. This approach ensures the imaging area is well registered between the 100-kHz and 200-kHz OCT imaging for each mouse. After both 100-kHz and-200 kHz OCT imaging sessions had been completed, the mouse was positioned in a separate chamber for recovery. A heat lamp was used to maintain mouse body temperature during the imaging session and the recovery stage. Subsequently, the same imaging protocol was performed for another mouse, until all five mice had been imaged.

2.4 Quantitative analysis of OCTA

AngioTool is a free, open-source software program that can be used to quantify blood vessel networks in microvascular images [28]. AngioTool provides some morphological and spatial parameters, including vessel area, the number of blood vessels, total vessel length, junction density, and the lacunarity. Studies have demonstrated the quantification of blood vessel networks with AngioTool in different animal studies such as in mouse brains, ratina, and human retina [29–33]. In this study, AngioTool allows the analysis of parameters including vessel area, total vessel length, and junction density for en face OCTA images, which are defined respectively as the area of the segmented vessels, the sum of Euclidean distances between the pixels of all the vessels, and the number of junctions per unit area. The parameters of the vessel diameter and vessel intensity in the AngioTool could be varied to dynamically update the vessel outline or detect the vessels at different spatial aspects. AngioTool can detect small blood vessels by applying smaller diameter parameter settings and eliminate the nonconnective pixels by removing the small particles and filling the holes. We set the vessel diameter parameters to 2 and 3, respectively, as the quantitative criteria for 100-kHz and 200-kHz OCT imaging of the ear skin from five mice.

3. Results

3.1 OCTA

Figure 4(a) presents a mouse ear skin photograph, where the ROI for OCT/OCTA is marked with a red box. The field of view (FOV) for the selected ROI is 3.85${\; } \times $ 3.85 mm². Figures 4(b-d) illustrate the maximum projected en face OCTA image over the ROI reconstructed with the OCT images collected at a frame rate and A-scan rate of 80 Hz (interscan time of 12.5 ms) and 100 kHz [Fig. 4(b)], 160 Hz (interscan time of 6.25 ms) and 200 kHz [Fig. 4(c)], and 80 Hz (interscan time of 12.5 ms) and 200 kHz [Fig. 4(d)], respectively. A frame rate of 80 Hz corresponded to an OCTA interscan time of 12.5 ms. The number of A-scans per B-scan frame was kept at 1,000. More detailed microvasculature information can be observed in Fig. 4(b) [red arrows, Fig. 4(b)] relative to Fig. 4(d), even though Figs. 4(b, d) show the same OCTA interscan time but acquired with a different A-scan rate (100 kHz vs. 200 kHz). In addition, when a comparison was made between Figs. 4(c) and 4(d), it was revealed that the longer OCTA interscan time allowed the detection of microvasculature with a relatively slower blood flow [green arrows, Fig. 4(d)] [26,27,34] at the cost of being more vulnerable to the motion artifacts due to breathing. By computing the OCTA images with different OCTA interscan times based on the VISTA algorithm, as described previously [27], it provides relative blood flow information on the OCTA images, as depicted in Fig. 4(e), based on the OCTA images, as demonstrated in Figs. 4(c, d). The vessels with a larger diameter exhibited relatively faster blood flow than did those with smaller vessel diameters.

Fig. 4. En face OCTA images of the mouse ear skin (mouse B) based on the maximum projection of the OCTA signals. (a) The photograph of the mouse ear skin where the imaged site for OCT/OCTA are marked with a red box. (b-d) OCTA images over the marked region in (a), which were reconstructed with an OCTA interscan time and OCT A-scan rate of 12.5 ms and 100 kHz, 6.25 ms and 200 kHz, as well as 12.5 ms and 200 kHz, respectively. (e) Reconstructed OCTA images with relative blood flow information obtained using the VISTA algorithm (c, d). Scale bars: 0.5 mm.

Download Full Size | PDF

3.2 Comparison of OCTA contrast

Figure 5 presents a comparison of en face OCTA images of mouse ear skin (mouse D) reconstructed by using the OCT images acquired with either an SS-OCT of 100 kHz or 200 kHz and the OCTA algorithm with the designated interscan time. The en face OCTA images were cropped to 2.31 × 2.31 mm² to avoid motion artifacts, such as those due to the breathing. These artifacts were represented as straight lines that usually occurred in the peripheral area of the original FOV. Thus, the OCTA images were cropped to avoid confounding the subsequent quantitative analysis of the OCTA images among different parameter settings. Figures 5(a–c) display raw en face OCTA images reconstructed with an OCTA interscan time and OCT A-scan rate of 12.5 ms and 100 kHz, 6.25 ms and 200 kHz, and 12.5 ms and 200 kHz, respectively. Figures 5(d–f) and Figs. 5(g–i) depict the analyzed results of Figs. 5(a–c) using AngioTool with the vessel diameters of 2 and 3, respectively. In Figs. 5(d–i), the contour and skeleton of the vessel identified for the analysis are denoted in yellow and red, respectively. Increasing the parameter value of the vessel diameter revealed that only a microvasculature with a large vessel diameter would be identified and included for the vessel contour and skeleton labeling using AngioTool. Hence, more skeleton features marked in red in Figs. 5(d–f) could be observed relative to Figs. 5(g–i).

Fig. 5. Comparison of the en face OCTA images of the mouse ear skin (mouse D) between different interscan times in terms of the 100-kHz and 200-kHz swept sources. Figures 5(a–c) illustrate raw en face OCTA images with an image size of approximately 2.31 × 2.31 mm² reconstructed with an OCTA interscan time and OCT A-scan rate of 12.5 ms and 100 kHz, 6.25 ms and 200 kHz, and 12.5 ms and 200 kHz, respectively. Figures 5(d–f) and (g–i) present the analyzed results of (a–c) obtained using AngioTool with vessel diameters of 2 and 3, respectively. After the AngioTool analysis algorithm was employed, the contour and skeleton of the microvasculature identified were marked in yellow and red, respectively. DP: diameter parameter Scale bars: 0.5 mm.

Download Full Size | PDF

Table 2 summarizes the quantitative analysis for the en face OCTA images from five mice obtained using the AngioTool. As shown in Table 2, for the aforementioned three parameters, OCTA images with a longer interscan time had a larger value than those with a shorter interscan time. The values for the OCTA images computed with an OCTA interscan time of 6.25 ms and an A-scan rate of 200 kHz were always smaller than those images with an OCTA interscan time of 12.5 ms and A-scan rate of 200 kHz and an OCTA interscan time of 12.5 ms and A-scan rate of 100 kHz. Also, with the same OCTA interscan time, the values of the three parameters for the OCTA images acquired at the 100 kHz were larger than those acquired at 200 kHz. The last row of Table 2 shows the P values from the one-way Kruskal–Wallis analysis. In this test, the system had three groups with interscan times and A-scan rates of 12.5 ms and 100 kHz, 6.25 ms and 200 kHz, and 12.5 ms and 200 kHz, respectively, and each group had a sample of five mice. The P values were calculated using MATLAB 2019b (MathWorks) for individual parameters, including vessel area, total vessel length, and junction density. The final results indicate a rejection of the null hypothesis and an acceptance of the alternative hypothesis, with significance at the 5% significance level. The null hypothesis was that the medians of three groups were the same, and the alternative hypothesis was that at least one population median of one group was not equal to the population median of at least one other group.

Table 2. Quantitative analysis results for the OCTA images of the ear skin of five mice using AngioTool.

View Table | View all tables in this article

Because three groups do not come from the overall same group under the 5% significance level, we again conducted a Wilcoxon signed-rank test to determine whether any two groups from the original three groups were statistically different. Table 3 shows that all P values associated with any two groups are under the 5% significance level in terms of vessel parameters. The P values were calculated using MATLAB 2019b (MathWorks). Thus, the results of Table 3 indicate again that the test could reject the null hypothesis and accept the alternative hypothesis at the 5% significance level, where the null hypothesis is that the difference between the two groups arises from a distribution with a zero median. The alternative hypothesis is that the difference between the two groups originates from a distribution with a nonzero median.

Table 3. P values from the statistical analysis using the Wilcoxon signed-rank test between two groups with different frame rates or A-scan rates.

View Table | View all tables in this article

Besides, the number of OCTA images averaged could affect its sensitivity to microvascular imaging. Therefore, we have computed the OCTA images for the 100-kHz OCT imaging with an interscan time of 12.5 ms by only using the first four of the five repetitive B-scan frames, i.e., only 3 correlation coefficient matrices were averaged to compute the OCTA image (Fig. 3). Note the same threshold value used to generate the noise mask for the OCTA images with 3 OCTA images averaged was used for the OCTA images with 4 OCTA images averaged for the 100-kHz OCT imaging. In order to quantitatively compare the changes of the OCTA images by reducing the number of OCTA images averaged for the 100-kHz OCT imaging, we also utilized AngioTool to assess the microvascular density exhibited in the OCTA images. Table 4 presents the comparison of the thresholded OCTA images for 100-kHz OCT imaging computed with 3 and 4 OCTA images averaged. As shown in Tables 4, a slight increase in most of the microvascular parameters was observed in the thresholded OCTA images computed with 3 OCTA images averaged than that with the 4 OCTA images.

Table 4. Quantitative analysis results using AngioTool for the OCTA images of the ear skin of five mice with interscan time 12.5 ms and A-scan rate 100 kHz in terms of three and four OCTA frames averaged.

View Table | View all tables in this article

Table 5 shows the quantitative comparison of microvasculature in terms of vessel area, total vessel length, and junction density between the en face OCTA images acquired at A-scan rates of 100 kHz and 200 kHz but with the same OCTA interscan time (12.5 ms), as listed in Table 2. To perform an objective comparison of the three aforementioned microvasculature parameters, we introduced the metric η to perform quantitative assessments of any of the three microvascular parameters between the analyzed en face OCTA images acquired with different A-scan rates. The metric is defined in the following equation.

(2)$$\mathrm{\eta } = \frac{{\textrm{Q}{\textrm{A}_{\textrm{100 kHz}}} - \textrm{Q}{\textrm{A}_{\textrm{200 kHz}}}}}{{\textrm{(Q}{\textrm{A}_{\textrm{100 kHz}}} + \textrm{Q}{\textrm{A}_{\textrm{200 kHz}}}\textrm{)/2}}}$$

In three microvascular parameters, the metric η is always larger than zero, suggesting that finer microvasculature can be detected in the en face OCTA images acquired at a lower A-scan rate. In addition, when we compared η values between the analyzed en face OCTA images with vessel diameters of 2 and 3, the η value was smaller for the later ones in terms of vessel area, total vessel length, and junction density. This suggests that for larger size blood vessels, using a 100-kHz light sources did not outperform the 200-kHz light source significantly.

Table 5. Quantitative comparison of microvasculatures between the en face OCTA images acquired at A-scan rates of 100 kHz and 200 kHz but with the same OCTA interscan time of the ear skin from five mice.

View Table | View all tables in this article

Finally, as mentioned previously, the en face OCTA images were cropped to exclude regions with strong motion artifacts for the subsequent quantitative microvasculature analysis with AngioTool. The selection of the ROI of the en face OCTA images was based on a previously discussed rationale. Figure 6 shows an example of the en face OCTA image where motion artifacts could be observed at the lower-left part of the FOV. We performed the same quantitative analysis on the cropped en face OCTA images from three randomly selected ROIs, as marked by red, purple, and green boxes in Fig. 6. As summarized in Table 6, the metric η was always larger than zero for the three microvascular parameters over the three ROIs. Also, η was smaller for the analyzed en face OCTA image with a vessel diameter of 3 than that with a vessel diameter of 2.

Fig. 6. Example en face OCTA image of the mouse ear skin (mouse C) where three ROIs were randomly selected as marked by (a) red, (b) purple, and (c) green boxes for the subsequent quantitative analysis of the microvasculature. Motion artifacts in images due to breathing can be observed at the lower-left corner of Figs. 6(a–c). Scale bar: 0.5 mm.

Download Full Size | PDF

4. Discussion

Of various OCTA algorithms, the intensity signal–based OCTA image [35,36] could be used to image vasculature with slow blood flow and an increased dynamic range. This could be achieved using the VISTA algorithm [26,27,37] based on the BM mode scanning mechanism [5]. Another approach is to change the scanning mechanism using a bidirectional scanning pattern without sacrificing the total scan duration [34]. In related studies, the OCTA algorithm has been used to calculate the decorrelation value, which ranges from 0 to 1, where 0 and 1 respectively indicate the lowest and highest dissimilarity [12,38]. By contrast, our system with sweep sources of either 200 kHz or 100 kHz can also reveal vasculature variations between different scan intervals, and the dynamic range could also be increased using the BM mode through the correlation mapping algorithm. According to the OCTA algorithm used in this study, as depicted in Fig. 3, the correlation coefficient with threshold and absolute values thereof ranged from 0 to 0.6. Also, the meaning of the value in terms of dissimilarity was opposite to that in the decorrelation method [11].

According to some researchers [34,39], the decorrelation value of OCTA can be defined as 1 minus the normalized autocorrelation function of the light field, as presented in Eq. (3),

(3)$${I_{de}} = 1 - \exp \textrm{(} - \frac{{|\tau |}}{{{\tau _c}}}\textrm{) }$$

where ${\tau _c}$ is correlation time and ${\tau _{}}$ is time difference corresponding to the frame interval or interscan time. In addition, the correlation coefficient of each pixel, as specified in Eq. (1), could be mapped into a normalized autocovariance function of the light field, as expressed in Eq. (4) [similar to Eq. (3)].

(4)$${I_{cm}} = \exp \textrm{(} - \frac{{2\tau }}{{{\tau _c}}}\textrm{)}$$

(5)$${\tau _c}\textrm{ }\cdot \textrm{ }({v_{flow}} + {v_{bulk}} + {v_{browian}}) = \xi $$

Given that ${\tau _c}$ has a linear relationship with motion (as described in [34]), inclusive of flow rate ${v_{flow}}$, bulk motion ${v_{bulk}}$, and Brownian motion ${v_{browian}}$ [as formulated in Eq. (5)], where $\xi $ is a constant, we can derive Eq. (6) by combining Eq. (3) and Eq. (5).

(6)$${I_{cm}} = \exp \textrm{[} - \frac{{\tau \textrm{ }\cdot \textrm{ }({v_{flow}} + {v_{bulk}} + {v_{browian}})}}{\xi }]$$

Although the interscan time τ was different between 100 kHz and 200 kHz OCT imaging, a longer interscan time (12.5 ms) provides more detailed blood vessel information than does a shorter interscan time (6.25 ms). This can be explained with reasoning similar to that in [34]. When interscan time τ increases, at the same blood flow rate, the correlation coefficient I_cm, as formalized in Eq. (6), decreases, making this value smaller than the threshold value. Equation (6) indicates that when interscan time increases (i.e., longer interscan time), a lower blood flow rate can be detected, thereby improving the slowest detectable flow. Similarly, Choi et al. [26] and Braaf et al. [40] also indicated a longer interscan time could improve the slowest detectable flow to facilitate the visualization of the capillary network in human eyes using decorrelation method OCTA and phase-resolved OCTA, respectively. Besides, with a shorter interscan time, the OCTA dynamic range between the slowest detectable flow and the fastest distinguishable flow could increase with the cost of compromising the slowest distinguishable flow [26]. Recently, Migacz et al. [41] also indicated the interscan time could play important role in contrast enhancement of OCTA images.

In general, the effect of interscan variation on the OCTA imaging will be most pronounced when the blood flow is not in the so-called saturation regime [26]. In particular, if the blood flow is flowing sufficiently fast, the OCTA signal will be, notwithstanding bulk motion or other higher order considerations, near to saturation regime and invariant to the interscan time. In this study, when comparing the OCTA images between the interscan time of 6.25 ms and 12.5 ms for the 200-kHz OCT imaging, the majority of the vessels appear to be at or near to the saturation regime (Fig. 4). Thus, an even shorter interscan time would likely allow greater increase in dynamic range.

On the other hand, the OCTA images from the 100-kHz OCT modality contained more detailed and clearer vasculatures than did the OCTA images from the 200-kHz OCT modality with the same B-scan interscan time τ (12.5 ms) and nearly the same sensitivity. Our results indicated that the correlation time τ_c or constant ξ of the 100-kHz OCT modality was less than that of the 200-kHz OCT modality, leading to an increased dynamic range, according to Eq. (4) or Eq. (6). Hence, the increased dynamic range also suggests that metric η listed in Tables 5 and 6 in terms of vessel area, total vessel length, and junction density for different vessel diameter parameters could all be positive values. Apart from the correlation time τ_c or constant ξ, the comparable or higher axial and lateral resolution of OCT images could benefit the visualization of capillary network [36]. However, the axial and lateral resolution were near the same for both 100-kHz and 200-kHz OCT system in our study. Besides, the number of OCTA images averaged to compute the final OCTA image per location was 3 and 4 for the 200-kHz and 100-kHz OCT imaging with an interscan time of 12.5 ms, respectively. As shown in Tables 4, an increase in most of the microvascular parameters was observed in the OCTA images computed with 3 OCTA images averaged than that with the 4 OCTA images averaged. It could be expected when the number of the OCTA images averaged decreases, the background noise could be increased due to the decreased signal-to-noise (SNR) ratio. Nevertheless, according to Tables 2 and 4, for either the number of the OCTA images averaged for 100-kHz OCT imaging, there exists a significant difference in the OCTA images computed between 100-kHz and 200-kHz OCT imaging with the 12.5 ms interscan time.

Table 6. Quantitative comparison of microvasculature between the en face OCTA images of mouse ear skin from three randomly selected ROIs shown in Fig. 6. Regions 1-3 correspond to ROIs marked by red, purple, and green boxes, respectively, in Fig. 6.

View Table | View all tables in this article

Some artifacts from bulk motion either caused by the respiration or the heart beating of mouse could not be easily removed without further image processing [42]. In the current study, although with the same OCTA interscan time, the OCTA images reconstructed from the 100-kHz OCT imaging showed more minute blood vessels of mouse ear skin than those with the 200-kHz OCT imaging. Because the peripheral areas of the mouse ear cannot easily and entirely be joined together with a sample holder, performing OCTA over the peripheral area of the original FOV made our images more vulnerable to influence from the aforementioned artifacts during 100-kHz OCT imaging than during 200-kHz OCT imaging. As illustrated in Fig. 6, motion artifacts are more prominent at the bottom part of the peripheral region, which is close to the mouse body. Thus, to perform an objective comparison of the microvascular parameters for the OCTA modalities with different A-scan rates, cropped en face OCTA images were used for the subsequent analysis with AngioTool to avoid compromising or confounding the quantitative analysis.

Physiological functions of mice vary with time, leading to blood flow rate variations. Thus, the blood flow rate might be altered during measurement with 100-kHz and 200-kHz OCT imaging, which could be another error source for the quantified values in Tables 2 to 6. The swept source with a quickly tunable A-scan rate provided benefits and reduced errors due to a longer interval between the two aforementioned experiments. Moreover, as AngioTool was used to conduct the quantified analysis, we did not eliminate all nonconnective pixels (background noise or nonconnective finer capillary) by removing small particles and filling the holes in the original OCTA images to fairly and accurately compare the results of OCTA images of ear skin from five mice. This limitation could also affect the values of vessel parameters listed in Tables 2 to 6 and the distributions of skeletons (Figs. 5(d-i)) of mice ear skin.

If the sample arm power for 200-kHz OCTA under safety specifications could be enhanced, a higher dynamic range may be achieved for observing the same tiny capillaries from a 100-kHz swept source with the same interscan time. Currently, the sample arm power transferred to the sample presented in Table 1 for both swept sources is still below the American National Standards Institute (ANSI) limit of 14.49 mW and 24.38 mW for human skin at the center wavelength of 1060 nm and bandwidth of 100 nm [43]. Therefore, our 200-kHz system configuration has a considerable margin for increasing the sample arm power of future OCTA of human skin. Future approaches will have advantages of higher speed, lower bulk motion effects, and either higher ranges or almost the same dynamic ranges of OCTA images compared with the original 100-kHz system.

5. Conclusion

In this study, we developed an SS-OCT imaging modality enabling either 100-kHz or 200-kHz OCT imaging by using two separate light sources. In addition, using the VISTA algorithm and the correlation mapping method, the intricate microvascular network of the mouse ear skin could be identified using OCTA. OCTA processes with different interscan times were compared; either 100-kHz or 200-kHz OCT showed finer microvasculature at a longer interscan time (e.g., 12.5 ms vs. 6.25 ms), suggesting an enhanced dynamic range of OCTA with the VISTA algorithm. VISTA and OCTA produced results for the ears of five mice; the framework allows the identification of finer microvasculature from 100-kHz OCT imaging than from 200-kHz OCT imaging with the same interscan time and a similar OCT system sensitivity. With the AngioTool, the quantitative analysis results have confirmed the previous observation through analyzing the en face OCTA images with the parameters, including vessel area, total vessel length, and junction density. In conclusion, the results of this study should lead to future improvements in the dynamic range of OCTA detection.

Funding

Japan Society for the Promotion of Science (18H01893); Japan Science and Technology Agency (JPMJMI18G8); National Taiwan University (NTU-IOTP-108PNTUS05, NTU-IOTP-109PNTUS05); Ministry of Science and Technology, Taiwan (107-2636-E-002-003, 108-2636-E-002-009, 109-2636-E-002-028).

Acknowledgments

The authors acknowledge the support from the Young Scholar Fellowship Program of the Ministry of Science and Technology (MOST) of Taiwan, Republic of China (ROC), (MOST-107-2636-E-002-003, 108-2636-E-002-009, and 109-2636-E-002-028), National Taiwan University (NTU-IOTP-108PNTUS05 and NTU-IOTP-109PNTUS05), Japan Science and Technology Agency (JPMJMI18G8) and Japan Society for the Promotion of Science (18H01893). Facility support was provided by the Molecular Imaging Center (MIC) at National Taiwan University. This article was subsidized by the Ministry of Science and Technology and National Taiwan University (NTU), Taiwan.

Disclosures

The authors declare that there are no conflicts of interest related to this article.

References

1. D. Huang, E. A. Swanson, C. P. Lin, J. S. Schuman, W. G. Stinson, W. Chang, M. R. Hee, T. Flotte, K. Gregory, and C. A. Puliafito, “Optical coherence tomography,” Science 254(5035), 1178–1181 (1991). [CrossRef]

2. E. A. Swanson and J. G. Fujimoto, “The ecosystem that powered the translation of OCT from fundamental research to clinical and commercial impact [Invited],” Biomed. Opt. Express 8(3), 1638–1664 (2017). [CrossRef]

3. R. A. Leitgeb, L. Schmetterer, W. Drexler, A. F. Fercher, R. J. Zawadzki, and T. Bajraszewski, “Real-time assessment of retinal blood flow with ultrafast acquisition by color Doppler Fourier domain optical coherence tomography,” Opt. Express 11(23), 3116–3121 (2003). [CrossRef]

4. S. Makita, Y. Hong, M. Yamanari, T. Yatagai, and Y. Yasuno, “Optical coherence angiography,” Opt. Express 14(17), 7821–7840 (2006). [CrossRef]

5. J. Fingler, D. Schwartz, C. H. Yang, and S. E. Fraser, “Mobility and transverse flow visualization using phase variance contrast with spectral domain optical coherence tomography,” Opt. Express 15(20), 12636–12653 (2007). [CrossRef]

6. S. M. R. Motaghiannezam, D. Koos, and S. E. Fraser, “Differential phase-contrast, swept-source optical coherence tomography at 1060 nm for in vivo human retinal and choroidal vasculature visualization,” J. Biomed. Opt. 17(2), 026011 (2012). [CrossRef]

7. R. Poddar, D. Y. Kim, J. S. Werner, and R. J. Zawadzki, “In vivo imaging of human vasculature in the chorioretinal complex using phase-variance contrast method with phase-stabilized 1-mu m swept-source optical coherence tomography,” J. Biomed. Opt. 19(12), 1 (2014). [CrossRef]

8. J. K. Barton and S. Stromski, “Flow measurement without phase information in optical coherence tomography images,” Opt. Express 13(14), 5234–5239 (2005). [CrossRef]

9. A. Mariampillai, B. A. Standish, E. H. Moriyama, M. Khurana, N. R. Munce, M. K. K. Leung, J. Jiang, A. Cable, B. C. Wilson, I. A. Vitkin, and V. X. D. Yang, “Speckle variance detection of microvasculature using swept-source optical coherence tomography,” Opt. Lett. 33(13), 1530–1532 (2008). [CrossRef]

10. J. Enfield, E. Jonathan, and M. Leahy, “In vivo imaging of the microcirculation of the volar forearm using correlation mapping optical coherence tomography (cmOCT),” Biomed. Opt. Express 2(5), 1184–1193 (2011). [CrossRef]

11. E. Jonathan, J. Enfield, and M. J. Leahy, “Correlation mapping method for generating microcirculation morphology from optical coherence tomography (OCT) intensity images,” J. Biophotonics 4(9), 583–587 (2010). [CrossRef]

12. Y. L. Jia, O. Tan, J. Tokayer, B. Potsaid, Y. M. Wang, J. J. Liu, M. F. Kraus, H. Subhash, J. G. Fujimoto, J. Hornegger, and D. Huang, “Split-spectrum amplitude-decorrelation angiography with optical coherence tomography,” Opt. Express 20(4), 4710–4725 (2012). [CrossRef]

13. R. K. Wang, S. L. Jacques, Z. Ma, S. Hurst, S. R. Hanson, and A. Gruber, “Three dimensional optical angiography,” Opt. Express 15(7), 4083–4097 (2007). [CrossRef]

14. Y. K. Tao, A. M. Davis, and J. A. Izatt, “Single-pass volumetric bidirectional blood flow imaging spectral domain optical coherence tomography using a modified Hilbert transform,” Opt. Express 16(16), 12350–12361 (2008). [CrossRef]

15. R. K. K. Wang, L. An, S. Saunders, and D. J. Wilson, “Optical microangiography provides depth-resolved images of directional ocular blood perfusion in posterior eye segment,” J. Biomed. Opt. 15(2), 020502 (2010). [CrossRef]

16. W. J. Choi, R. Reif, S. Yousefi, and R. K. Wang, “Improved microcirculation imaging of human skin in vivo using optical microangiography with a correlation mapping mask,” J. Biomed. Opt. 19(3), 036010 (2014). [CrossRef]

17. C. L. Chen, W. S. Shi, and W. R. Gao, “Imaginary part-based correlation mapping optical coherence tomography for imaging of blood vessels in vivo,” J. Biomed. Opt. 20(11), 116009 (2015). [CrossRef]

18. M. Miura, S. Makita, S. Azuma, Y. Yasuno, S. Sugiyama, T. Mino, T. Yamaguchi, T. Agawa, T. Iwasaki, Y. Usui, N. A. Rao, and H. Goto, “Evaluation of retinal pigment epithelium layer change in Vogt-Koyanagi-Harada disease with multicontrast optical coherence tomography,” Invest. Ophthalmol. Visual Sci. 60(10), 3352–3362 (2019). [CrossRef]

19. E. Borrelli, R. Sacconi, M. Brambati, F. Bandello, and G. Querques, “In vivo rotational three-dimensional OCTA analysis of microaneurysms in the human diabetic retina,” Sci. Rep. 9(1), 16789 (2019). [CrossRef]

20. O. Jonas, H. Jon, and B. E. J. Gregor, “Advances in optical coherence tomography in dermatology—a review,” J. Biomed. Opt. 23(12), 1–10 (2018). [CrossRef]

21. J. Lee, W. H. Jang, S. Shim, B. Kim, W.-S. Jang, J. K. Myung, S. Park, and K. H. Kim, “Characterization of early-stage cutaneous radiation injury by using optical coherence tomography angiography,” Biomed. Opt. Express 11(5), 2652–2664 (2020). [CrossRef]

22. A. J. Deegan, F. Talebi-Liasi, S. Z. Song, Y. D. Li, J. J. Xu, S. J. Men, M. M. Shinohara, M. E. Flowers, S. J. Lee, and R. K. K. Wang, “Optical coherence tomography angiography of normal skin and inflammatory dermatologic conditions,” Lasers Surg. Med. 50(3), 183–193 (2018). [CrossRef]

23. S. Azuma, S. Makita, D. Kasaragod, S. Sugiyama, M. Miura, and Y. Yasuno, “Clinical multi-functional OCT for retinal imaging,” Biomed. Opt. Express 10(11), 5724–5743 (2019). [CrossRef]

24. M.-T. Tsai, I. C. Lee, Z.-F. Lee, H.-L. Liu, C.-C. Wang, Y.-C. Choia, H.-Y. Chou, and J.-D. Lee, “In vivo investigation of temporal effects and drug delivery induced by transdermal microneedles with optical coherence tomography,” Biomed. Opt. Express 7(5), 1865–1876 (2016). [CrossRef]

25. Y. Li, P. Tang, S. Song, A. Rakymzhan, and R. K. Wang, “Electrically tunable lens integrated with optical coherence tomography angiography for cerebral blood flow imaging in deep cortical layers in mice,” Opt. Lett. 44(20), 5037–5040 (2019). [CrossRef]

26. W. Choi, E. M. Moult, N. K. Waheed, M. Adhi, B. Lee, C. D. Lu, T. E. de Carlo, V. Jayaraman, P. J. Rosenfeld, J. S. Duker, and J. G. Fujimoto, “Ultrahigh-speed, swept-source optical coherence tomography angiography in nonexudative age-related macular degeneration with geographic atrophy,” Ophthalmology 122(12), 2532–2544 (2015). [CrossRef]

27. S. B. Ploner, E. M. Moult, W. Choi, N. K. Waheed, B. Lee, E. A. Novais, E. D. Cole, B. Potsaid, L. Husvogt, J. Schottenhamml, A. Maier, P. J. Rosenfeld, J. S. Duker, J. Hornegger, and J. G. Fujimoto, “Toward quantitative optical coherence tomography angiography: visualizing blood flow speeds in ocular pathology using variable interscan time analysis,” Retina 36(Suppl 1), S118–S126 (2016). [CrossRef]

28. E. Zudaire, L. Gambardella, C. Kurcz, and S. Vermeren, “A computational tool for quantitative analysis of vascular networks,” PLoS One 6(11), e27385 (2011). [CrossRef]

29. L. Beckmann, X. Zhang, N. A. Nadkarni, Z. Cai, A. Batra, D. P. Sullivan, W. A. Muller, C. Sun, R. Kuranov, and H. F. Zhang, “Longitudinal deep-brain imaging in mouse using visible-light optical coherence tomography through chronic microprism cranial window,” Biomed. Opt. Express 10(10), 5235–5250 (2019). [CrossRef]

30. M. Foddis, K. Winek, K. Bentele, S. Mueller, S. Blumenau, N. N. Reichhart, S. Crespo-Garcia, D. Harnett, A. Ivanov, A. Meisel, A. Joussen, O. Strauss, D. Beule, U. Dirnagl, and C. Sassi, “An exploratory investigation of brain collateral circulation plasticity after cerebral ischemia in two experimental C57BL/6 mouse models,” J. Cereb. Blood Flow Metab. 40(2), 276–287 (2020). [CrossRef]

31. J. H. Meyer, P. P. Larsen, C. Strack, W. M. Harmening, T. U. Krohne, F. G. Holz, and S. Schmitz-Valckenberg, “Optical coherence tomography angiography (OCT-A) in an animal model of laser-induced choroidal neovascularization,” Exp. Eye Res. 184, 162–171 (2019). [CrossRef]

32. L. S. Carroll, H. Uehara, D. Fang, S. Choi, X. Zhang, M. Singh, Z. Sandhu, P. M. Cummins, T. M. Curtis, A. W. Stitt, B. J. Archer, and B. K. Ambati, “Intravitreal AAV2.COMP-Ang1 attenuates deep capillary plexus expansion in the aged diabetic mouse retina,” Invest. Ophthalmol. Visual Sci. 60(7), 2494–2502 (2019). [CrossRef]

33. Y. Sugano, T. Sekiryu, M. Furuta, R. Tomita, H. Shintake, H. Maehara, and A. Ojima, “Morphometrical evaluation of the choriocapillaris imaged by swept-source optical coherence tomography angiography,” Clin. Ophthalmol. 12, 2267–2276 (2018). [CrossRef]

34. X. Wei, T. T. Hormel, S. H. Pi, Y. K. Guo, Y. F. Jian, and Y. L. Jia, “High dynamic range optical coherence tomography angiography (HDR-OCTA),” Biomed. Opt. Express 10(7), 3560–3571 (2019). [CrossRef]

35. A. Q. Zhang, Q. Q. Zhang, C. L. Chen, and R. K. K. Wang, “Methods and algorithms for optical coherence tomography-based angiography: a review and comparison,” J. Biomed. Opt. 20(10), 100901 (2015). [CrossRef]

36. R. F. Spaide, J. G. Fujimoto, N. K. Waheed, S. R. Sadda, and G. Staurenghi, “Optical coherence tomography angiography,” Prog. Retinal Eye Res. 64, 1–55 (2018). [CrossRef]

37. M. Arya, R. Rashad, O. Sorour, E. M. Moult, J. G. Fujimoto, and N. K. Waheed, “Optical coherence tomography angiography (OCTA) flow speed mapping technology for retinal diseases,” Expert Rev. Med. Devices 15(12), 875–882 (2018). [CrossRef]

38. H. C. Lee, O. O. Ahsen, K. Liang, Z. Wang, C. Cleveland, L. Booth, B. Potsaid, V. Jayaraman, A. E. Cable, H. Mashimo, R. Langer, G. Traverso, and J. G. Fujimoto, “Circumferential optical coherence tomography angiography imaging of the swine esophagus using a micromotor balloon catheter,” Biomed. Opt. Express 7(8), 2927–2942 (2016). [CrossRef]

39. A. F. Fercher and J. D. Briers, “Flow visualization by means of single-exposure speckle photography,” Opt. Commun. 37(5), 326–330 (1981). [CrossRef]

40. B. Braaf, K. A. Vermeer, K. V. Vienola, and J. F. de Boer, “Angiography of the retina and the choroid with phase-resolved OCT using interval-optimized backstitched B-scans,” Opt. Express 20(18), 20516–20534 (2012). [CrossRef]

41. J. V. Migacz, I. Gorczynska, M. Azimipour, R. Jonnal, R. J. Zawadzki, and J. S. Werner, “Megahertz-rate optical coherence tomography angiography improves the contrast of the choriocapillaris and choroid in human retinal imaging,” Biomed. Opt. Express 10(1), 50–65 (2019). [CrossRef]

42. Y. J. Hong, S. Makita, F. Jaillon, M. J. Ju, E. J. Min, B. H. Lee, M. Itoh, M. Miura, and Y. Yasuno, “High-penetration swept source Doppler optical coherence angiography by fully numerical phase stabilization,” Opt. Express 20(3), 2740–2760 (2012). [CrossRef]

43. L. I. o. America, “American National Standards for Safe Use of Lasers in Health Care,” ANSI Z136.1-2014, 83–87 (2014).