The nervous system, based on a complex network of nerves and cells, carries messages by transmitting signals to and from different parts of the body. The supply of blood flow in the nervous system is critical to maintain physiological functions. Optical coherence tomography (OCT) has the ability to noninvasively image the microvascular networks and accurately quantify the blood flow in vivo with high spatial and temporal resolutions in three dimensions. It has been used to measure changes in blood supplies and assess physiological functions before and after occurrences of diseases, drug administration, and external stimulation in the nervous system, especially in the brain. In this paper, we discuss current challenges of Doppler OCT and angiography in the data processing algorithms and imaging systems for neuroscience research. The advancements and feasible solutions for current challenges are proposed.

The nervous system is a complex network consisting of nerves and cells which functions as signal transmission among different parts of the biological body. The brain, as the center of the nervous system, exerts centralized control to the other organs. The brain consists of a huge number of neurons connected to each other and functional vascular networks for blood supply to the neural networks. Changes in blood supply may deactivate or activate the functions of neurons, alter the structures of neural networks, and adjust the patterns of signal transmission. The quantification and mapping of blood flow are important for understanding the brain function and research on occurrences and development of diseases, drug administration, and responses to external stimulation in the brain. The imaging of blood flow provides a powerful tool for the visualization of vascular networks.

In clinical applications, many techniques have been used for neuroimaging, such as magnetic resonance imaging (MRI),1–3 computed tomography (CT),4–6 and positron emission tomography (PET).7–9 MRI can evaluate the anatomy of a whole brain and image cerebral perfusion based on serial measurements of concentrations of blood oxygen or tracer agents.2 In a CT system, cross-sectional images of the whole brain are generated by the reconstruction of X-ray intensities measured by the detectors in a circular orbit, and the contrasts come from the differences in X-ray attenuation of different tissues. In a perfusion study using CT, contrast agents are injected into the body and cause transient hyperattenuation while they pass through the detected cerebral tissues.10 PET generates semiquantitative brain images by localizing radiotracers with a radioisotope and specific ligands. When the specific ligands are utilized by the active cells in the cerebral tissues, the gamma rays generated from the radioisotope are detected, and thus, the specific ligands are located.9 These techniques have the advantage of high penetration and the ability of three-dimensional (3D) imaging; therefore, the blood supply and tissue activities in a whole brain can be imaged. However, the spatial resolutions of conventional MRI, PET, and CT techniques are over ∼500 μm.11 Ultrasound imaging (US) has a better spatial resolution of ∼100 μm and can image the vascular networks in a depth over ∼1 cm. However, it remains challenging for US to image small vessels and capillaries because most of the vessels in the brain have diameters smaller than 100 μm. Optical imaging techniques present significant advantages for spatial resolution.

For neuroimaging, several optical techniques have been developed with different penetration depths, spatial resolutions, and imaging speeds, including optical coherence tomography (OCT),12–15 photoacoustic imaging (PAI),16–19 fluorescence microscopy (FM),20–23 near-infrared spectral imaging (NIRS),24–27 and laser speckle imaging (LSI).28–31 NIRS assesses cerebral hemodynamics based on temporal variations of oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) concentrations measured by their absorption of near-infrared light (700-1000 nm).32 Active cortical regions have more blood flow and present increased levels of HbO and decreased levels of HbR.32 However, NIRS cannot visualize the vascular network due to a relatively low spatial resolution. LSI uses coherent light to illuminate the brain cortex and observes a speckle pattern with a camera.33 The relative changes in flow velocities can be mapped by measurements of the variations in speckle contrasts.34 It is not easy for LSI to quantify the blood flow because LSI is a wide-field imaging technique, and the interference from the surrounding tissues could distort the measurements.35,36 Fluorescence microscopy provides high and super spatial resolutions for cortical tissues and cerebral vascular networks by detecting fluorescent dyes, fluorescent proteins, or autofluorescence.20,22,37 However, the imaging depth is limited by high fluorescence scattering in tissues. PAI produces images with optical contrast, ultrasonic resolution, and ultrasound penetration depth in which a low-energy pulse laser induces ultrasonic waves in a tissue and an ultrasound transducer detects the ultrasonic waves.12,16 When blood has stronger absorption at the selected wavelength of a pulse laser in comparison with other tissues, the blood flow can be imaged with a better contrast. OCT performs cross-sectional imaging by the use of a broadband light source with a short coherence length, based on the principle of low-coherence interferometry or white light interferometry.38–41 OCT is a powerful technique for noninvasive and 3D volumetric imaging with a spatial resolution of 1-15 μm, a penetration depth of 1-3 mm in highly scattering tissues, and an A-line speed from kHz to MHz. The OCT imaging field of view covers 5-15 mm by transversely scanning the incident optical beam, and OCT axial imaging is performed by measuring the magnitude and echo time delay of the backscattered light.42 The techniques for neuroimaging are compared in Fig. 1.

FIG. 1.

Comparisons of the penetration depth, spatial resolution, and image speed among various techniques for neuroimaging, including magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), ultrasound imaging (US), optical coherence tomography (OCT), photoacoustic imaging (PAI), fluorescence microscopy (FM), near-infrared spectral imaging (NIRS), and laser speckle imaging (LSI).

FIG. 1.

Comparisons of the penetration depth, spatial resolution, and image speed among various techniques for neuroimaging, including magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), ultrasound imaging (US), optical coherence tomography (OCT), photoacoustic imaging (PAI), fluorescence microscopy (FM), near-infrared spectral imaging (NIRS), and laser speckle imaging (LSI).

Close modal

A typical OCT system based on a Michelson interferometer has a low-coherence broadband light source, a sample light path, a reference light path, a coupler for light interference, and a photodetector. In a swept-source OCT system, a swept source laser successively generates low-coherence light at a single wavelength, and an amplified photodetector successively monitors the interference light at each wavelength. Instead of the swept source, a superluminescent diode is used in a spectral-domain OCT system which simultaneously outputs a broadband spectrum. The interference light at different wavelengths is separated by a grating and is simultaneously captured by using a camera. After the intensities of interference light are recorded at a series of wavelengths by the OCT system, they will be processed by a Fourier transform and the complex OCT signal F will be achieved at a specific spatial position. The 3D dataset of complex F will be used for the blood flow measurements.

Since the first report of Doppler OCT in 1997,43 a continuous increase in the research on methods and applications of Doppler OCT and angiography has been present.44 The noninvasive, high-resolution nature of Doppler OCT and angiography allows in vivo applications in neuroimaging,45,46 ophthalmology,47–54 and dermatology.55–60 An ideal OCT system for neuroimaging should provide high-contrast images of microvascular networks, quantify accurate flow velocities in the networks, cover deep imaging range, process the data with high speed for real-time analysis, etc. However, some challenges in the data processing algorithms and imaging systems still remain to be solved. In this paper, we discuss current challenges of Doppler OCT and angiography and propose some advancements and feasible solutions to address the challenges in neuroimaging.

Doppler OCT and angiography can quantify blood flow and map microvascular networks based on the processing of OCT complex data. The OCT complex number F consists of a phase component φ and an amplitude component A, which can be expressed as A × exp(). Both the phase and amplitude components carry information about the motion of particles (i.e., red blood cells in blood flow). Therefore, Doppler OCT and angiography can be achieved by measurement of phases, analysis of amplitudes, and calculation of complex data incorporating phase and amplitude components. Methods for Doppler OCT and angiography are summarized in Table I. Generally, the principles of these methods are based on the calculation of differences in OCT data over time, which are captured at the same 3D spatial position. In practice, there are two procedures for data analysis, including inter-frame analysis of neighboring B-frames in the slow scan direction and inter-A-line analysis of adjacent A-lines in the fast scan direction. When the distance between two sampling positions is much smaller in comparison with the optical beam size, the two sampling positions can be approximately regarded as the same. The principles of calculation are similar between inter-A-line and inter-frame analyses; however, the inter-frame calculation has a much longer equivalent sampling time interval ΔT than the calculation based on the inter-A-lines.

TABLE I.

Comparisons of various data processing algorithms for Doppler OCT and angiography.

IncorporatedFlowMathematicEquation index
MethodEquationcomponentquantificationmethodin this paper
Doppler phase shift V×cos(θ) Phase Quantitative Subtraction (1
=Δf×λ2×n 
Phase variance V×sinθ Phase Quantitative Standard (5
=8×λ×σfπ×n×NA deviation 
Intensity-based σi2=1 Amplitude Semiquantitative Decorrelation (14
Doppler variance 2×m=1M1Fm×Fm+1m=1M1Fm2+m=1M1Fm+12 
Amplitude- σi2=1 Amplitude Semiquantitative Decorrelation (15
decorrelation 1M1×m=1M12×Fm×Fm+1Fm2+Fm+12 
Speckle variance σs2=1M Amplitude Semiquantitative Variance (16
×m=1MFm1M×m=1MFm2 
Intensity Iflow=1M1 Amplitude Semiquantitative Subtraction (17
differentiation ×m=1M1FmFm+1 
Intensity relative Rsd Amplitude Semiquantitative Relative standard (18
standard deviation = 1M1×m=1M(Fm1M×m=1MFm)21M×m=1MFm deviation 
Phase-resolved σc2=1 Amplitude Semiquantitative Decorrelation (19
Doppler variance 2×m=1M1Fm×Fm+1*m=1M1Fm2+m=1M1Fm+12 and phase 
Complex Iflow Amplitude Semiquantitative Subtraction (20
differentiation =1M1×m=1M1FmFm+1 and phase 
IncorporatedFlowMathematicEquation index
MethodEquationcomponentquantificationmethodin this paper
Doppler phase shift V×cos(θ) Phase Quantitative Subtraction (1
=Δf×λ2×n 
Phase variance V×sinθ Phase Quantitative Standard (5
=8×λ×σfπ×n×NA deviation 
Intensity-based σi2=1 Amplitude Semiquantitative Decorrelation (14
Doppler variance 2×m=1M1Fm×Fm+1m=1M1Fm2+m=1M1Fm+12 
Amplitude- σi2=1 Amplitude Semiquantitative Decorrelation (15
decorrelation 1M1×m=1M12×Fm×Fm+1Fm2+Fm+12 
Speckle variance σs2=1M Amplitude Semiquantitative Variance (16
×m=1MFm1M×m=1MFm2 
Intensity Iflow=1M1 Amplitude Semiquantitative Subtraction (17
differentiation ×m=1M1FmFm+1 
Intensity relative Rsd Amplitude Semiquantitative Relative standard (18
standard deviation = 1M1×m=1M(Fm1M×m=1MFm)21M×m=1MFm deviation 
Phase-resolved σc2=1 Amplitude Semiquantitative Decorrelation (19
Doppler variance 2×m=1M1Fm×Fm+1*m=1M1Fm2+m=1M1Fm+12 and phase 
Complex Iflow Amplitude Semiquantitative Subtraction (20
differentiation =1M1×m=1M1FmFm+1 and phase 

1. Phase-resolved Doppler OCT measurements

Based on the Doppler principle, the frequency shift Δf of the backscattered light from a moving particle has a relationship with the velocity V of the particle along the flow direction, which is described by the following equation:13,57
(1)
where n is the tissue refractive index, λ is the central wavelength of light in a vacuum, θ is the Doppler angle defined as the angle between the incident optical beam and the flow direction, and V × cos(θ) is the flow velocity along the OCT beam direction, which is shown in Fig. 2(a). As the frequency shift Δf can be calculated by the phase change Δφ within a sampling time interval ΔT based on the following equation:15,
(2)
the flow velocity can be determined by the phase measurements using the following equation:57 
(3)
The phase change Δφ can be calculated by the OCT complex data using the following equation:56 
(4)
where Fm and Fm+1 are the complex OCT data captured at the same position but at a different time, Im() and Re() are the imaginary part and real part of an OCT complex signal, respectively, and F* is the conjugate complex of F.
FIG. 2.

Schematic of phase-resolved Doppler OCT measurements. (a) The flow velocity with the axial and transverse components. (b) Flow velocity measurement based on dual-beam-angle OCT. (c) Flux measurement based on the en face integration of axial velocities.

FIG. 2.

Schematic of phase-resolved Doppler OCT measurements. (a) The flow velocity with the axial and transverse components. (b) Flow velocity measurement based on dual-beam-angle OCT. (c) Flux measurement based on the en face integration of axial velocities.

Close modal
From Eq. (1), the velocity cannot be quantified when the Doppler angle is close to 90°. In order to measure the transverse flow velocity, a phase variance method has been developed.61 When the backscattering light from the moving particles enters the different sides of an objective with a large numeric aperture, the frequency shifts Δf will change due to the differences of the Doppler angles. Based on the analysis of frequency shift bandwidth, the flow velocity can be calculated by the following equation:61,
(5)
where σf is the standard deviation (SD) of the frequency shift Δf, NA is the effective numerical aperture of the scan objective and V × sin(θ) is the flow velocity perpendicular to the OCT beam direction, which is shown in Fig. 2(a). The standard deviation σf of frequency shifts can be calculated by the following equation:56 
(6)
where Δf¯ is the average Doppler shift and Pf) is the Doppler power spectrum. Equivalently, σf2 can also be calculated by the following equation:
(7)
where M is the repetition number of measurements at the same position and Δfm is the frequency shift at the mth measurement. When the frequency shift Δf is calculated by the phase change Δφ using Eq. (2), the standard deviation σf of the frequency shifts can be calculated by the standard deviation σp of the phase changes using the following equation:62–64 
(8)

Either the axial velocity measured by the phase shifts or the transverse velocity measured by the standard deviations of the phase changes can be used to calculate the velocity along the flow direction when the Doppler angle is determined between the incident optical beam and flow direction, as shown in Fig. 2(a).15 In order to determine the Doppler angles, Qi et al. developed a vessel reconstruction approach based on vessel segmentation and skeletonization in 3D, which is described in Fig. 3.46 Incorporating the calculation of the Doppler angles and the measurements of the phase shifts, the flow direction and absolute velocities can be determined automatically in each position of the entire middle cerebral artery (MCA) branches. Makita et al. measured the 3D vessel geometry by incorporating an en face vessel image and two representative cross-sectional flow images.65 Wang et al. scanned across the blood vessel twice with a displacement between two scanning planes and reconstructed the geometry of vessels by connecting the successive vessel centers in two planes.66 You et al. applied a gradient vessel tracking method to numerically approach the flow angles within a region of interest.67 These methods were used for the determination of Doppler angles in larger branches which, however, might not be effective in smaller vessels and capillaries.

FIG. 3.

Flow velocity measurements in major MCA branches of the rat brain. (a) En face vascular mapping using the intensity-based Doppler variance method. (b) Automatic vessel skeleton reconstruction. (c) Doppler angles represented by the cosine values. (d) Reconstructed flow velocities distributed along the entire vessels. Reproduced with permission from Qi et al., Biomed. Opt. Express 7, 601 (2016). Copyright 2016 Optical Society of America.

FIG. 3.

Flow velocity measurements in major MCA branches of the rat brain. (a) En face vascular mapping using the intensity-based Doppler variance method. (b) Automatic vessel skeleton reconstruction. (c) Doppler angles represented by the cosine values. (d) Reconstructed flow velocities distributed along the entire vessels. Reproduced with permission from Qi et al., Biomed. Opt. Express 7, 601 (2016). Copyright 2016 Optical Society of America.

Close modal
In order to avoid the difficulty in determining Doppler angles, Pedersen et al. developed a dual-beam-angle OCT system for the quantification of flow velocities68 in which the axial flow velocities were detected by two optical beams with a known separation angle α as shown in Fig. 2(b). As the relationship between the Doppler angles θ1 and θ2 for Beams 1 and 2, respectively, can be expressed as θ2 = θ1 + α, the Doppler angle θ1 can be calculated according to68 
(9)
based on the solution of the following equation:
(10)
after the measurements of the axial velocities v1 and v2 along optical Beams 1 and 2, respectively.
Without the predetermination of Doppler angles, Srinivasan et al. proposed quantitative measurements of blood flux based on the en face integration of axial flow velocities at a specific depth, which is shown in Fig. 2(c).69 The blood flux U in a vessel can be calculated by the velocity V along the flow direction and the cross-sectional area Scross of the vessel, which is simplified below,
(11)
As the velocity V can be determined by Vaxial/cos(θ) and the area Scross can be calculated by SEnFace × cos(θ), Eq. (11) can be described as:
(12)
where Vaxial is the axial velocity along the optical beam and SEnFace is the vessel area in the en face transverse plane.

By the measurements of the axial velocity Vaxial using phase shifts and the measurements of the vessel area in the en face transverse plane at a given depth, the blood flux can be quantified.

Doppler OCT imaging based on phase measurements has the ability to quantify the blood flow. In the case where visualization of microvascular networks is important and accurate flow quantification is not critical, intensity-based Doppler OCT and angiography are proposed.

2. Intensity-based Doppler OCT and angiography

Intensity-based Doppler OCT and angiography utilize the amplitude component of the OCT complex signal to visualize the vascular networks instead of the phase component. Therefore, intensity-based methods can be used in a phase-instable system, and data processing will be much simpler without the correction of phase wrapping. Zhao et al. proposed a Doppler variance method for blood flow imaging, as shown in the following equation:56,
(13)
based on the calculation of decorrelation. Liu et al. developed an intensity-based Doppler variance algorithm by incorporating averaging windows M, which is expressed as70,
(14)
where only amplitudes of the OCT complex signals are considered. The intensity-based Doppler variance method can be used to sensitively detect the transverse motion.71–73 Later, Jia et al. described an amplitude-decorrelation algorithm based on a different averaging procedure, which is shown in the following equation:74 
(15)
Mariampillai et al. developed and optimized the speckle variance detection of microvasculature based on the calculation of the variances of structural OCT intensities among M repetitive measurements, which is described in the following equation:75,76
(16)
The use of inter-frame analysis with a relatively lower frame rate resulted in a larger time interval between neighboring frames. The larger time interval increased the speckle variance contrast which, however, may cause saturation of the variances and loss of a quantitative relationship between the variance and flow velocity.
In order to simplify the data processing and speed up the microvascular imaging, Huang et al. achieved angiography by performing direct differentiation of the OCT intensity using the following equation:77 
(17)
The operation of differentiation can be regarded as high pass filtering and, thus, can suppress the signals from the stationary objects.78 In practice, repetition measurements can be made by repeated B-scanning at one lateral location, and the subtraction of OCT intensities between two consecutive B-scans is calculated.
The intensity-based methods can be used for the visualization of vascular networks, and each of these methods has its unique advantages.79,80 However, the quantification of flow velocities remains a challenge. The flow velocities can be quantified by intensity-based OCT measurements after calibration in a range. The intensity-based Doppler variance based on inter-A-line analysis was demonstrated to be insensitive to Doppler angles when the Doppler angle was close to 90°.81 For a complex vascular network, the Doppler angles may not be limited to around 90°. Zhu et al. developed a method for the mapping of microvascular networks and quantification of flow velocities by calculating the relative standard deviation (SD) of OCT amplitudes.82 The relative SD was demonstrated to be angle-insensitive within a much wider range based on inter-A-line analysis and to have a nearly linear relationship with the flow velocities which is expressed as the following equation:82 
(18)
After calibration, the flow velocity can be quantified in vivo without an explicit knowledge of Doppler angles.

3. Algorithms based on complex signal analysis

From an OCT system, the complex data will be achieved after Fourier transformation of interference signals. Either phase information or amplitude information can be used for flow quantification and vascular mapping. Some methods using complex signal analysis have incorporated both phase and amplitude information. Based on Eq. (13), Zhao et al. performed an averaging of complex data to develop a phase-resolved Doppler variance method,56 which is expressed below:
(19)
Equation (19) incorporates the differences of the phases and amplitudes in the averaging window.81,83
Based on the differential calculation of neighboring B frames, An et al. proposed an OCT angiography method as shown in the following equation:55 
(20)
in which the magnitude was taken and averaged after performing differentiation of the complex signals. Similar to Eq. (17), the subtraction operation can be approximately regarded as high-pass filtering of the data.

The vascular mapping based on the analysis of complex signals involves phase and amplitude items and, therefore, may achieve more information about the flow. However, it is more sensitive to bulk motion and requires a phase stable system and more post-processing steps.

The data processing algorithms for neuroimaging should simultaneously achieve at least two goals: mapping the microvascular network and quantifying the flow velocity in the network. Each individual goal has been achieved by different algorithms; however, simultaneous achievements of both goals remain a challenge.

1. Challenges and solutions for phase-resolved Doppler OCT

Phase-resolved Doppler OCT can quantify flow velocities with a high temporal resolution. Yu et al. utilized the Doppler phase shift method to measure flow dynamics before and after localized ischemia induced by photocoagulation (light-induced clotting) of Rose Bengal photodynamic therapy (PDT) in mouse cortices.84 With fast, repeated B-scans across the vessels of interest, phase shift measurements recorded the flow velocities for over 10 times within a cardiac cycle, as shown in Fig. 4. The parameter of the resistance index (RI), which was calculated as (SD)/S for assessing the changes of the flow velocities in each cardiac cycle, was used to monitor the changes in vascular conditions before and after PDT. A higher RI value was associated with a higher resistance state of a vessel accompanied by the localized ischemic stroke. Wang et al. quantified the blood flow velocities in the pial arteries of rat sensory cortices before and after drug administration, light and electric stimulations with high velocity sensitivity and high imaging speed.85 Significant changes were measured in blood flow after injection of a drug with the function of vasodilation, light stimulation in front of the rat’s eye and electric stimulation of the sciatic nerve in the rat’s paw. Ren et al. measured the flow velocities in cerebral microvasculature after the use of cocaine in the mouse somatosensory cortices to assess cocaine-induced cerebral microischemia.86 They revealed that cocaine doses within the range administered by drug abusers induced long lasting cerebral microischemia, which were exacerbated with repeated cocaine use. However, there are some limitations for phase-resolved Doppler OCT to map microvascular networks.

FIG. 4.

Laser speckle imaging and phase-resolved Doppler OCT measurements before and after a photocoagulation event caused by Rose Bengal photodynamic therapy (PDT). [(a)–(c)] Laser speckle flow index maps. [(d)–(f)] Flow changes in vessel 1. [(g)–(i)] Flow changes in vessel 2. Reproduced with permission from Yu et al., J. Biomed. Opt. 15, 066006 (2010). Copyright 2010 Society of Photo-Optical Instrumentation Engineers (SPIE).

FIG. 4.

Laser speckle imaging and phase-resolved Doppler OCT measurements before and after a photocoagulation event caused by Rose Bengal photodynamic therapy (PDT). [(a)–(c)] Laser speckle flow index maps. [(d)–(f)] Flow changes in vessel 1. [(g)–(i)] Flow changes in vessel 2. Reproduced with permission from Yu et al., J. Biomed. Opt. 15, 066006 (2010). Copyright 2010 Society of Photo-Optical Instrumentation Engineers (SPIE).

Close modal

First, the measurements of Doppler phase shifts can only detect the velocity components along the OCT beam when the Doppler angles are not predetermined between the OCT beam and the vessels. Especially, when the Doppler angles are varied in a region of interest, the velocity components along the OCT beam cannot be used to compare the absolute velocities in the flow direction. In addition, the flow velocities cannot be detected accurately when the Doppler angle is close to 90°. In some cases, the OCT beam is nearly perpendicular to the microvascular beds, and thus, some segments of the vessels cannot be detected by phase-resolved Doppler OCT. To measure the absolute velocities and avoid the risk that the OCT beam is perpendicular to the vessels, two or more optical beams could be built in the OCT system to reach the vessels at different angles. Optimized design of beam incident angles may eliminate the possibility that the Doppler angle is close to 90°.

Second, phase wrapping results in an extra need for phase correction. Before the development of high-efficiency algorithms for phase wrapping correction, the Doppler phase shift method has not provided high-contrast images of microvascular networks in a wide dynamic range of flow velocities. Especially, the phase wrapping correction is more challenging in smaller vessels and capillaries. Akamatsu et al. compared the robust recruitment of leptomeningeal collateral flow after MCA occlusion in C57BL/6 mice using phase-resolved Doppler OCT.45 Spatial distribution of the flow velocities in C57BL/6 mice was mapped before and after MCA occlusion, which is shown in Fig. 5.45 Phase wrapping was present in the microvascular images and only velocity components along the OCT beam were mapped as the Doppler angles were not measured. In order to overcome the difficulties in accurate phase quantification, high-efficiency algorithms should be developed for phase wrapping correction. Alternatively, a feasible solution is the combination of images with different dynamic ranges of velocities. In each image, the wrapping signals are removed and only unwrapped signals are extracted.

FIG. 5.

Spatial distribution of axial flow velocities before and after MCA occlusion in a C57BL/6 mouse. The flow toward the optical beam is designated as red, and the opposite direction is designated as green. Scale bar, 1 mm. Reproduced with permission from Akamatsu et al., J. Neurosci. 35, 3851 (2015). Copyright 2015 the authors.

FIG. 5.

Spatial distribution of axial flow velocities before and after MCA occlusion in a C57BL/6 mouse. The flow toward the optical beam is designated as red, and the opposite direction is designated as green. Scale bar, 1 mm. Reproduced with permission from Akamatsu et al., J. Neurosci. 35, 3851 (2015). Copyright 2015 the authors.

Close modal

Finally, phase shift measurements require a phase-stable OCT imaging system. The distortion of phase measurements will decrease the signal-to-noise ratio of microvascular network images. To reduce the noise in the phase measurements, the phase stability of swept sources should be improved and spectral domain OCT systems can be used for the phase measurements. Phase distortion induced by trigger desynchronization may also be corrected by spectral phase encoding and instantaneous correlation among the A-scans in a swept-source OCT system.87 

2. Challenges and solutions for intensity-based measurements

For the analysis of cerebrovascular physiology, intensity-based algorithms have been demonstrated as promising methods to map microvascular networks in cerebral cortices with high spatial resolution. As animal models, rats and mice are widely used in neuroscience research. The microvasculatures of a mouse cortex and a rat cortex are shown in Fig. 6, which were visualized by the intensity-based Doppler variance algorithm.88 The intensity-based methods have been used to assess the flow change in cortical microvasculature. Using an intensity-based Doppler variance, Qi et al. compared the vascular networks before and after MCA occlusion on rats and revealed the consequences of ischemic stroke due to MCA occlusion.46 Intensity-based methods can also be used to assess the lumen diameters. Lin et al. used intensity-based OCT imaging to reveal amyloid-β-dependent vascular impairment in Alzheimer’s mouse model.89 It was found that superficial cortical vessels were thicker and more dilated in the 20-month controls than in the 20-month 3xTg-AD mice, which is shown in Fig. 7. From the OCT microvasculature images, the vessel volume fraction can be quantified by calculating the ratio of a vessel volume to a tissue volume. The volume fraction of superficial vessels showed a 29% decrease in 20-month 3xTg-AD mice than in 20-month controls from Fig. 7.89 Baran et al. applied the intensity-based OCT method to evaluate the changes in vessel lumen diameter among pial and penetrating arterioles across the penumbra region in the parietal cortices before and after MCA occlusion.90 The lumen diameters changed in response to the stroke due to MCA occlusion in the penetrating arterioles; however, they were only slightly affected in the pial arterioles. Microvascular mapping using intensity-based measurements has a higher contrast compared with the phase-resolved Doppler method; however, the ability of velocity quantification is limited for intensity-based measurements.

FIG. 6.

Microvascular mapping for (a) a rat cerebral cortex and (b) a mouse cerebral cortex using the intensity-based Doppler variance algorithm. Reproduced with permission from G. Liu and Z. Chen, Chin. Opt. Lett. 11, 011702 (2013). Copyright 2013 Chinese Optics Letters.

FIG. 6.

Microvascular mapping for (a) a rat cerebral cortex and (b) a mouse cerebral cortex using the intensity-based Doppler variance algorithm. Reproduced with permission from G. Liu and Z. Chen, Chin. Opt. Lett. 11, 011702 (2013). Copyright 2013 Chinese Optics Letters.

Close modal
FIG. 7.

Microvasculature imaging of cortices in 20-month control and 3xTg-AD mice using the intensity-based Doppler variance algorithm. Superficial vessels (∼200 µm) are designated as white, and deeper vessels are designated as pink. Reproduced with permission from Lin et al., Neurophotonics 1, 011005 (2014). Copyright 2014 the authors.

FIG. 7.

Microvasculature imaging of cortices in 20-month control and 3xTg-AD mice using the intensity-based Doppler variance algorithm. Superficial vessels (∼200 µm) are designated as white, and deeper vessels are designated as pink. Reproduced with permission from Lin et al., Neurophotonics 1, 011005 (2014). Copyright 2014 the authors.

Close modal

First, there is no direct quantitative relationship between the flow velocities and the values from intensity-based measurements. In order to explore the opportunities for quantification of flow velocities using intensity-based methods, Liu et al. demonstrated that the values from intensity-based Doppler variance measurements increased with the increase of the flow velocities at lower velocity regions and reached saturation at higher flow velocity regions.58 The variance values were able to quantify the flow velocity with a predetermined calibration curve at relatively lower velocity regions. The relative SD values also had a nearly linear relationship with the flow velocities in a range, and the flow velocities could be quantified after calibration. For the quantification of flow velocities, calibration is usually required, which will complicate the intensity-based measurements.

Second, the dependence of the values from intensity-based measurements on the angles between the vessels and the light beam has not been quantified directly. The variance values based on inter-A-line analysis have been demonstrated to be insensitive to Doppler angles close to 90°. The relative SD values have been demonstrated to be angle-insensitive within a much wider range of vessel angles based on inter-A-line analysis. For more accurate measurements of flow velocities, the influence of the vessel angles on the intensity-based values should be assessed in more detail.

Finally, there is a conflict over the selection of time intervals between quantification of flow velocities and mapping of microvascular networks. Usually, the time interval should be relatively shorter for flow quantification but long enough for the mapping of microvascular networks, especially the capillaries, which is discussed in Fig. 8. The relationship between the relative SD and the flow velocity depends on the time interval of adjacent A-lines in the inter-A-line calculation and depends on the time interval of neighboring frames in the inter-frame calculation, according to the analysis of a Gaussian function model.82 In Figs. 8(a)–8(c), the time intervals are doubled successively. The linear range varies with the changes of time intervals which are located in a faster velocity range when the time interval is smaller as shown in Fig. 8(a) and located in a slower velocity range when the time interval is larger as shown in Fig. 8(c). The points marked by the asterisks have a slower flow velocity, which is outside the nearly linear ranges with 1× and 2× time intervals, but can be quantified with a 4× time interval. The points marked by the circles have a faster flow velocity, which can be quantified with a 1× time interval but present saturated relative SD values with 2× and 4× time intervals. The increase of the time intervals will improve the detection sensitivity for slower flow and cause signal saturation for faster flow. When the time interval is large enough, even the slower flow will generate saturated relative SD values, and a high-contrast image can be achieved for the mapping of capillary networks. Figure 8(d) shows the vascular mapping of rat cortical branches with different time intervals based on relative SD measurements.82 

FIG. 8.

[(a)–(c)] Relationship between relative standard deviation (SD) and the flow velocity based on the analysis of a Gaussian function model. (a) Model with 1× time interval. (b) Model with 2× time interval. (c) Model with 4× time interval. The points marked by the asterisks and by the circles have the same flow velocities, respectively. (d) Vascular mapping of rat MCA branches with different time intervals. (d) was reproduced with permission from Zhu et al., Appl. Phys. Lett. 111, 181101 (2017). Copyright 2017 AIP Publishing LLC.

FIG. 8.

[(a)–(c)] Relationship between relative standard deviation (SD) and the flow velocity based on the analysis of a Gaussian function model. (a) Model with 1× time interval. (b) Model with 2× time interval. (c) Model with 4× time interval. The points marked by the asterisks and by the circles have the same flow velocities, respectively. (d) Vascular mapping of rat MCA branches with different time intervals. (d) was reproduced with permission from Zhu et al., Appl. Phys. Lett. 111, 181101 (2017). Copyright 2017 AIP Publishing LLC.

Close modal

Similar to the relative SD measurements, the relationship between the decorrelation value and the flow velocity can be well fitted with an exponential function.81 There is a slowest sensitivity threshold and a fastest saturation limit. The flow under the sensitivity threshold generates decorrelation values indistinguishable from background noise. The flow above the saturation limit produces similar decorrelation values and is indistinguishable from one another.51 The changes in time intervals will alter the slowest sensitivity threshold and the fastest saturation limit. An increase of time intervals results in the saturation of decorrelation values and a higher contrast for the slow flow. In most cases, the quantification of the flow velocities is performed based on inter-A-line calculation with a relatively shorter time interval, and the mapping of the microvascular networks is performed based on inter-frame analysis with a relatively larger time interval.

In order to address the challenge of selecting time intervals, a feasible solution is over sampling. By over sampling, different time intervals can be achieved in one measurement, and thus, the decorrelation values and relative SD values can be calculated with different time intervals. The dynamic range can be extended by incorporating measurements with different time intervals for velocity quantification, and the imaging contrast can be enhanced with a long time interval for the mapping of microvascular networks.

The algorithms for Doppler OCT and angiography are promising for use in the quantification of blood flow and the mapping of microvascular networks with high temporal and spatial resolutions in the study of neuroscience. However, there are also some challenges with imaging systems, such as limited imaging depth, motion artifacts, relatively slow imaging speed, and low imaging contrast for small capillaries. Therefore, further improvements are still required to meet these challenges.

First, the cortical tissues are highly optical scattering materials, and the penetration depth of OCT light is limited to a smaller range of 1-2 mm, compared with the size of a whole brain. Only superficial vessels near the surface of a cortex can be imaged clearly. However, some responses of blood flow to ischemic conditions, drug administration, and stimulation are present in deeper cortices. Chen et al. reported a 1.3 µm swept-source OCT system with spectral phase encoding and instantaneous correlation among the A-scans for the reduction of phase errors in deeper tissues, and the flow was readily detected at depths down to 3.2 mm in a mouse brain.87 Recently Li et al. developed an OCT system with a 1.7 µm swept-source laser, which can penetrate deeper tissues because of the longer wavelength, compared with the conventional 1.3 µm swept-source laser.91,92 Photoacoustic imaging and ultrasound imaging can detect much deeper tissues but suffer from lower resolution compared with OCT. The combination of OCT with photoacoustic imaging or ultrasound imaging may provide high-resolution images of microvascular networks in cortices and blood flow dynamics in deep cortical tissues.

Second, the motion artifacts are troubling factors in Doppler OCT and angiography93 because the blood flow is normally much slower than the physiological bulk motion. The bulk motion will cause a larger phase change which may hide the smaller phase change due to the blood flow. In addition, the large motion will also cause the repeated measurement data to not be captured in the same position, and thus, the Doppler OCT and angiography cannot be analyzed correctly. In order to extract the smaller phase change generated from the blood flow, the phase change due to the bulk motion should be removed. As the bulk motion usually induces a constant phase shift in each position, the motion artifacts can be corrected by subtracting the constant phase term from the data in an A-line. Yu et al. developed a histogram-based method for bulk-motion correction.52 As the diameters of vessels are relatively small compared to the imaging depth in tissue, the blood flow will generate phase changes in a few locations of an A-line and the tissue motion will contribute a constant phase change in more locations of an A-line. Thus, the peak in the phase change histogram of an A-line represents the phase change due to bulk motion. The distortion due to the bulk motion can be corrected by subsequently subtracting the constant phase change in the A-line. Algorithms insensitive to the bulk motion have also been developed. Liu et al. validated that the impact of bulk motion can be ignored when both lateral averaging and depth averaging are performed by the Doppler variance method.94 The large bulk motion can also be eliminated during image acquisition using a motion tracking system which generates almost motion-free OCT data for subsequent analysis.95,96

Third, the imaging speed for flow mapping and quantification depends on the capture time and processing efficiency. Fast data capture is critical for in vivo imaging when the bulk motion is significant during the measurements. There are two ways to achieve higher A-line rates, including the increase of single-beam A-line rates and the parallelization of illumination and acquisition. An increase of single-beam A-line rates relies on the use of faster swept lasers in swept-source OCT systems or higher speed cameras in spectral-domain OCT systems. In swept-source OCT systems, the vertical-cavity surface-emitting laser (VCSEL) with an A-line rate of 200 kHz and the Fourier domain mode locking (FDML) laser with an A-line rate up to MHz have been commercialized for OCT imaging.97 Zhi et al. achieved 4D OCT angiography with a frame rate over 3 kHz and a volume rate of 4.7 volumes/s (512 × 200 × 720 voxels) by using a 1.6 MHz FDML swept source.98 The ultrahigh speed complementary metal-oxide-semiconductor (CMOS) line scan camera can yield an A-line rate up to 312.5 kHz in a spectral-domain OCT system.99 Parallelization of illumination and acquisition allows imaging of multiple lateral points at one time and further accelerates data capture speeds which has been used in full-field OCT systems.100–102 In order to improve data processing efficiency, hardware units, such as digital signal processors (DSP), field programmable gate arrays (FPGA), and graphics processing units (GPU), have been applied so that the data can be processed in near real-time.103,104 The robustness and simplification of analysis algorithms also play significant roles in the acceleration of flow mapping and quantification.

Finally, because capillaries are relatively small with slow flow, the imaging of capillaries remains a challenge due to low contrast. In order to improve the ability to image small capillaries in a cortex, an increase of the spatial resolution in an OCT system and an enhancement of the contrast between the slow flow and the stationary background are potential solutions. (1) Based on the principle of OCT imaging, the axial resolution can be improved by the use of a laser source with a shorter central wavelength and broader bandwidth. Using a Ti:sapphire laser with a spectrum centered at 800 nm and a bandwidth of 170 nm, Werkmeister et al. developed ultrahigh-resolution OCT imaging with an axial resolution of approximately 1.2 μm.105 The recently developed visible OCT could further increase the axial resolution by switching the light source from near-infrared light to visible light.106 The light with a shorter wavelength will generally have more highly scattering coefficients in biological tissues. (2) The imaging contrast can be enhanced by an extension of the sampling time interval and an increase of optical scattering in tissues. From Fig. 8, the imaging contrast is enhanced when a larger time interval is selected in the measurements. Usually, the inter-frame analysis based on the calculation among neighboring frames will yield a higher contrast for capillary imaging, compared with inter-A-line analysis based on the calculation among adjacent A-lines. (3) The addition of highly optical scattering materials will increase the sensitivity for measurements of capillary flows. Using the intravenous injection of an intralipid solution, which is safe for use as parenteral nutrition, Pan et al. improved the detection sensitivity for capillary networks, as shown in Fig. 9.107 Figures 9(a) and 9(b) show flow mapping before and after the injection of intralipid. The capillaries can be clearly imaged after the injection of intralipid. It is noteworthy that the penetration depth will be reduced by more highly scattering though more highly scattering can generate a higher imaging contrast of microvascular mapping.

FIG. 9.

Flow quantification using Doppler OCT. (a) Flow mapping before injection of intralipid. (b) Flow mapping after the injection of intralipid. Reproduced with permission from Pan et al., Neuroimage 103, 492 (2014). Copyright 2014 the authors.

FIG. 9.

Flow quantification using Doppler OCT. (a) Flow mapping before injection of intralipid. (b) Flow mapping after the injection of intralipid. Reproduced with permission from Pan et al., Neuroimage 103, 492 (2014). Copyright 2014 the authors.

Close modal

The measurement of blood flow is important for the study of physiological functions in a nervous system. Benefiting from noninvasive and 3D volume imaging of OCT with high temporal and spatial resolutions, Doppler OCT and angiography can not only quantify the blood flow in a cortex but also map the networks consisting of arteries, veins, and capillaries when various algorithms are incorporated into OCT complex data. Some challenges in data processing algorithms and imaging systems remain to be solved. Doppler OCT and angiography can be applied to monitor the dynamics of blood flow in vivo under physiological conditions and have promised to be powerful tools for neuroimaging.

This work was supported by grants from the National Institutes of Health (Nos. R01HL-125084, R01HL-127271, R01EY-026091, R01EY-021529, P41EB-015890, and R01EY028662). Dr. Zhongping Chen has a financial interest in OCT Medical Imaging, Inc., which, however, did not support this work.

1.
B. E.
McGehee
,
J. M.
Pollock
, and
J. A.
Maldjian
,
J. Magn. Reson. Imaging
36
,
1257
(
2012
).
2.
J. R.
Petrella
and
J. M.
Provenzale
,
AJR, Am. J. Roentgenol.
175
,
207
(
2000
).
4.
D. T.
Ginat
and
R.
Gupta
,
Annu. Rev. Biomed. Eng.
16
,
431
(
2014
).
5.
Y. W.
Lui
,
E. R.
Tang
,
A. M.
Allmendinger
, and
V.
Spektor
,
Am. J. Neuroradiol.
31
,
1552
(
2010
).
6.
J. M.
Wardlaw
and
O.
Mielke
,
Radiology
235
,
444
(
2005
).
8.
J. J.
Vaquero
and
P.
Kinahan
,
Annu. Rev. Biomed. Eng.
17
,
385
(
2015
).
9.
C.
Slough
,
S. C.
Masters
,
R. A.
Hurley
, and
K. H.
Taber
,
J. Neuropsychiatry Clin. Neurosci.
28
,
A4
(
2016
).
10.
E. M.
de Lucas
,
E.
Sánchez
,
A.
Gutiérrez
,
A. G.
Mandly
,
E.
Ruiz
,
A. F.
Flórez
,
J.
Izquierdo
,
J.
Arnáiz
,
T.
Piedra
,
N.
Valle
,
I.
Bañales
, and
F.
Quintana
,
Radiographics
28
,
1673
(
2008
).
11.
T. J.
Sejnowski
,
P. S.
Churchland
, and
J. A.
Movshon
,
Nat. Neurosci.
17
,
1440
(
2014
).
12.
Y.
Cho
,
G.
Zheng
,
G.
Augustine
,
D.
Hochbaum
,
A.
Cohen
,
T.
Knopfel
,
F.
Pisanello
,
F.
Pavone
,
I.
Vellekoop
,
M.
Booth
,
S.
Hu
,
J.
Zhu
,
Z.
Chen
, and
Y.
Hoshi
,
J. Opt.
18
,
093007
(
2016
).
13.
Z.
Chen
and
J.
Zhang
, in
Optical Coherence Tomography: Technology and Applications
, 2nd ed., edited by
W.
Drexler
and
J. G.
Fujimoto
(
Springer International Publishing
,
Cham
,
2015
), p.
1289
.
14.
U.
Baran
and
R. K.
Wang
,
Neurophotonics
3
,
010902
(
2016
).
15.
Z.
Chen
and
J.
Zhu
, in
Neurophotonics and Brain Mapping
, 1st ed., edited by
Y.
Chen
and
B.
Kateb
(
Taylor & Francis
,
Boca Raton
,
2017
), p.
159
.
16.
17.
J.
Yao
,
L.
Wang
,
J. M.
Yang
,
K. I.
Maslov
,
T. T.
Wong
,
L.
Li
,
C. H.
Huang
,
J.
Zou
, and
L. V.
Wang
,
Nat. Methods
12
,
407
(
2015
).
18.
M.
Nasiriavanaki
,
J.
Xia
,
H.
Wan
,
A. Q.
Bauer
,
J. P.
Culver
, and
L. V.
Wang
,
Proc. Natl. Acad. Sci. U. S. A.
111
,
21
(
2014
).
19.
P.
Zhang
,
L.
Li
,
L.
Lin
,
P.
Hu
,
J.
Shi
,
Y.
He
,
L.
Zhu
,
Y.
Zhou
, and
L. V.
Wang
,
J. Biophotonics
11
,
e201700024
(
2018
).
20.
J. N.
Kerr
and
W.
Denk
,
Nat. Rev. Neurosci.
9
,
195
(
2008
).
21.
C. J.
Engelbrecht
,
R. S.
Johnston
,
E. J.
Seibel
, and
F.
Helmchen
,
Opt. Express
16
,
5556
(
2008
).
22.
J. H.
Park
,
W.
Sun
, and
M.
Cui
,
Proc. Natl. Acad. Sci. U. S. A.
112
,
9236
(
2015
).
23.
K.
Ohki
,
S.
Chung
,
Y. H.
Ch'ng
,
P.
Kara
, and
R. C.
Reid
,
Nature
433
,
597
(
2005
).
25.
E.
Gratton
,
V.
Toronov
,
U.
Wolf
,
M.
Wolf
, and
A.
Webb
,
J. Biomed. Opt.
10
,
11008
(
2005
).
26.
A.
Sassaroli
,
M.
Pierro
,
P.
Bergethon
, and
S.
Fantini
,
IEEE J. Sel. Top. Quantum Electron.
18
,
1478
(
2012
).
27.
A.
Vermeij
,
A. H.
van Beek
,
M. G.
Olde Rikkert
,
J. A.
Claassen
, and
R. P.
Kessels
,
PLoS One
7
,
e46210
(
2012
).
29.
A. B.
Parthasarathy
,
E. L.
Weber
,
L. M.
Richards
,
D. J.
Fox
, and
A. K.
Dunn
,
J. Biomed. Opt.
15
,
066030
(
2010
).
30.
C. C.
Lay
,
M. F.
Davis
,
C. H.
Chen-Bee
, and
R. D.
Frostig
,
PLoS One
5
,
e11270
(
2010
).
31.
P.
Zakharov
,
A. C.
Völker
,
M. T.
Wyss
,
F.
Haiss
,
N.
Calcinaghi
,
C.
Zunzunegui
,
A.
Buck
,
F.
Scheffold
, and
B.
Weber
,
Opt. Express
17
,
13904
(
2009
).
32.
A.
Azari-Anderson
,
F. A.
Chowdhry
,
Y.
Ardeshirpour
,
N. S.
Karamzadeh
,
E. G.
Smith
,
V.
Chernomordik
, and
A. H.
Gandjbakhche
, in
Neurophotonics and Brain Mapping
, 1st ed., edited by
Y.
Chen
and
B.
Kateb
(
Taylor & Francis
,
Boca Raton
,
2017
), p.
41
.
33.
B.
Choi
,
W.
Tan
,
W.
Jia
,
S. M.
White
,
W. J.
Moy
,
B. Y.
Yang
,
J.
Zhu
,
Z.
Chen
,
K. M.
Kelly
, and
J. S.
Nelson
,
IEEE J. Sel. Top. Quantum Electron.
22
,
307
(
2016
).
34.
R.
Farraro
,
O.
Fathi
, and
B.
Choi
,
J. Biomed. Opt.
21
,
94001
(
2016
).
35.
C.
Lal
,
A.
Banerjee
, and
N. U.
Sujatha
,
J. Biomed. Opt.
18
,
111419
(
2013
).
36.
M. A.
Davis
,
S. M.
Kazmi
, and
A. K.
Dunn
,
J. Biomed. Opt.
19
,
086001
(
2014
).
37.
M.
Takahashi
,
T.
Urushihata
,
H.
Takuwa
,
K.
Sakata
,
Y.
Takado
,
E.
Shimizu
,
T.
Suhara
,
M.
Higuchi
, and
H.
Ito
,
Front. Neurosci.
11
,
723
(
2017
).
38.
J. G.
Fujimoto
,
M. E.
Brezinski
,
G. J.
Tearney
,
S. A.
Boppart
,
B.
Bouma
,
M. R.
Hee
,
J. F.
Southern
, and
E. A.
Swanson
,
Nat. Med.
1
,
970
(
1995
).
39.
J. G.
Fujimoto
,
Nat. Biotechnol.
21
,
1361
(
2003
).
40.
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
,
Science
254
,
1178
(
1991
).
41.
W.
Drexler
,
M.
Liu
,
A.
Kumar
,
T.
Kamali
,
A.
Unterhuber
, and
R. A.
Leitgeb
,
J. Biomed. Opt.
19
,
071412
(
2014
).
42.
J. G.
Fujimoto
and
W.
Drexler
, in
Optical Coherence Tomography: Technology and Applications
, 2nd ed., edited by
W.
Drexler
and
J. G.
Fujimoto
(
Springer International Publishing
,
Cham
,
2015
), p.
3
.
43.
Z.
Chen
,
T. E.
Milner
,
S.
Srinivas
,
X.
Wang
,
A.
Malekafzali
,
M. J.
van Gemert
, and
J. S.
Nelson
,
Opt. Lett.
22
,
1119
(
1997
).
44.
C. L.
Chen
and
R. K.
Wang
,
Biomed. Opt. Express
8
,
1056
(
2017
).
45.
Y.
Akamatsu
,
Y.
Nishijima
,
C. C.
Lee
,
S. Y.
Yang
,
L.
Shi
,
L.
An
,
R. K.
Wang
,
T.
Tominaga
, and
J.
Liu
,
J. Neurosci.
35
,
3851
(
2015
).
46.
L.
Qi
,
J.
Zhu
,
A. M.
Hancock
,
C.
Dai
,
X.
Zhang
,
R. D.
Frostig
, and
Z.
Chen
,
Biomed. Opt. Express
7
,
601
(
2016
).
47.
J. C.
Lee
,
B. J.
Wong
,
O.
Tan
,
S.
Srinivas
,
S. R.
Sadda
,
D.
Huang
, and
A. A.
Fawzi
,
Invest. Ophthalmol. Visual Sci.
54
,
6104
(
2013
).
48.
O.
Tan
,
Y.
Wang
,
R. K.
Konduru
,
X.
Zhang
,
S. R.
Sadda
, and
D.
Huang
,
J. Vis. Exp.
67
,
e3524
(
2012
).
49.
S.
Yazdanfar
,
A. M.
Rollins
, and
J. A.
Izatt
,
Opt. Lett.
25
,
1448
(
2000
).
50.
S.
Huang
,
Z.
Piao
,
J.
Zhu
,
F.
Lu
, and
Z.
Chen
,
J. Biomed. Opt.
20
,
76003
(
2015
).
51.
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
,
Ophthalmology
122
,
2532
(
2015
).
52.
L.
Yu
and
Z.
Chen
,
J. Biomed. Opt.
15
,
016029
(
2010
).
53.
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
,
Retina
36
(
Suppl. 1
),
S118
(
2016
).
54.
M. R.
Munk
,
H.
Giannakaki-Zimmermann
,
L.
Berger
,
W.
Huf
,
A.
Ebneter
,
S.
Wolf
, and
M. S.
Zinkernagel
,
PLoS One
12
,
e0177059
(
2017
).
55.
L.
An
,
J.
Qin
, and
R. K.
Wang
,
Opt. Express
18
,
8220
(
2010
).
56.
Y.
Zhao
,
Z.
Chen
,
C.
Saxer
,
Q.
Shen
,
S.
Xiang
,
J. F.
de Boer
, and
J. S.
Nelson
,
Opt. Lett.
25
,
1358
(
2000
).
57.
Y.
Zhao
,
Z.
Chen
,
C.
Saxer
,
S.
Xiang
,
J. F.
de Boer
, and
J. S.
Nelson
,
Opt. Lett.
25
,
114
(
2000
).
58.
G.
Liu
,
W.
Jia
,
V.
Sun
,
B.
Choi
, and
Z.
Chen
,
Opt. Express
20
,
7694
(
2012
).
59.
G.
Liu
,
W.
Jia
,
J. S.
Nelson
, and
Z.
Chen
,
Lasers Surg. Med.
45
,
628
(
2013
).
60.
U.
Baran
,
W. J.
Choi
, and
R. K.
Wang
,
Skin Res. Technol.
22
,
238
(
2016
).
61.
H.
Ren
,
K. M.
Brecke
,
Z.
Ding
,
Y.
Zhao
,
J. S.
Nelson
, and
Z.
Chen
,
Opt. Lett.
27
,
409
(
2002
).
62.
D. Y.
Kim
,
J.
Fingler
,
J. S.
Werner
,
D. M.
Schwartz
,
S. E.
Fraser
, and
R. J.
Zawadzki
,
Biomed. Opt. Express
2
,
1504
(
2011
).
63.
J.
Fingler
,
D.
Schwartz
,
C.
Yang
, and
S. E.
Fraser
,
Opt. Express
15
,
12636
(
2007
).
64.
A.
Zhang
,
Q.
Zhang
,
C. L.
Chen
, and
R. K.
Wang
,
J. Biomed. Opt.
20
,
100901
(
2015
).
65.
S.
Makita
,
T.
Fabritius
, and
Y.
Yasuno
,
Opt. Lett.
33
,
836
(
2008
).
66.
Y.
Wang
,
B. A.
Bower
,
J. A.
Izatt
,
O.
Tan
, and
D.
Huang
,
J. Biomed. Opt.
12
,
041215
(
2007
).
67.
J.
You
,
C.
Du
,
N. D.
Volkow
, and
Y.
Pan
,
Biomed. Opt. Express
5
,
3217
(
2014
).
68.
C. J.
Pedersen
,
D.
Huang
,
M. A.
Shure
, and
A. M.
Rollins
,
Opt. Lett.
32
,
506
(
2007
).
69.
V. J.
Srinivasan
,
S.
Sakadzić
,
I.
Gorczynska
,
S.
Ruvinskaya
,
W.
Wu
,
J. G.
Fujimoto
, and
D. A.
Boas
,
Opt. Express
18
,
2477
(
2010
).
70.
G.
Liu
,
L.
Chou
,
W.
Jia
,
W.
Qi
,
B.
Choi
, and
Z.
Chen
,
Opt. Express
19
,
11429
(
2011
).
71.
J.
Zhu
,
Y.
Miao
,
L.
Qi
,
Y.
Qu
,
Y.
He
,
Q.
Yang
, and
Z.
Chen
,
Appl. Phys. Lett.
110
,
201101
(
2017
).
72.
J.
Zhu
,
L.
Qi
,
Y.
Miao
,
T.
Ma
,
C.
Dai
,
Y.
Qu
,
Y.
He
,
Y.
Gao
,
Q.
Zhou
, and
Z.
Chen
,
Sci. Rep.
6
,
35499
(
2016
).
73.
J.
Zhu
,
Y.
Qu
,
T.
Ma
,
R.
Li
,
Y.
Du
,
S.
Huang
,
K. K.
Shung
,
Q.
Zhou
, and
Z.
Chen
,
Opt. Lett.
40
,
2099
(
2015
).
74.
Y.
Jia
,
O.
Tan
,
J.
Tokayer
,
B.
Potsaid
,
Y.
Wang
,
J. J.
Liu
,
M. F.
Kraus
,
H.
Subhash
,
J. G.
Fujimoto
,
J.
Hornegger
, and
D.
Huang
,
Opt. Express
20
,
4710
(
2012
).
75.
A.
Mariampillai
,
B. A.
Standish
,
E. H.
Moriyama
,
M.
Khurana
,
N. R.
Munce
,
M. K.
Leung
,
J.
Jiang
,
A.
Cable
,
B. C.
Wilson
,
I. A.
Vitkin
, and
V. X.
Yang
,
Opt. Lett.
33
,
1530
(
2008
).
76.
A.
Mariampillai
,
M. K.
Leung
,
M.
Jarvi
,
B. A.
Standish
,
K.
Lee
,
B. C.
Wilson
,
A.
Vitkin
, and
V. X.
Yang
,
Opt. Lett.
35
,
1257
(
2010
).
77.
Y.
Huang
,
Q.
Zhang
,
M.
Thorell
,
L.
An
,
M.
Durbin
,
M.
Laron
,
U.
Sharma
,
G.
Gregori
,
P.
Rosenfeld
, and
R.
Wang
,
Ophthalmic Surg., Lasers Imaging Retina
45
,
382
(
2014
).
78.
Z.
Zhi
,
W.
Cepurna
,
E.
Johnson
,
T.
Shen
,
J.
Morrison
, and
R. K.
Wang
,
Biomed. Opt. Express
2
,
579
(
2011
).
79.
I.
Gorczynska
,
J. V.
Migacz
,
R. J.
Zawadzki
,
A. G.
Capps
, and
J. S.
Werner
,
Biomed. Opt. Express
7
,
911
(
2016
).
80.
M. S.
Mahmud
,
D. W.
Cadotte
,
B.
Vuong
,
C.
Sun
,
T. W.
Luk
,
A.
Mariampillai
, and
V. X.
Yang
,
J. Biomed. Opt.
18
,
50901
(
2013
).
81.
G.
Liu
,
A. J.
Lin
,
B. J.
Tromberg
, and
Z.
Chen
,
Biomed. Opt. Express
3
,
2669
(
2012
).
82.
J.
Zhu
,
B.
Zhang
,
L.
Qi
,
L.
Wang
,
Q.
Yang
,
Z.
Zhu
,
T.
Huo
, and
Z.
Chen
,
Appl. Phys. Lett.
111
,
181101
(
2017
).
83.
L.
Wang
,
Y.
Wang
,
S.
Guo
,
J.
Zhang
,
M.
Bachman
,
G.
Li
, and
Z.
Chen
,
Opt. Commun.
242
,
345
(
2004
).
84.
L.
Yu
,
E.
Nguyen
,
G.
Liu
,
B.
Choi
, and
Z.
Chen
,
J. Biomed. Opt.
15
,
066006
(
2010
).
85.
C.
Wang
,
Y.
Yang
,
Z.
Ding
,
J.
Meng
,
K.
Wang
,
W.
Yang
, and
Y.
Xu
,
J. Biomed. Opt.
16
,
046001
(
2011
).
86.
H.
Ren
,
C.
Du
,
Z.
Yuan
,
K.
Park
,
N. D.
Volkow
, and
Y.
Pan
,
Mol. Psychiatry
17
,
1017
(
2012
).
87.
W.
Chen
,
J.
You
,
X.
Gu
,
C.
Du
, and
Y.
Pan
,
Sci. Rep.
6
,
38786
(
2016
).
88.
G.
Liu
and
Z.
Chen
,
Chin. Opt. Lett.
11
,
011702
(
2013
).
89.
A. J.
Lin
,
G.
Liu
,
N. A.
Castello
,
J. J.
Yeh
,
R.
Rahimian
,
G.
Lee
,
V.
Tsay
,
A. J.
Durkin
,
B.
Choi
,
F. M.
LaFerla
,
Z.
Chen
,
K. N.
Green
, and
B. J.
Tromberg
,
Neurophotonics
1
,
011005
(
2014
).
90.
U.
Baran
,
Y.
Li
, and
R. K.
Wang
,
Neurophotonics
2
,
025006
(
2015
).
91.
Y.
Li
,
J.
Jing
,
E.
Heidari
,
J.
Zhu
,
Y.
Qu
, and
Z.
Chen
,
Sci. Rep.
7
,
14525
(
2017
).
92.
Y.
Li
,
N. T.
Sudol
,
Y.
Miao
,
J. C.
Jing
,
J.
Zhu
,
F.
Lane
, and
Z.
Chen
, “
1.7 micron optical coherence tomography for vaginal tissue characterization in vivo
,”
Lasers Surg. Med.
(published online).
93.
R. F.
Spaide
,
J. G.
Fujimoto
, and
N. K.
Waheed
,
Retina
35
,
2163
(
2015
).
94.
G.
Liu
,
W.
Qi
,
L.
Yu
, and
Z.
Chen
,
Opt. Express
19
,
3657
(
2011
).
95.
Q.
Zhang
,
Y.
Huang
,
T.
Zhang
,
S.
Kubach
,
L.
An
,
M.
Laron
,
U.
Sharma
, and
R. K.
Wang
,
J. Biomed. Opt.
20
,
066008
(
2015
).
96.
B.
Braaf
,
K. V.
Vienola
,
C. K.
Sheehy
,
Q.
Yang
,
K. A.
Vermeer
,
P.
Tiruveedhula
,
D. W.
Arathorn
,
A.
Roorda
, and
J. F.
de Boer
,
Biomed. Opt. Express
4
,
51
(
2013
).
97.
T.
Klein
and
R.
Huber
,
Biomed. Opt. Express
8
,
828
(
2017
).
98.
Z.
Zhi
,
W.
Qin
,
J.
Wang
,
W.
Wei
, and
R. K.
Wang
,
Opt. Lett.
40
,
1779
(
2015
).
99.
B.
Potsaid
,
I.
Gorczynska
,
V. J.
Srinivasan
,
Y.
Chen
,
J.
Jiang
,
A.
Cable
, and
J. G.
Fujimoto
,
Opt. Express
16
,
15149
(
2008
).
100.
T.
Bonin
,
G.
Franke
,
M.
Hagen-Eggert
,
P.
Koch
, and
G.
Hüttmann
,
Opt. Lett.
35
,
3432
(
2010
).
101.
D. J.
Fechtig
,
B.
Grajciar
,
T.
Schmoll
,
C.
Blatter
,
R. M.
Werkmeister
,
W.
Drexler
, and
R. A.
Leitgeb
,
Biomed. Opt. Express
6
,
716
(
2015
).
102.
C.
Pfäffle
,
H.
Spahr
,
D.
Hillmann
,
H.
Sudkamp
,
G.
Franke
,
P.
Koch
, and
G.
Hüttmann
,
Biomed. Opt. Express
8
,
1499
(
2017
).
103.
K.
Zhang
and
J. U.
Kang
,
Opt. Express
18
,
23472
(
2010
).
104.
D. H.
Choi
,
H.
Hiro-Oka
,
K.
Shimizu
, and
K.
Ohbayashi
,
Biomed. Opt. Express
3
,
3067
(
2012
).
105.
R. M.
Werkmeister
,
S.
Sapeta
,
D.
Schmidl
,
G.
Garhöfer
,
G.
Schmidinger
,
V.
Aranha Dos Santos
,
G. C.
Aschinger
,
I.
Baumgartner
,
N.
Pircher
,
F.
Schwarzhans
,
A.
Pantalon
,
H.
Dua
, and
L.
Schmetterer
,
Biomed. Opt. Express
8
,
1221
(
2017
).
106.
X.
Shu
,
L.
Beckmann
, and
H.
Zhang
,
J. Biomed. Opt.
22
,
1
(
2017
).
107.
Y.
Pan
,
J.
You
,
N. D.
Volkow
,
K.
Park
, and
C.
Du
,
Neuroimage
103
,
492
(
2014
).