Three-Dimensional Visualization of Ocular Vascular Pathology by Optical Coherence Angiography In Vivo

Abstract

Purpose.: To demonstrate the clinical application of a noninvasive, three-dimensional, vascular imaging technique called Doppler optical coherence angiography (OCA). To evaluate the vascular architecture of polypoidal choroidal vasculopathy (PCV) using Doppler OCA.

Methods.: The authors prospectively examined the eyes of four healthy subjects and 15 PCV patients. Three-dimensional vascular flow imaging was performed using high-speed, high-resolution, and high-penetration spectral-domain Doppler optical coherence tomography. Two-dimensional images of the retina, choroid, and vascular lesions were obtained simultaneously.

Results.: Distribution of blood flow detected by Doppler OCA imaging corresponded well with that by indocyanine angiographic imaging. PCV lesions were localized in the space between the retinal pigment epithelium and the Bruch's membrane.

Conclusions.: The authors found using Doppler OCA that PCV lesions are similar in architecture to choroidal neovascularization in age-related macular degeneration. Doppler OCA facilitates rapid and noninvasive examination of exudative macular diseases.

Exudative macular diseases include polypoidal choroidal vasculopathy (PCV) and age-related macular degeneration, which are the major causes of severe loss of central vision. ¹ PCV, which was first described as a distinct clinical entity by Yannuzzi et al., ² is characterized by numerous, recurrent, bilateral, asymmetric, and serosanguineous detachments in the retinal pigment epithelium (RPE). In clinical studies of PCV, polypoidal vascular lesions and branching vascular networks in the inner choroid were described as characteristic findings of PCV ^{3 –8} ; however, the origin and location of these lesions remain controversial. ^6,9 Some studies speculate that the lesions are located in the sub-RPE and represent a type of choroidal neovascularization (CNV), ^7,10,11 whereas others speculate that they are located in the choroid and represent pathologic changes in the choroidal vasculature. ^12,13 To verify these speculations, determination of the precise location and examination of the three-dimensional (3D) structure of PCV lesions are required. 

In clinical practice, fluorescein angiography (FA) and indocyanine-green angiography (ICGA) are the chief diagnostic modalities used to evaluate PCV lesions. ^3,4,6,9 However, FA and ICGA poorly resolve the depth of these lesions. Occasionally, FA and ICGA are based on scanning laser ophthalmoscopy and inherently enable depth resolution because of the confocal effect. However, their axial resolution is limited to 300 μm. ¹⁴ The poor axial resolution of these devices impedes evaluation of the 3D structure of the PCV lesions. 

Optical coherence tomography (OCT) generates cross-sectional images by measuring the echo time delay and magnitude of backscattered or back-reflected light. ¹⁵ OCT has achieved micrometer-level axial resolution in cross-sectional retinal imaging. Commercial OCT, using a light source of 840 nm, was an essential diagnostic tool for the clinical evaluation of PCV lesions. ^10,12,16,17 The alternative wavelength for retinal OCT is 1.0 μm, which allows for deeper penetration into tissue. ^{18 –21} OCT using a 1.0-μm band light source, the so-called high-penetration OCT, ²² allows for better visualization of the PCV lesions beneath the RPE. ²²  

OCT could provide important information about PCV lesions ^{10,12,16,22,23} ; however, conventional OCT, including high-penetration OCT, has an inherent weakness for evaluating vascular lesions. Conventional OCT, which we refer to as scattering OCT, is only sensitive to backscattering light intensity and cannot detect blood flow information. Because of this limitation, conventional scattering OCT cannot discriminate vascular lesions from the surrounding tissues; hence, the precise location of the PCV lesions cannot be determined. One alternative to scattering OCT is Doppler OCT, an extension of OCT that is capable of measuring the Doppler shifts arising from blood flow. ^24,25 This technique provides depth-resolved, cross-sectional images of the retina and choroid while preserving blood flow-derived information. ^26,27 Two different applications of Doppler OCT were reported for evaluation of retinal circulation. One application was objective evaluation of blood flow and flow volume. ^28,29 These studies showed that, using the data derived from the Doppler OCT, absolute retinal blood flow speed and volume could be calculated in the human retina. ^28,29 Another application was 3D ocular vascular imaging. ^{30 –33} Using Doppler signals as a contrast source, scattering OCT could noninvasively pinpoint blood flow location. High-speed Doppler OCT imaging that enables 3D visualization of the chorioretinal vasculature is termed Doppler optical coherence angiography (OCA). ³⁰  

Doppler OCA may facilitate comprehensive evaluation of the PCV lesions. In this study, we present the first clinical application of Doppler OCA and describe the usefulness of high-penetration Doppler OCT for evaluating 3D vascular architecture in exudative macular diseases. 

Subjects and Methods

Subjects

As controls, we examined four eyes of four healthy Japanese volunteers (four men; age range, 28–49 years; mean age, 35.0 years). For PCV, we evaluated 15 eyes of 15 Japanese patients with PCV (13 men, two women; age range, 60–86 years; mean age,71.9 years). Eyes with severe cataracts or other eye diseases that interfered with Doppler OCT image quality were excluded from this study. The clinical diagnosis of PCV was made by identification of orange-red subretinal lesions by fundus examination and polypoidal lesions in ICGA. All eyes of patients with PCV had pigment epithelium detachments (PEDs). Among them, four eyes had hemorrhagic PEDs, and six eyes had subretinal hemorrhage. 

All experiments were performed according to the tenets of the Declaration of Helsinki and were approved by the Institutional Review Boards of the University of Tsukuba and Tokyo Medical University. Informed consent for the examination was obtained from all participants. 

Doppler OCA

The prototype Doppler spectral-domain OCT system was built by the Computational Optic Group at the University of Tsukuba. This system is based on the Michelson interferometer, as shown in Figure 1. The light source is a superluminescent diode with a central wavelength of 1020 nm and a bandwidth of 100 nm (Superlum, Carrigtwohill, Ireland). A high-speed indium gallium arsenide line-scanning camera with 1024 pixels (Goodrich, Charlotte, NC) was used as the detection system. The measurement speed was 47,000 depth-scans/s, and the depth resolution was measured to be 4.3 μm deep in the tissue. The interferometer was attached to a semi-custom fundus-scanning head based on a 3D OCT system (OCT-1000; Topcon Corp., Tokyo, Japan). A raster-scanning protocol with 1500 depth-scans × 128 B-scans covering a 6.0 × 6.0-mm region on the retina was used for volumetric scans. The scanning depth was 1.3 mm, and the acquisition speed of each measurement was 4.1 second/volume. 

Figure 1.

 

Schematic diagram of the optical coherence tomography imaging system. VND, variable neutral density filter; P, polarizer; PC, polarization controller; SLD, superluminescent diode; InGaAs, indium gallium arsenide.

The Doppler shift of OCT signals was calculated using the phase difference between the adjacent depth-scans where the phase difference was obtained by complex division of adjacent depth-scans. The time interval between the two sequential depth-scans was 21.3 μs, and the integration time of the line camera was 13.9 μs. According to these parameters, the maximum detectable Doppler shift was defined as 23.5 kHz, and the corresponding axial velocity component in the eye was 8.68 mm/s. The dominant factor, which limits the minimum detectable axial velocity, is the ratio of spatial sampling spacing to the beam-spot diameter on tissues. ³⁴ The minimum detectable axial velocity was expected to be 0.44 mm/s. 

The bidirectional Doppler image is displayed by a gray color code in which positive flow (from posterior to anterior) is displayed as white and negative flow (anterior to posterior) is displayed as black. The Doppler signals at pixels, which have smaller structural OCT signal intensity than a predefined threshold, are displayed as null velocity because the Doppler signal in these pixels are mostly noise. The color Doppler image was created by overlaying a bidirectional Doppler signal on its corresponding structural OCT. The Doppler signal is color-coded by red-blue color maps, in which positive flow is displayed as red and negative flow is displayed as blue. 

To visualize retinal, choroidal, and sub-RPE regions in different colors, the 3D volume is segmented both automatically and manually. For healthy eyes, the volume is automatically segmented to the retina and choroid. Here the retina is defined as the region between the internal limiting membrane and the RPE, and the choroid is defined as the region beneath the Bruch's membrane. For eyes with PCV, the space between the RPE and the Bruch's membrane was manually segmented. 

For 3D visualization, the ocular axial motion during the acquisition was compensated by the following methods. ³⁰ Inter-B-scanning motion was detected and canceled by using a fast correlation-based algorithm. For 3D visualization of Doppler OCA, the power of the Doppler shift was volume-rendered. Here the power of the Doppler shift was defined as the squared power of the bidirectional Doppler signal. 

Results

Healthy Eyes

Figures 2, 3, and 4 show OCA images of a representative subject, a 49-year-old healthy man. Figures 2B to 2D show the cross-sectional structural and Doppler OCT images. Figure 2B shows the structural OCT in which the backscattering property of the retina is visualized and is identical with conventional OCT. In the structural OCT, all retinal layers typical of high-resolution OCT are observed. ^35,36 In the bidirectional Doppler OCT image (Fig. 2C) and the color Doppler OCT images (Fig. 2D), correspondence between the locations of vessels and blood flow is clearly seen. 

Figure 2.

 

Doppler OCT images of the right eye of a healthy 49-year-old man. An en face projection OCT image (A) enables precise registration of OCT data with the ICGA image. White line: scanning line of B-scan OCT images (B–D). A color Doppler OCT image (D) was created from the intensity OCT image (B) and the bidirectional Doppler OCT image (C). Doppler signal in the retina (arrow) and the choroid (arrowhead) can be clearly seen (C, D).

Figure 3.

 

Stereoscopic view of the three-dimensional Doppler OCA image of a healthy eye in Figure 2. A pair of images is presented for crossed-eye viewing. The distribution of the Doppler signals from the retina (white) and the choroid (green) are displayed.

Figure 4.

 

Comparison of the Doppler OCA image and the ICGA image. A projection image of Doppler OCA was registered to the arterial phase of the ICGA image (A) using an individual color code (retina, blue; choroid, green). White line: area of Doppler OCA imaging. Doppler OCA image (B) corresponds well with the arterial phase of the ICGA image (A).

For a more comprehensive and detailed investigation, 3D vascular imaging was performed in which the eye was scanned by a raster-scanning protocol for 4.1 seconds. The squared power of the bidirectional Doppler signal of the measured blood volume was then volume-rendered as shown in Figure 3. For better understanding, the measured 3D volume was automatically segmented into the retina and the choroid, each segment was rendered using different colors (white for retina and green for choroid), and the 3D vascular architecture was readily and intuitively observed. 

An additional benefit of 3D investigation is that two-dimensional en face projections can be created from the 3D volume. An en face projection of structural OCT is known to reveal a clear retinal vascular pattern (Fig. 2A). Furthermore, in the case of OCA, we can create an en face projection of Doppler signals that is comparable to conventional angiographies. In the example shown in Figure 4, the en face projection of squared intensity of bidirectional Doppler signal of the retina (blue) and the choroid (green) are overlaid on an arterial-phase ICGA. Coregistration of the Doppler signal with the retinal/choroidal arteries is clearly observed. These findings were reproducible in all four healthy eyes. 

PCV

Images in Figures 5, 6, and 7 were obtained from the right eye of a 70-year-old man. Fundus color photographs showed multiple PEDs with orange-red spots in the macula (Fig. 5A). In the ICGA images, branching vascular networks overlying the large choroidal vessels were observed in the early phase (Fig. 5B), whereas polypoidal lesions adjacent to the vascular network were readily observed in the late phase (Fig. 5C). Structural OCT at the center of the branching vascular network showed undulation of the RPE line in the area of the branching vascular network in ICGA (Fig. 5D). Beneath the undulating RPE line, a hyperreflective smooth line representing Bruch's membrane was clearly visible. The space between the undulating RPE line and Bruch's membrane was filled with a patchy, moderately hyperreflective mass. This hyperreflective mass was thought to represent PCV lesions in the study using conventional structural OCT. ^22,23 However, structural OCT cannot provide any evidence of the existence of abnormal vasculature. The color Doppler OCT image (Fig. 5E) revealed the presence of blood flow at the hyperreflective mass between the RPE and Bruch's membrane. 

Figure 5.

 

Color fundus, ICGA, and OCT images of PCV obtained from the right eye of a 70-year-old man. Color fundus photographs (A) show multiple PEDs with orange-red spots (white arrows) in the macula. In the ICGA images, branching vascular networks overlying the large choroidal vessels (white arrow) were evident in the early phase (B), whereas polypoidal lesions adjacent to the vascular network (white arrows) were readily seen in the late phase (C). Black line: area of Doppler OCT imaging; white line: scanning line of B-scan OCT images (D, E). The intensity B-scan OCT image (D) shows the space between the RPE and Bruch's membrane. Bidirectional color Doppler B-scan OCT image (E) shows the presence of blood flow at the hyperreflective mass between the RPE and Bruch's membrane.

Figure 6.

 

Stereoscopic view of the three-dimensional Doppler OCA image of an eye affected by PCV in Figure 5. A pair of images is presented for crossed-eye viewing. The distribution of the Doppler signals from the retina, choroid, and the space between the RPE and Bruch's membrane are color coded with white, green, and red, respectively.

Figure 7.

 

Doppler OCA image from the eye affected by PCV in Figures 5 and 6 was superimposed on the CGA image using individual color codes (retina, blue; choroid, green; space between RPE and Bruch's membrane, red). Black line: area of Doppler OCA imaging. Distribution of the Doppler signals in the space between the RPE and Bruch's membrane corresponds well with the area of abnormal vascular network in the ICGA image.

The 3D structure of PCV lesions was examined by Doppler OCA. For layer-by-layer analysis, the internal limiting membrane line was automatically segmented, whereas the RPE line and Bruch's membrane line were manually segmented. Based on this segmentation, the Doppler OCA volume was separated into three parts: retina (from the internal limiting membrane to the RPE), the space between the RPE and Bruch's membrane (from the RPE to Bruch's membrane), and the choroid (beneath Bruch's membrane). Figure 6 shows stereoscopic volume-rendering of the Doppler OCA volume where the Doppler signals of the retina, choroid, and the space between the RPE and Bruch's membrane were color-coded with, respectively, white, green, and red. En face projection of the Doppler OCA was overlaid on ICGA, as shown in Figure 7, where the Doppler signals at the retina, choroid, and RPE–Bruch's membrane space are represented, respectively, by blue, green, and red. Distribution of the Doppler signals in the space between the RPE and Bruch's membrane corresponded well with the area of abnormal vascular networks in ICGA images. 

The images in Figures 8 and 9 were obtained from the right eye of a 69-year-old woman and the left eye of a 77-year-old man, respectively. In these images, the same color codes as in Figures 5 and 7 were used. In these cases, the vessels in the branching vascular network were narrower than first case (Figs. 8B, 9B). Despite the narrow vessels of the vascular network, the presence of blood flow in the space between the RPE and Bruch's membrane was confirmed in color Doppler OCT images (Figs. 8E, 9E). Distribution of the Doppler signals in the space between the RPE and Bruch's membrane corresponded well with the area of abnormal vascular networks in the ICGA images (Figs. 8F, 9F). This Doppler OCA finding indicates the presence of PCV lesions between the RPE and Bruch's membrane, a finding that was confirmed in all 15 eyes of patients with PCV. 

Figure 8.

 

Color fundus, ICGA, OCT, and OCA images of PCV obtained from the right eye of a 69-year-old woman. Color fundus photographs (A) show multiple PEDs with orange-red spots (white arrow) in the macula. In the ICGA images, branching vascular networks overlying the large choroidal vessels (white arrow) are evident in the early phase (B), and polypoidal lesions adjacent to the vascular network (white arrows) are readily seen in the late phase (C). Black line: area of Doppler OCT imaging; white line: scanning line of B-scan OCT images (D, E). An intensity B-scan OCT image (D) shows the space between the RPE and Bruch's membrane. A bidirectional color Doppler B-scan OCT image (E) shows the presence of blood flow at the hyperreflective mass between the RPE and Bruch's membrane. A Doppler OCA image superimposed on the ICGA images (F) shows the distribution of the Doppler signals in the space between the RPE and Bruch's membrane; this distribution corresponds well with the area of the abnormal vascular network in the ICGA image. The same color codes as in Figures 7 were used in the Doppler OCA image. Black line: area of Doppler OCA imaging.

Figure 9.

 

Color fundus, ICGA, OCT, and OCA images of PCV obtained from the left eye of a 77-year-old man. Color fundus photographs (A) show multiple PEDs with subretinal hemorrhage. An orange-red spot (white arrow) is observed in the macula. In the ICGA images, branching vascular networks overlying the large choroidal vessels (white arrow) are evident in the early phase (B), and polypoidal lesions adjacent to the vascular network (white arrows) are readily seen in the late phase (C). Black line: area of Doppler OCT imaging; white line: scanning line of B-scan OCT images (D, E). An intensity B-scan OCT image (D) shows the space between the RPE and Bruch's membrane. Bidirectional color Doppler B-scan OCT image (E) shows the presence of blood flow in the space between the RPE and Bruch's membrane. Doppler OCA image superimposed on ICGA images (F) shows the distribution of the Doppler signals in the space between the RPE and Bruch's membrane; this distribution corresponds well with the area of the abnormal vascular network in the ICGA image. The same color codes as in Figure 7 were used in the Doppler OCA image. Black line: area of Doppler OCA imaging.

Discussion

In this study, we used Doppler OCA imaging to investigate vascular architecture. Doppler OCA provides 3D architectural information of the retinal and choroid vasculature and corresponds well with the ICGA images. In eyes affected by PCV, Doppler signals were detected in the space between the RPE and Bruch's membrane, representing the presence of blood flow in this space. The distribution of this blood flow corresponded to the abnormal vascular networks seen in the ICGA images. 

There is controversy surrounding the 3D structure of PCV lesions. ^6,7 Histopathologic studies showed that PCV lesions were localized beneath ³⁷ or inside ¹¹ Bruch's membrane. Although histopathologic studies were conducted using a surgically removed fragmentary sample ¹³ or of end-stage PCV with severe complications, ^11,37 comprehensive analysis of the 3D structure of PCV lesions could not be achieved. Doppler OCA revealed the presence of branching vascular networks in the space between the RPE and Bruch's membrane. This finding supported the previous studies with scattering OCT. ^{10,16,17,23,38} Although this does not necessarily mean that all PCV lesions exist only between the RPE and Bruch's membrane, there is no doubt that the space between Bruch's membrane and the RPE is crucial in PCV. This feature coincides with the vascular formation of type 1 CNV. ³⁹  

Some aspects of PCV lesions differ from those of typical CNV. Branching vascular networks distribute in a large area and sometimes extend beyond the vascular arcade. ⁴⁰ Branching vascular networks were accompanied by polyp formation. This wide distribution and polyp formation are uncommon findings in typical CNV. Further studies are necessary to evaluate the pathophysiological differences between PCV lesions and typical CNV lesions. 

This study revealed the clinical application of Doppler OCA in exudative macular diseases. Using Doppler signals as a contrast source, Doppler OCA enables the 3D visualization of the chorioretinal vasculature in vivo. Layer-by-layer and 3D volumetric analyses of the chorioretinal vascular architecture are possible using Doppler OCA imaging. These analyses could not be performed by conventional ocular vascular flow imaging modalities such as FA, ICGA, scanning laser Doppler flowmetry, ⁴¹ laser speckle photography, ⁴² or laser speckle flowgraphy. ⁴³ One great advantage of Doppler OCA is that it may be used to evaluate exudative macular diseases. Precise localization of vascular lesions is crucial to evaluate the pathophysiological features of PCV. Determining the depth of the CNV location is important for predicting prognoses and therapeutic effects. ³⁹  

Doppler OCA technology has the potential for further application. Doppler blood flow measurement only requires additional data analyses derived from conventional Fourier-domain OCT, and additional OCT hardware modifications are not necessary. Doppler OCT can measure the blood flow velocity and volume and potentially facilitate objective evaluation of blood flow in PCV lesions. In this study, Doppler OCA could detect only some parts of the choroidal or retinal vasculature; hence, ICGA and FA are still required to more thoroughly evaluate the entire structure of vascular lesions. However, the clinical applications of ICGA and FA have been limited because of the possibility of severe complications, ⁴⁴ patient discomfort, and relatively long measurement time. Doppler OCA is noninvasive and has a short measurement time (typically a few seconds; 4.1 seconds in this study). More sensitive Doppler OCA techniques are in development, ⁴⁵ and their introduction would enable even better visualization of the chorioretinal vasculature in the future. Doppler OCA may potentially function as a noninvasive alternative to FA and ICGA used noninvasively to assess macular diseases. 

Footnotes

 Supported in part by the Japan Science and Technology Agency through a program of the Development of Systems and Technology for Advanced Measurement and Analysis.

Footnotes

 Disclosure: M. Miura, None; S. Makita, Topcon (F), P; T. Iwasaki, None; Y. Yasuno, Topcon (F), P