Next Article in Journal
Localised Objective Characterisation Assessment of Lymphoedema (LOCAL): Using High-Frequency Ultrasound, Bioelectrical Impedance Spectroscopy and Volume to Evaluate Superficial Tissue Composition
Previous Article in Journal
Peripherical Blood hsa-miR-335-5p Quantification as a Prognostic, but Not Diagnostic, Marker of Gastric Cancer
Previous Article in Special Issue
The Surviving, Not Thriving, Photoreceptors in Patients with ABCA4 Stargardt Disease
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Diagnostics
Volume 14
Issue 15
10.3390/diagnostics14151615
diagnostics-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles thumb_up
...
Endorse textsms
...
Comment
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Compact Linear Flow Phantom Model for Retinal Blood-Flow Evaluation

by
Achyut J. Raghavendra
1,2,*,
Abdelrahman M. Elhusseiny
3,
Anant Agrawal
2,
Zhuolin Liu
2,
Daniel X. Hammer
2 and
Osamah J. Saeedi
1
1
Department of Ophthalmology and Visual Sciences, University of Maryland School of Medicine, Baltimore, MD 21201, USA
2
Center for Devices and Radiological Health (CDRH), U.S. Food and Drug Administration (FDA), Silver Spring, MD 20993, USA
3
Department of Ophthalmology, Harvey and Bernice Jones Eye Institute, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA
*
Author to whom correspondence should be addressed.
Diagnostics 2024, 14(15), 1615; https://doi.org/10.3390/diagnostics14151615
Submission received: 31 May 2024 / Revised: 12 July 2024 / Accepted: 17 July 2024 / Published: 26 July 2024
(This article belongs to the Special Issue High-Resolution Retinal Imaging: Hot Topics and Recent Developments)
Browse Figures
Review Reports Versions Notes

Abstract

:
Impaired retinal blood flow is associated with ocular diseases such as glaucoma, macular degeneration, and diabetic retinopathy. Among several ocular imaging techniques developed to measure retinal blood flow both invasively and non-invasively, adaptive optics (AO)-enabled scanning laser ophthalmoscopy (AO-SLO) resolves individual red blood cells and provides a high resolution with which to measure flow across retinal microvasculature. However, cross-validation of flow measures remains a challenge owing to instrument and patient-specific variability in each imaging technique. Hence, there is a critical need for a well-controlled clinical flow phantom for standardization and to establish blood-flow measures as clinical biomarkers for early diagnosis. Here, we present the design and validation of a simple, compact, portable, linear flow phantom based on a direct current motor and a conveyor-belt system that provides linear velocity tuning within the retinal microvasculature range (0.5–7 mm/s). The model was evaluated using a sensitive AO-SLO line-scan technique, which showed a <6% standard deviation from the true velocity. Further, a clinical SLO instrument showed a linear correlation with the phantom’s true velocity (r2 > 0.997). This model has great potential to calibrate, evaluate, and improve the accuracy of existing clinical imaging systems for retinal blood flow and aid in the diagnosis of ocular diseases with abnormal blood flow.

1. Introduction

Retinal blood flow is a crucial ocular biomarker for prevalent conditions that cause blindness worldwide. Dysregulation of retinal circulation has been linked to the pathogenesis of various ocular diseases, including age-related macular degeneration, glaucoma, and diabetic retinopathy [1,2,3,4]. Numerous studies have shown the association between retinal vascular alterations and systemic conditions such as hypertension, diabetes, stroke, and dementia [5,6]. Hanaguri and colleagues have demonstrated that retinal blood-flow regulation alterations may precede neural dysfunction in mice with type II diabetes mellitus [7].
Several ocular imaging techniques have been developed to measure retinal blood flow, such as optical coherence tomography angiography (OCTA), laser speckle imaging, laser Doppler velocimetry, laser Doppler flowmetry, and erythrocyte-mediated angiography (EMA) [8,9,10,11]. Among these, adaptive optics (AO)-enabled devices, which compensate for ocular aberrations and offer high lateral resolution, have enabled direct visualization of moving erythrocytes [12,13,14,15], the fine details of the vessel wall [4], and precise measurement of lumen diameter [16,17]. Burns et. al. first demonstrated an elegant method to capture red blood cell motion in space–time images using AO-SLO in line-scanning mode [18,19]. Bedggood and Metha described the use of an AO flood illumination system to assess erythrocyte velocity in retinal vasculature [13]. This approach enables the generation of high-contrast perfusion maps, offering detailed insights into the lumen of the retinal vascular network [20]. AO has also been combined with other imaging modalities, including OCT or EMA, to provide cellular-level resolution and additional blood-flow measurement capabilities [10,21]. Although more invasive, EMA enables the direct visualization of sparsely labeled indocyanine green (ICG)-loaded erythrocyte ghost cells using a clinical SLO instrument, providing precise quantification of ocular blood flow [22].
The quantification of blood-flow metrics from these imaging techniques often suffers from patient-specific variability and generalizability. Ocular imaging systems generally use the eye as an objective to visualize structures in the posterior segment. Consequently, flow rate measurements are influenced by normal variations in axial length, corneal curvature, and intraocular pressure, among other anatomical and physiological parameters [23,24,25]. Additionally, there is no universally accepted standard for acquiring absolute quantitative measures of blood flow, making it a challenge to compare results obtained across different ocular imaging devices.
Other major medical imaging modalities, such as X-ray computed tomography, magnetic resonance imaging, and ultrasound, use physical models known as “phantoms” to standardize system performance and optimize image quality. However, there is no accepted phantom for use in retinal vascular imaging. Several groups have developed retinal phantoms to assess structural imaging performance, mimicking the optical properties of the multi-layered retina [26,27,28]. For imaging flow, a phantom model has been created with microfluidic channels and an auxiliary pump in a circulation system to achieve various flow speeds [29]. However, their bulky setup, wet operation, and clogging issues in capillary-sized channels are potential disadvantages. A simple but reliable model is needed for the routine performance assessment of emerging clinical instruments that measure blood flow. Several groups have demonstrated that retinal blood-flow modelling demands multi-scale and multi-physics models to be physiologically realistic [30,31]. However, these theoretical predictions need validation with in vivo measurements, as demonstrated by Liu et al. [32]. Our work aims to aid in the functional assessment and calibration of in vivo imaging techniques.
In this study, we introduce a compact linear flow phantom based on a direct current (DC) motor and a conveyor-belt system. The phantom was validated on an AO-SLO device configured for line-scanning to measure retinal blood flow. Our flexible design allowed for the investigation of varying vessel angles with the AO-SLO line-scan technique. The phantom was also tested with a clinical SLO instrument configured for EMA to demonstrate its advantage in avoiding wet components and other complexities of alternate phantoms.

2. Materials and Methods

2.1. Phantom Fabrication

A photograph and schematic of the assembled linear flow phantom are shown in Figure 1a,c. A high-torque 12V DC motor (Fuzhou Bringsmart Intelligent Tech. Co., Ltd., Fuzhou, China) with encoder and reduction gearbox (1:600) is coupled with a 6 mm wide timing belt via a GT2 synchronous timing pulley (12 mm diameter with 20 teeth). A 20 mm or 30 mm focal length achromat lens mounted in front of the belt serves as a focusing element, mimicking the human eye axial length range. The optical axis is orthogonal to a manual rotation mount (RSP1, Thorlabs), which can rotate the belt from 0 to 90°, allowing for operation over the full range of retinal vessel orientations. An idle pulley is centered with the optical axis and maintains constant distance between the belt and achromatic lens. A motor housing (Figure 1b) and supporting pieces were 3D-printed with Ultimaker 3 (Ultimaker BV, Utrecht, The Netherlands). A DC motor controller (1065B, Phidget Inc., Calgary, AB, Canada) and custom LabVIEW program (2018, National Instruments, Austin, TX, USA) were used to control the input voltage signal to the DC motor and monitor the rotational speed in real time at 10 Hz. The linear belt speed could be adjusted precisely within the range of physiological microvasculature flow velocities in the retina (0.5–7 mm/s) [33].
To mimic flowing red blood cells, polystyrene microspheres with a 10 μm nominal diameter (Cat# 24294-2, Polysciences, Warrington, PA, USA) were deposited on a flexible plastic slide to form a uniform monolayer using a drop-casting technique within a humid chamber, as shown in the schematic in Figure 2a. Near-infrared (IR) fluorescent beads (Degradex PLGA microspheres, CAT# LGFN10K, Sigma-Aldrich, St. Louis, MO, USA) were added to the polystyrene microsphere suspension in various number ratios motivated by the bead concentration ratios used in EMA imaging [34]. Bead density measurement was performed by Fiji (ImageJ version 2.3.0/1.54f) analysis of brightfield and fluorescence microscopy images taken with a 10× objective. Brightfield and fluorescence microscopy were both required to quantify the relative density ratios of non-fluorescent and fluorescent beads. Bead density assessment included at least ten different regions within each coated slide. After verification of the deposited beads on the slide under the micrscope, it was affixed to the belt surface, bead-side up, with double-sided tape.

2.2. Phantom Calibration

The phantom was evaluated for bi-directional velocity linearity within the operational voltage (−12 V to +12 V) by observing the linear distance traveled by a marker on the belt with a PixelLink camera (PL-D7620MU-T) operating at 120 frames per second. Calibration curves were generated with the complete phantom assembly, which was used as the true linear velocity for all subsequent experiments. System repeatability was tested by measuring linear velocities for up to 5 s under continuous conveyor-belt operation.

2.3. Phantom Evaluation

The Food and Drug Administration multimodal AO system is described in detail elsewhere [35]. Only the AO-SLO channel in line-scan mode was used to evaluate the phantom. The line-scan mode produces alternate raster and space–time images by intermittently holding the slow galvanometer scanner stationary [18]. The AO-SLO channel uses a superluminiscent diode (SLD, λc = 756 nm, ∆λ = 20 nm) for imaging. Another SLD (λc = 830 nm, ∆λ = 60 nm) was used as the Shack–Hartmann Wavefront Sensor (SHWS) beacon. A 30 μm diameter pinhole (1.5 Airy disc diameter) in the SLO detection path was used to resolve the beads via the confocal AO-SLO approach. The phantom was mounted on a three-axis translation stage, and optical alignment was performed with an adjustable iris for optimal positioning. The plastic slide with beads offered a diffuse reflection necessary to create a uniform array of bright spots on the SHWS, which was essential to achieve optimum wavefront correction. The specular reflection from the plastic microscope slide surface, which overwhelms the bead signal, was avoided by adding a slight tilt to the phantom optical alignment. Focus was set on the bead monolayer above the belt/slide surface. With a 1.5° × 1.5° FOV (raster mode), 100 interleaved frames were collected at 27 Hz with a resonant scanner frequency of 13.6 kHz in AO-SLO line-scan mode. During acquisition, the phantom belt was rotated by angle α relative to the fast-scan direction, which mimics vessel orientation in the retina and is required to create measurable streaks using the line-scan technique. Three videos were collected for each experimental setup to calculate measurement repeatability. A custom MATLAB (MathWorks, Natick, MA, USA) program was developed to derive the slope of the streaks (Δx/Δt) caused by individual beads on space–time images [36]. The velocity of individual beads was then computed as follows:
v = (Δx/Δt) × sec(α),
To investigate the effect of varying values of α on velocity calculation, videos were collected for each rotational stage position between 0 to 90° in steps of 10°, with all other parameters kept constant. After each modification, care was taken to check for focal plane and optimum wavefront correction to resolve the beads. The known focal length of the lens in air (n = 1) was used to convert the scan angle to the lateral distance at the focal plane using paraxial approximation.

2.4. Clinical SLO Imaging

The Heidelberg Spectralis HRA 2 (Heidelberg Engineering, Heidelberg, Germany), a clinical confocal SLO, was used in indocyanine green angiography (ICGA) mode with the excitation at 785 nm and the emission at 805–840 nm, values typically used in the EMA technique [34]. The phantom was mounted and optically aligned on the chin rest. A five-second video was collected with 15° FOV at 15.4 fps for each experimental setting. Collected video frames were further analyzed in Fiji (ImageJ version 2.3.0/1.54f) software. Instrument scaling output (µm/pixel) was used for image calibration. Automated bead tracking was performed for the velocity analysis using TrackMate, a Fiji tool developed by Tinevez and colleagues [37]. The algorithm detects beads in every frame using a Hessian feature detector. Frame-to-frame bead tracking was performed with a simple linear assignment problem (LAP)-tracker algorithm with optimized boundary conditions [38]. The mean velocity was then computed for each video by averaging at least twenty individual particle traces from the TrackMate results. All statistical analyses were performed in Excel (Microsof 365, Microsoft, Redmond, WA, USA) including mean, standard deviation (SD), Pearson correlation coefficient (r), and regression analysis. The p-value threshold was set to 0.05 for statistical significance.

3. Results

3.1. Calibration Curves

Figure 3a shows the linearity of the velocity measured with varying motor voltages from −12 to 12 V, where r2 > 0.999 with no load (p < 0.0001). After phantom assembly with the DC motor, the linearity was slightly lower, with r2 > 0.989 (p < 0.0001). Note that the DC motor required at least ±0.5 V to begin moving. The repeatability test showed SD < 4.8% (n = 3) at 6 V.

3.2. AO-SLO Imaging

For phantom evaluation, videos were collected from the AO-SLO channel in line-scan mode at different motor voltages within the linear range (single AO-SLO full frame shown in Figure 4a). In line-scan mode, space–time images contain streaks from the moving beads as shown in Figure 4b. A summary of mean velocities for various phantom velocity settings calculated from Equation (1) is shown in Figure 4c. Linear regression analysis showed a slope = 1.041 and r2 value = 0.9983 (F1,3 = 1792; p < 0.05).
The phantom was further used to investigate the systematic effect of variations in vessel angles on AO-SLO line-scan accuracy. Using a constant belt velocity of 3.82 mm/s, we observed a non-linear increase in measured line-scan velocity above a 45° vessel angle, as shown in Figure 5a. Non-linear regression analysis with a cubic polynomial yielded the following best fit curve: y = 8.44 × 10−5x3 − 0.007x2 + 0.223x + 1.9 with r2 = 0.9934 (p < 0.05). Additionally, the observed streak length decreased with an increase in the vessel angle, as shown in Figure 5b.

3.3. Clinical SLO Imaging/EMA

To test our phantom with a clinical SLO instrument in ICGA mode, as used in EMA imaging, videos at 15.4 fps were collected while the phantom velocity was varied. Figure 6a shows a single SLO frame of the fluorescent beads with the phantom stationary. For videos collected at various phantom velocities, a maximum intensity projection of all video frames was taken to produce a single composite image, illustrating the bead motion in each video, as shown in Figure 6b. Phantom linear velocity thus creates clear trackable streaks from the fluorescent beads, where the distance between bead positions along the track can be used to compute the measured velocity. Results using TrackMate showed a linear relationship between the measured velocity and the true set phantom velocity with a slope = 0.9455 and r2 = 0.9975 (F1,3 = 1792; p < 0.05), as shown in Figure 6c.

4. Discussion

In this study, we developed a simple and compact linear flow phantom for accurate retinal blood-flow velocity evaluation. The calibration curves showed excellent linearity after assembly, except for those close to the maximum voltage rating of the DC motor compared to no-load conditions. This difference is attributed to friction loss. Our phantom offered a linear velocity range between 0.5–7 mm/s, which corresponds to physiological flow velocities observed in the retinal microcirculation, and the modular design of our phantom allows for easy modifications for other ranges found in other non-biomedical applications. The bi-directional nature of the DC motor offered the ability to easily study different flow directions, as observed in arteries and veins in the retinal circulation. Our phantom, validated with AO-SLO and clinical SLO, achieves high accuracy with measured slopes close to unity. The low standard deviation of residuals for AO-SLO (0.1127 mm/s) and clinical SLO (0.1074 mm/s) demonstrates high precision.
For the AO-SLO line-scan approach, our phantom offers an optimal bead density for flow determination and was tested at various phantom velocities. We measured a linear correlation between phantom and AO-SLO derived mean velocities that was near unity. Previous reports showed that AO-SLO line-scan derived velocity measurement requires vessel selection such that the intersection angle between the vessel and scanner is always less than 45° for high accuracy [11,18]. Typically, a steeper streak and vessel angle results in reduced slope-determination sensitivity and shorter streaks on space–time images for a given flow velocity, respectively [39]. This happens because the time taken for erythrocytes to cross the scanning line is inversely proportional to the component of the erythrocyte velocity perpendicular to the fast scanning direction, as shown by Zhong et al. [18]. This limits the number of vessel segments where the flow can be quantified in any imaged region. With our phantom model, equipped with a flexible rotation stage, we systematically studied the effect of vessel angle on measured velocities, and showed a non-linear increase in error beyond 45°, up to ~400% error for angles near 90°. To correct this, a non-linear regression model fit to calibration data collected with the phantom could derive a correction factor for vessel angles between 40° to 90°. This could potentially extend the measurement of microvasculature flow rates to all vessels in an imaged region without any hardware modification.
EMA offers direct quantitation of blood flow by tracing individual ICG-labelled red blood cells with a clinical SLO instrument. A recent study combined AO-SLO and EMA to quantify retinal blood-flow measurements in healthy human subjects [36]. The single-file flow observed with EMA was emulated with fluorescent beads, added at the ratio of 1:3 to the total beads, in our phantom and tested on a clinical SLO instrument. Results showed linear correlation between the phantom and clinical SLO-derived mean velocities, albeit with a slight underestimation. This deviation could be attributed to an error in the scaling calibration provided by the clinical SLO instrument, which assumes a human eye model compared to our phantom model in air. For accurate velocity measures in humans, it is imperative to know the imaged region’s true scaling (µm/pixel), which depends on the optics of the human eye. Many eye models have been proposed and used for the spatial correction of fundus images with varying degrees of complexity and parameterization [40], including axial length (AL) [41,42], lens thickness, refractive error [43], and anterior chamber depth [42,43]. Garraway-Heath et al. showed that accurate quantitation could be achieved using only AL [44]. Any deviation in AL from the true size could lead to an erroneous magnification factor and could impact critical clinical decisions in disease management [45,46,47].
Retinal phantoms play a crucial role in objective performance validation of imaging systems because subject variability is removed. Several groups have developed multilayer phantoms to emulate retinal structures [48,49,50]. However, most published phantoms are limited to structural assessment and do not mimic vascular blood flow. In another study, a biomimetic 3D-printed vascular phantom was developed via direct laser writing but lacked capillary single-file flow modeling [51]. Traditionally, circular polyethylene tubes or glass capillaries have been used in flow phantoms [52]. Although easy to fabricate, blood-vessel patterns and available lumen diameters are limited. Many microfluidic phantoms have been developed to accurately assess microvasculature and use intralipids to mimic flowing red blood cells with optimal concentrations. Materials like PDMS exhibit thermal shrinkage, which affects microfluidic channel lumen after curing and could lead to undesired flow patterns. In addition, bulky pump setups and underlying fluid dynamics add complexity to the phantom. Our phantom assembly uses off-the-shelf parts that are inexpensive, resulting in a total cost of less than $500, excluding the cost of the 3D printing and computer. While our phantom lacks anthropomorphic features, it avoids wet components and offers a simple, well-controlled model for accurate functional assessment. One of the limitations of our model is that the optical elements do not match the refractive index of the human eye. In addition, our model does not represent the vessel within retinal layers in depth. Furthermore, the current phantom only models relatively simple flow based upon the average blood velocity and does not consider many complex flow properties, hemodynamic effects, and physiological parameters that may impact actual retinal blood flow (red blood cell shape, Reynold’s number, blood as a multiphase non-Newtonian fluid, etc.). Some of these limitations could be addressed in future iterations of the phantom.
In the future, our phantom could be adapted for quantitative performance validation of both clinical and non-clinical imaging instruments. Additionally, the phantom is equipped to easily drive voltage signals in the software so that pulsatile blood flow could be mimicked to enable accurate calibration of dynamic parameters of retinal blood flow during the cardiac cycle. Our model can be used to calibrate, evaluate, and assess the velocity accuracy of current clinical imaging systems designed for retinal blood-flow measurement. Moreover, it can play a crucial role in assessing devices designed to characterize abnormal blood-flow patterns associated with many ocular diseases.

5. Conclusions

We presented a simple, compact, cost-effective, dry linear flow phantom for evaluating optical imaging devices that measure retinal blood velocity. The phantom model was optimized for bead density, including sparse fluorescent beads mimicking labeling in the EMA technique. The calibration showed a linearity in the 0.5–7 mm/s velocity range. Further, the model was calibrated and evaluated using a sensitive AO-SLO line-scan technique. The effects of varying vessel angles on AO-SLO results were systematically studied to improve the accuracy of retinal blood-flow quantitation. This model has great potential to calibrate and improve the accuracy of existing clinical imaging systems that measure retinal blood flow to aid in ocular disease diagnosis.

Author Contributions

Conceptualization, A.J.R., A.A. and O.J.S.; methodology, A.J.R., A.A., Z.L., D.X.H. and O.J.S.; software, A.J.R., A.A., Z.L. and D.X.H.; formal analysis, A.J.R.; investigation, A.J.R., A.A. and O.J.S.; resources, A.J.R., A.A., D.X.H. and O.J.S.; data curation, A.J.R., A.A. and O.J.S..; writing—original draft preparation, A.M.E. and A.J.R.; writing—review and editing, A.A., Z.L., D.X.H. and O.J.S.; visualization, A.J.R. and A.M.E.; supervision, D.X.H. and O.J.S.; project administration, A.J.R., A.A. and O.J.S.; funding acquisition, A.J.R., A.A. and O.J.S. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the U.S. Food and Drug Administration through the Center of Excellence in Regulatory Science and Innovation (M-CERSI) program and by the National Eye Institute/National Institutes of Health (R01EY031731).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Acknowledgments

We are grateful to the staff from Additive Manufacturing for Medical Products (AMMP) Laboratory for aid in 3D-printing components for this project.

Conflicts of Interest

O.J.S. has received research and non-research support from Heidelberg Engineering. All other authors declare no conflicts of interest.

Commercial Disclaimer

The mention of commercial products, their sources, or their use in connection with material reported herein is not to be construed as either an actual or implied endorsement of such products by the U.S. Department of Health and Human Services.

References

  1. Sehi, M.; Goharian, I.; Konduru, R.; Tan, O.; Srinivas, S.; Sadda, S.R.; Francis, B.A.; Huang, D.; Greenfield, D.S. Retinal blood flow in glaucomatous eyes with single-hemifield damage. Ophthalmology 2014, 121, 750–758. [Google Scholar] [CrossRef] [PubMed]
  2. Cherecheanu, A.P.; Garhofer, G.; Schmidl, D.; Werkmeister, R.; Schmetterer, L. Ocular perfusion pressure and ocular blood flow in glaucoma. Curr. Opin. Pharmacol. 2013, 13, 36–42. [Google Scholar] [CrossRef]
  3. Taylor, T.R.P.; Menten, M.J.; Rueckert, D.; Sivaprasad, S.; Lotery, A.J. The role of the retinal vasculature in age-related macular degeneration: A spotlight on OCTA. Eye 2024, 38, 442–449. [Google Scholar] [CrossRef] [PubMed]
  4. Mujat, M.; Sampani, K.; Patel, A.H.; Sun, J.K.; Iftimia, N. Cellular-level analysis of retinal blood vessel walls based on phase gradient images. Diagnostics 2023, 13, 3399. [Google Scholar] [CrossRef] [PubMed]
  5. Witt, N.; Wong, T.Y.; Hughes, A.D.; Chaturvedi, N.; Klein, B.E.; Evans, R.; McNamara, M.; Thom, S.A.M.; Klein, R. Abnormalities of retinal microvascular structure and risk of mortality from ischemic heart disease and stroke. Hypertension 2006, 47, 975–981. [Google Scholar] [CrossRef] [PubMed]
  6. Cheung, C.Y.; Ikram, M.K.; Chen, C.; Wong, T.Y. Imaging retina to study dementia and stroke. Prog. Retin. Eye Res. 2017, 57, 89–107. [Google Scholar] [CrossRef] [PubMed]
  7. Hanaguri, J.; Yokota, H.; Watanabe, M.; Yamagami, S.; Kushiyama, A.; Kuo, L.; Nagaoka, T. Retinal blood flow dysregulation precedes neural retinal dysfunction in type 2 diabetic mice. Sci. Rep. 2021, 11, 18401. [Google Scholar] [CrossRef] [PubMed]
  8. Nagaoka, T.; Sato, E.; Takahashi, A.; Yokota, H.; Sogawa, K.; Yoshida, A. Impaired retinal circulation in patients with type 2 diabetes mellitus: Retinal laser Doppler velocimetry study. Investig. Ophthalmol. Vis. Sci. 2010, 51, 6729–6734. [Google Scholar] [CrossRef] [PubMed]
  9. Bedggood, P.; Metha, A. Adaptive optics imaging of the retinal microvasculature. Clin. Exp. Optom. 2020, 103, 112–122. [Google Scholar] [CrossRef] [PubMed]
  10. Li, J.; Wang, D.; Pottenburgh, J.; Bower, A.J.; Asanad, S.; Lai, E.W.; Simon, C.; Im, L.; Huryn, L.A.; Tao, Y.; et al. Visualization of erythrocyte stasis in the living human eye in health and disease. iScience 2023, 26, 105755. [Google Scholar] [CrossRef]
  11. Joseph, A.; Guevara-Torres, A.; Schallek, J. Imaging single-cell blood flow in the smallest to largest vessels in the living retina. Elife 2019, 8, e45077. [Google Scholar] [CrossRef] [PubMed]
  12. Gu, B.; Wang, X.; Twa, M.D.; Tam, J.; Girkin, C.A.; Zhang, Y. Noninvasive in vivo characterization of erythrocyte motion in human retinal capillaries using high-speed adaptive optics near-confocal imaging. Biomed. Opt. Express 2018, 9, 3653–3677. [Google Scholar] [CrossRef]
  13. Bedggood, P.; Metha, A. Direct visualization and characterization of erythrocyte flow in human retinal capillaries. Biomed. Opt. Express 2012, 3, 3264–3277. [Google Scholar] [CrossRef] [PubMed]
  14. Tam, J.; Tiruveedhula, P.; Roorda, A. Characterization of single-file flow through human retinal parafoveal capillaries using an adaptive optics scanning laser ophthalmoscope. Biomed. Opt. Express 2011, 2, 781–793. [Google Scholar] [CrossRef] [PubMed]
  15. Martin, J.A.; Roorda, A. Direct and noninvasive assessment of parafoveal capillary leukocyte velocity. Ophthalmology 2005, 112, 2219–2224. [Google Scholar] [CrossRef] [PubMed]
  16. Chui, T.Y.; Gast, T.J.; Burns, S.A. Imaging of vascular wall fine structure in the human retina using adaptive optics scanning laser ophthalmoscopy. Investig. Ophthalmol. Vis. Sci. 2013, 54, 7115–7124. [Google Scholar] [CrossRef] [PubMed]
  17. Koch, E.; Rosenbaum, D.; Brolly, A.; Sahel, J.A.; Chaumet-Riffaud, P.; Girerd, X.; Rossant, F.; Paques, M. Morphometric analysis of small arteries in the human retina using adaptive optics imaging: Relationship with blood pressure and focal vascular changes. J. Hypertens. 2014, 32, 890–898. [Google Scholar] [CrossRef]
  18. Zhong, Z.; Petrig, B.L.; Qi, X.; Burns, S.A. In vivo measurement of erythrocyte velocity and retinal blood flow using adaptive optics scanning laser ophthalmoscopy. Opt. Express 2008, 16, 12746–12756. [Google Scholar] [CrossRef]
  19. Burns, S.A.; Elsner, A.E.; Sapoznik, K.A.; Warner, R.L.; Gast, T.J. Adaptive optics imaging of the human retina. Prog. Retin. Eye Res. 2019, 68, 1–30. [Google Scholar] [CrossRef]
  20. Bedggood, P.; Metha, A. Mapping flow velocity in the human retinal capillary network with pixel intensity cross correlation. PLoS ONE 2019, 14, e0218918. [Google Scholar] [CrossRef]
  21. Liu, Z.; Saeedi, O.; Zhang, F.; Villanueva, R.; Asanad, S.; Agrawal, A.; Hammer, D.X. Quantification of Retinal Ganglion Cell Morphology in Human Glaucomatous Eyes. Investig. Ophthalmol. Vis. Sci. 2021, 62, 34. [Google Scholar] [CrossRef] [PubMed]
  22. Chen, V.Y.; Le, C.T.; Pottenburgh, J.; Siddiqui, A.; Park, A.; Asanad, S.; Magder, L.; Im, L.T.; Saeedi, O.J. A pilot study assessing retinal blood flow dysregulation in glaucoma using erythrocyte mediated velocimetry. Transl. Vision. Sci. Technol. 2022, 11, 19. [Google Scholar] [CrossRef] [PubMed]
  23. Burns, S.A.; Elsner, A.E.; Gast, T.J. Imaging the Retinal Vasculature. Annu. Rev. Vis. Sci. 2021, 7, 129–153. [Google Scholar] [CrossRef] [PubMed]
  24. Riva, C.E.; Sinclair, S.H.; Grunwald, J.E. Autoregulation of retinal circulation in response to decrease of perfusion pressure. Investig. Ophthalmol. Vis. Sci. 1981, 21 Pt 1, 34–38. [Google Scholar]
  25. Warner, R.L.; Gast, T.J.; Sapoznik, K.A.; Carmichael-Martins, A.; Burns, S.A. Measuring Temporal and Spatial Variability of Red Blood Cell Velocity in Human Retinal Vessels. Investig. Ophthalmol. Vis. Sci. 2021, 62, 29. [Google Scholar] [CrossRef] [PubMed]
  26. Agrawal, A.; Connors, M.; Beylin, A.; Liang, C.P.; Barton, D.; Chen, Y.; Drezek, R.A.; Pfefer, T.J. Characterizing the point spread function of retinal OCT devices with a model eye-based phantom. Biomed. Opt. Express 2012, 3, 1116–1126. [Google Scholar] [CrossRef] [PubMed]
  27. Kedia, N.; Liu, Z.; Sochol, R.D.; Tam, J.; Hammer, D.X.; Agrawal, A. 3-D printed photoreceptor phantoms for evaluating lateral resolution of adaptive optics imaging systems. Opt. Lett. 2019, 44, 1825–1828. [Google Scholar] [CrossRef] [PubMed]
  28. Okubo, T.; Mino, T. Design and fabrication of a long-term stable model eye for OCT retinal imaging. In Proceedings of the SPIE Optical Coherence Tomography and Coherence Domain Optical Methods in Biomedicine XXVII, San Francisco, CA, USA, 28 January–3 February 2023. [Google Scholar]
  29. Lee, H.-J.; Samiudin, N.M.; Lee, T.G.; Doh, I.; Lee, S.-W. Retina phantom for the evaluation of optical coherence tomography angiography based on microfluidic channels. Biomed. Opt. Express 2019, 10, 5535–5548. [Google Scholar] [CrossRef] [PubMed]
  30. Chiaravalli, G.; Guidoboni, G.; Sacco, R.; Radell, J.; Harris, A. A multi-scale/multi-physics model for the theoretical study of the vascular configuration of retinal capillary plexuses based on OCTA data. Math. Med. Biol. A J. IMA 2022, 39, 77–104. [Google Scholar] [CrossRef]
  31. Guidoboni, G.; Harris, A.; Cassani, S.; Arciero, J.; Siesky, B.; Amireskandari, A.; Tobe, L.; Egan, P.; Januleviciene, I.; Park, J. Intraocular pressure, blood pressure, and retinal blood flow autoregulation: A mathematical model to clarify their relationship and clinical relevance. Investig. Ophthalmol. Vis. Sci. 2014, 55, 4105–4118. [Google Scholar] [CrossRef]
  32. Liu, D.; Wood, N.; Witt, N.; Hughes, A.; Thom, S.; Xu, X. Computational analysis of oxygen transport in the retinal arterial network. Curr. Eye Res. 2009, 34, 945–956. [Google Scholar] [CrossRef]
  33. Sampson, D.M.; Dubis, A.M.; Chen, F.K.; Zawadzki, R.J.; Sampson, D.D. Towards standardizing retinal optical coherence tomography angiography: A review. Light. Sci. Appl. 2022, 11, 63. [Google Scholar] [CrossRef] [PubMed]
  34. Saeedi, O.; Ou, M.; Kalarn, S.; Jones, A.; Toledo, L.; Im, L.; Quigley, H.A. Erythrocyte Mediated Angiography: Use of a Novel Method to Determine Erythrocyte Dynamics in the Glaucomatous Optic Nerve. Investig. Ophthalmol. Vis. Sci. 2017, 58, 3130. [Google Scholar]
  35. Liu, Z.; Tam, J.; Saeedi, O.; Hammer, D.X. Trans-retinal cellular imaging with multimodal adaptive optics. Biomed. Opt. Express 2018, 9, 4246–4262. [Google Scholar] [CrossRef]
  36. Raghavendra, A.; Chen, V.; Liu, Z.; Hammer, D.; Saeedi, O. Quantitative retinal blood flow measurement in humans using adaptive optics scanning laser ophthalmoscopy and erythrocyte mediated angiography. In Proceedings of the SPIE Ophthalmic Technologies XXXIII, San Francisco, CA, USA, 28 January–3 February 2023. [Google Scholar]
  37. Tinevez, J.Y.; Perry, N.; Schindelin, J.; Hoopes, G.M.; Reynolds, G.D.; Laplantine, E.; Bednarek, S.Y.; Shorte, S.L.; Eliceiri, K.W. TrackMate: An open and extensible platform for single-particle tracking. Methods 2017, 115, 80–90. [Google Scholar] [CrossRef]
  38. Jaqaman, K.; Loerke, D.; Mettlen, M.; Kuwata, H.; Grinstein, S.; Schmid, S.L.; Danuser, G. Robust single-particle tracking in live-cell time-lapse sequences. Nat. Methods 2008, 5, 695–702. [Google Scholar] [CrossRef]
  39. Chhatbar, P.Y.; Kara, P. Improved blood velocity measurements with a hybrid image filtering and iterative Radon transform algorithm. Front. Neurosci. 2013, 7, 106. [Google Scholar] [CrossRef] [PubMed]
  40. Huang, X.; Anderson, T.; Dubra, A. Retinal magnification factors at the fixation locus derived from schematic eyes with four individualized surfaces. Biomed. Opt. Express 2022, 13, 3786–3808. [Google Scholar] [CrossRef] [PubMed]
  41. Rudnicka, A.R.; Burk, R.O.; Edgar, D.F.; Fitzke, F.W. Magnification characteristics of fundus imaging systems. Ophthalmology 1998, 105, 2186–2192. [Google Scholar] [CrossRef] [PubMed]
  42. Bennett, A.G.; Rudnicka, A.R.; Edgar, D.F. Improvements on Littmann’s method of determining the size of retinal features by fundus photography. Graefe’s Arch. Clin. Exp. Ophthalmol. 1994, 232, 361–367. [Google Scholar] [CrossRef]
  43. Bengtsson, B.; Krakau, C. Some essential optical features of the Zeiss fundus camera. Acta Ophthalmol. 1977, 55, 123–131. [Google Scholar] [CrossRef] [PubMed]
  44. Garway-Heath, D.F.; Rudnicka, A.R.; Lowe, T.; Foster, P.J.; Fitzke, F.W.; Hitchings, R.A. Measurement of optic disc size: Equivalence of methods to correct for ocular magnification. Br. J. Ophthalmol. 1998, 82, 643–649. [Google Scholar] [CrossRef] [PubMed]
  45. Sampson, D.M.; Gong, P.; An, D.; Menghini, M.; Hansen, A.; Mackey, D.A.; Sampson, D.D.; Chen, F.K. Axial length variation impacts on superficial retinal vessel density and foveal avascular zone area measurements using optical coherence tomography angiography. Investig. Ophthalmol. Vis. Sci. 2017, 58, 3065–3072. [Google Scholar] [CrossRef] [PubMed]
  46. Chen, C.Y.; Huang, E.J.C.; Kuo, C.N.; Wu, P.L.; Chen, C.L.; Wu, P.C.; Wu, S.H.; King, Y.C.; Lai, C.H. The relationship between age, axial length and retinal nerve fiber layer thickness in the normal elderly population in Taiwan: The Chiayi eye study in Taiwan. PLoS ONE 2018, 13, e0194116. [Google Scholar] [CrossRef] [PubMed]
  47. Savini, G.; Barboni, P.; Parisi, V.; Carbonelli, M. The influence of axial length on retinal nerve fibre layer thickness and optic-disc size measurements by spectral-domain OCT. Br. J. Ophthalmol. 2011, 96, 57–61. [Google Scholar] [CrossRef] [PubMed]
  48. Chen, Z.; Milner, T.E.; Dave, D.; Nelson, J.S. Optical Doppler tomographic imaging of fluid flow velocity in highly scattering media. Opt. Lett. 1997, 22, 64–66. [Google Scholar] [CrossRef] [PubMed]
  49. Baxi, J.; Calhoun, W.; Sepah, Y.J.; Hammer, D.X.; Ilev, I.; Joshua Pfefer, T.; Nguyen, Q.D.; Agrawal, A. Retina-simulating phantom for optical coherence tomography. J. Biomed. Opt. 2014, 19, 021106. [Google Scholar] [CrossRef] [PubMed]
  50. Rowe, T.S.; Zawadzki, R.J. New developments in eye models with retina tissue phantoms for ophthalmic optical coherence tomography. In Proceedings of the SPIE Optical Diagnostics and Sensing XII: Toward Point-of-Care Diagnostics, and Design and Performance Validation of Phantoms Used in Conjunction with Optical Measurement of Tissue IV, San Francisco, CA, USA, 21–26 January 2012. [Google Scholar]
  51. Horng, H.; O’Brien, K.; Lamont, A.; Sochol, R.D.; Pfefer, T.J.; Chen, Y. 3D printed vascular phantoms for high-resolution biophotonic image quality assessment via direct laser writing. Opt. Lett. 2021, 46, 1987–1990. [Google Scholar] [CrossRef]
  52. Tokayer, J.; Jia, Y.; Dhalla, A.-H.; Huang, D. Blood flow velocity quantification using split-spectrum amplitude-decorrelation angiography with optical coherence tomography. Biomed. Opt. Express 2013, 4, 1909–1924. [Google Scholar] [CrossRef]
Diagnostics 14 01615 g001
Figure 1. Linear flow phantom: (a) photograph of the phantom consisting of a motor and timing belt pulley system on a manual rotation mount attached to an optical post; (b) a high-torque DC motor with a synchronous timing belt, pulley, and 3D-printed parts used in the assembly; and (c) the schematic shows the optical system, including iris, lens, and timing belt on an idle pulley, such that the surface of the slide with beads is at the focal plane. The angular velocity of the pulley, ω °/s attached to the motor shaft, translates to linear velocity v mm/s.
Figure 1. Linear flow phantom: (a) photograph of the phantom consisting of a motor and timing belt pulley system on a manual rotation mount attached to an optical post; (b) a high-torque DC motor with a synchronous timing belt, pulley, and 3D-printed parts used in the assembly; and (c) the schematic shows the optical system, including iris, lens, and timing belt on an idle pulley, such that the surface of the slide with beads is at the focal plane. The angular velocity of the pulley, ω °/s attached to the motor shaft, translates to linear velocity v mm/s.
Diagnostics 14 01615 g001
Diagnostics 14 01615 g002
Figure 2. (a) Slide preparation for the phantom includes aqueous drop casting with an optimized concentration of polystyrene microspheres with fluorescent beads on a clean, flexible plastic slide. The preparation was dried overnight to achieve a uniform monolayer inside a humidity chamber. (b) White-light brightfield microscopy of the slide shows a monolayer mixture of fluorescent and non-fluorescent deposited beads with a density of ~580 particles/mm2. (c) The same region imaged with fluorescence shows the near-IR fluorescent beads only with a density of ~40 particles/mm2. Bead density measurement was performed using the ImageJ analyze particles tool at ten different regions on the slide.
Figure 2. (a) Slide preparation for the phantom includes aqueous drop casting with an optimized concentration of polystyrene microspheres with fluorescent beads on a clean, flexible plastic slide. The preparation was dried overnight to achieve a uniform monolayer inside a humidity chamber. (b) White-light brightfield microscopy of the slide shows a monolayer mixture of fluorescent and non-fluorescent deposited beads with a density of ~580 particles/mm2. (c) The same region imaged with fluorescence shows the near-IR fluorescent beads only with a density of ~40 particles/mm2. Bead density measurement was performed using the ImageJ analyze particles tool at ten different regions on the slide.
Diagnostics 14 01615 g002
Diagnostics 14 01615 g003
Figure 3. Performance of the phantom in the DC motor voltage range of −12 V to 12 V showed (a) a linear relation (r2 > 0.999) with rotational velocity (ω °/s) without load and (b) with complete phantom assembly showed a linear relation (r2 > 0.989) with belt linear velocity. Linearity for phantom use was measured between ±1.5 to 11 volts, corresponding to 0.5 to 7 mm/s belt linear velocity.
Figure 3. Performance of the phantom in the DC motor voltage range of −12 V to 12 V showed (a) a linear relation (r2 > 0.999) with rotational velocity (ω °/s) without load and (b) with complete phantom assembly showed a linear relation (r2 > 0.989) with belt linear velocity. Linearity for phantom use was measured between ±1.5 to 11 volts, corresponding to 0.5 to 7 mm/s belt linear velocity.
Diagnostics 14 01615 g003
Diagnostics 14 01615 g004
Figure 4. AO-SLO imaging of phantom with 20 mm focal length achromat. (a) A stationary AO-SLO raster image of beads. Scale bar = 50 μm. (b) A corresponding space–time image in AO-SLO line-scan mode shows streaks due to the motion of beads at a given phantom linear velocity. The orange box outline shows the sliding region of interest encompassing streaks, analyzed by the custom algorithm for slope calculation. (c) Mean velocity was computed for three independent video acquisitions for each input phantom velocity.
Figure 4. AO-SLO imaging of phantom with 20 mm focal length achromat. (a) A stationary AO-SLO raster image of beads. Scale bar = 50 μm. (b) A corresponding space–time image in AO-SLO line-scan mode shows streaks due to the motion of beads at a given phantom linear velocity. The orange box outline shows the sliding region of interest encompassing streaks, analyzed by the custom algorithm for slope calculation. (c) Mean velocity was computed for three independent video acquisitions for each input phantom velocity.
Diagnostics 14 01615 g004
Diagnostics 14 01615 g005
Figure 5. (a) The effect of varying vessel angle dependence in the line-scan approach is evident from the large error above 45°. The phantom used a 30 mm focal length achromat lens. Error bars indicated standard deviation over three independent measurements. (b) Streaks observed for each vessel angle setting at the same phantom velocity on space–time images show angle-dependent streak length.
Figure 5. (a) The effect of varying vessel angle dependence in the line-scan approach is evident from the large error above 45°. The phantom used a 30 mm focal length achromat lens. Error bars indicated standard deviation over three independent measurements. (b) Streaks observed for each vessel angle setting at the same phantom velocity on space–time images show angle-dependent streak length.
Diagnostics 14 01615 g005
Diagnostics 14 01615 g006
Figure 6. (a) Photograph of the clinical SLO used to test the phantom. A single frame from an ICGA video showed well-resolved fluorescent beads sampled at 15.4 fps; (b) a single composite image for each phantom velocity setting (0.8–5.8 mm/s) is produced by applying a maximum intensity projection across all frames in each video to visualize the motion traces of fluorescent beads. Scale bar is 200 µm; and (c) analysis of varying phantom velocity showed a linear trend (r2 > 0.997). Error bars indicate standard deviation over three independent measurements.
Figure 6. (a) Photograph of the clinical SLO used to test the phantom. A single frame from an ICGA video showed well-resolved fluorescent beads sampled at 15.4 fps; (b) a single composite image for each phantom velocity setting (0.8–5.8 mm/s) is produced by applying a maximum intensity projection across all frames in each video to visualize the motion traces of fluorescent beads. Scale bar is 200 µm; and (c) analysis of varying phantom velocity showed a linear trend (r2 > 0.997). Error bars indicate standard deviation over three independent measurements.
Diagnostics 14 01615 g006
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Raghavendra, A.J.; Elhusseiny, A.M.; Agrawal, A.; Liu, Z.; Hammer, D.X.; Saeedi, O.J. Compact Linear Flow Phantom Model for Retinal Blood-Flow Evaluation. Diagnostics 2024, 14, 1615. https://doi.org/10.3390/diagnostics14151615

AMA Style

Raghavendra AJ, Elhusseiny AM, Agrawal A, Liu Z, Hammer DX, Saeedi OJ. Compact Linear Flow Phantom Model for Retinal Blood-Flow Evaluation. Diagnostics. 2024; 14(15):1615. https://doi.org/10.3390/diagnostics14151615

Chicago/Turabian Style

Raghavendra, Achyut J., Abdelrahman M. Elhusseiny, Anant Agrawal, Zhuolin Liu, Daniel X. Hammer, and Osamah J. Saeedi. 2024. "Compact Linear Flow Phantom Model for Retinal Blood-Flow Evaluation" Diagnostics 14, no. 15: 1615. https://doi.org/10.3390/diagnostics14151615

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Article metric data becomes available approximately 24 hours after publication online.
Back to TopTop