Open Access
Add to BibSonomyAbstract
Melanin is the dominant light absorber in retinal pigment epithelium (RPE). The loss of RPE melanin is a sign of ocular senescence and is both a risk factor and a symptom of age-related macular degeneration (AMD). Quantifying the RPE melanin concentration provides insight into the pathological role of RPE in ocular aging and the onset and progression of AMD. The main challenge in accurate quantification of RPE melanin concentration is to distinguish this ten-micrometer-thick cell monolayer from the underlying choroid, which also contains melanin but carries different pathognomonic information. In this work, we investigated a three-dimensional photoacoustic microscopic (PAM) method with high axial resolution, empowered by broad acoustic detection bandwidth, to distinguish RPE from choroid and quantify melanin concentrations in the RPE ex vivo. We first conducted numerical simulation on photoacoustic generation in the RPE, which suggested that a PAM system with at least 100-MHz detection bandwidth provided sufficient axial resolution to distinguish the melanin in RPE from that in choroid. Based on simulation results, we integrated a transparent broadband micro-ring resonator (MRR) based detector in a homebuilt PAM system. We imaged ex vivo RPE-choroid complexes (RCCs) from both porcine and human eyes and quantified the absolute melanin concentrations in the RPE and choroid, respectively. In our study, the measured melanin concentrations were 14.7 mg/mL and 17.0 mg/mL in human and porcine RPE, and 12 mg/mL and 61 mg/mL in human and porcine choroid, respectively. This study suggests that broadband PAM is capable of quantifying the RPE melanin concentration from RCCs ex vivo.
© 2017 Optical Society of America
1. Introduction
Retinal pigment epithelium (RPE) is a monolayer of pigmented cells beneath the neural retina and tightly attached to the underlying choroid. The RPE plays a crucial role in nourishing photoreceptors, recycling photopigment and maintaining normal visual cycle [1]. The brownish/black color of the RPE originates from its intracellular melanin (eumelanin, dominantly) [2], which serves as a photo-protector and a radical scavenger to protect neural retina from photooxidation [3]. The RPE melanin cannot be reproduced postnatally and its concentration decreases due to life-long exposure to light and oxidative stress [4, 5]. The gradual loss of the RPE melanin compromises its photoprotection mechanism, indicates ocular senescence and makes the eye susceptible to degeneration diseases [4]. For example, the RPE melanin concentration carries pathognomonic signs of age-related macular degeneration (AMD), which is the leading blinding disease in developed world [6, 7]. Subjects with low RPE melanin concentration are more susceptible to the degeneration, which, once developed, leads to further loss of melanin and the RPE cells in macular area [8, 9]. Quantification of absolute melanin concentration in the RPE can significantly enrich our knowledge regarding melanin’s role in aging and onset of different ocular diseases, especially AMD.
Despite the importance of RPE melanin and the close attention it draws from both fundamental scientists and clinicians, the specific RPE melanin concentration was reported in few studies, largely due to lack of simple, efficient and accessible measurement methods. In the 1980s, researchers counted the number of melanin granules in RPE layer on high-magnification micrographs of representative thin ocular tissue sections [10]. Later on, optical absorption (OA) based methods were applied to measure the optical attenuation through either isolated RPE melanin solution [11] or intact RPE tissue [12, 13], which was then related to melanin concentration given extinction coefficient of melanin. Recently proposed electron spin resonance (ESR) spectroscopy could give absolute melanin quantity in the tested sample by comparing measured spectra amplitude of RPE cell extracts with that of synthetic melanin [4]. Though feasible, accurate quantification of the RPE melanin concentration by either OA or ESR method remained practically challenging. The RPE is only around ten-micrometer-thick and is tightly attached to choroid, a hundreds-of-micrometer-thick connective tissue layer containing large amount of melanin and blood vessels [14]. The choroidal melanin carries little pathological information and is considered of less interest [15]. In both OA and ESR methods, researchers needed to completely exclude choroidal tissues in the tested sample in order to make the measurement accurate and specific. Yet, physically peeling the delicate cell monolayer of RPE from underlying choroid was troublesome and demanding due to their close adjacency. In most cases, investigators could only obtain the integrated melanin concentration in RPE-choroid complexes (RCCs) [13].
Recently, several non-invasive imaging based methods for RPE melanin quantification were also proposed since they could provide melanin distribution with cellular level spatial resolution in vivo. However, these methods were still affected by choroidal melanin. Fundus spectral analysis method [16] employed a foveal reflection model and used the fundus reflectance to calculate intraocular melanin optical density. Though providing consistent results on macular pigment estimation, this method could not give accurate RPE melanin concentration due to the oversimplified scattering-free modeling [16]. Near-infrared autofluorescence (NIR-AF) imaging [17] was proposed by researchers who found coincidence of higher melanin concentration and stronger AF signal excited by NIR light. Yet, rigorous quantitative relationship was not established and the autofluorescence in retina was not specific to the RPE melanin [17]. In addition, the above two methods could not measure three-dimensional (3D) tissue properties and they did not rule out influence from choroidal melanin due to lack of axial resolution. To overcome the inherent limitation of 2D imaging, Baumann, et al. related the depolarization contrast from 3D polarization-sensitive optical coherence tomography (PS-OCT) to RPE melanin concentration and demonstrated detecting RPE melanin difference in normal and albino human subjects. However, quantitatively measuring RPE melanin concentration has yet to be demonstrated [18, 19].
Photoacoustic microscopy (PAM) is a 3D imaging method relying on optical absorption contrast. In PAM, the photoacoustic (PA) wave is generated by thermal-elastic expansion in the sample after being excited by focused pulsed laser illumination. The amplitude of the PA wave is proportional to the concentration of local optical absorbers, with 100% relative sensitivity [20, 21]. The time delay between the optical excitation and the acoustic signal detection is determined by the distance between the PA source and the ultrasonic detector. Therefore, the temporal profile of the received PA signal can be used to reconstruct the axial distribution of the absorbers. Similar to pure ultrasound imaging, the axial resolution of PAM depends on the bandwidth of detected acoustic signal. With broadband ultrasonic wave generated at the excitation laser focal point, the axial resolution of PAM is primarily determined by the detection bandwidth of the ultrasonic detector. PAM has been mostly applied to quantitative inspection of endogenous chromophores, including hemoglobin and melanin [22]. Applications of PAM on epidermal melanin content quantification and melanoma imaging were investigated [23, 24]. Photoacoustic ophthalmoscopy (PAOM) was developed to visualize RPE melanin in vivo, demonstrating distinct difference of RPE PA generation between normal and albino mice [25]. However, quantitative RPE melanin measurement using PAM has not been performed, due to the lack of calculation model to transform the PA signal to RPE melanin concentration, as well as insufficient axial resolution (23 µm) [25] to separate RPE from choroid using piezoelectric detectors [14].
Recently, we developed a broadband 3D isometric optical resolution PAM using an optically transparent micro-ring resonator (MRR) based ultrasonic detector on a commercial inverted microscope platform [26]. The MRR ultrasonic detector was fabricated on a microscope coverslip with the total device thickness of 225 μm. Such a miniaturized ultrasonic detector could be seamlessly integrated into microscopic modalities while providing a large ultrasonic detection bandwidth (280 MHz at −6 dB) for PA imaging. We demonstrated single cell 3D PA imaging with an axial resolution of 2.12 μm, which could potentially resolve the RPE layer from choroid [26].
In this work, we used this broadband PAM to quantify the melanin concentration in the RPE without physically separating it from choroid. We first developed a numerical model to simulate the PA signal of the RCC under different detection bandwidths and demonstrated that broadband detection with at least 100-MHz bandwidth could provide sufficient resolving power to separate RPE from choroid, and guarantee linear dependence of PA amplitude on RPE melanin concentration. We then imaged isolated porcine and human RCC samples and showed the capability of separating RPE and choroid using our broadband PAM. Furthermore, we calibrated the PAM system by measuring PA amplitude of RPE phantoms with controlled melanin concentrations. Based on the calibration, the melanin concentrations of RPE and choroidal were measured separately.
2. Methods and materials
2.1 Simulation
We first numerically investigated the influence of acoustic detection bandwidth on the quantification of RPE melanin using PAM. Figures 1(a) and 1(b) illustrate PA generation in RCC. We used a one dimensional (1D) model to simulate PA signal generation [27]. The incident direction of PAM excitation light was perpendicular to a 10-µm thick RPE layer. The beam propagated through the RPE to reach a 200-µm thick choroid layer. We modeled the RPE and the choroid as homogeneous slabs since their optical properties (absorption and scattering) could be considered uniform under the imaging conditions used in our experiments (532-nm excitation wavelength, 0.1 NA, 2.8-µm focal spot diameter). The corresponding focal depth of excitation light was about 49 μm, well covering RPE layer and the possible penetration distance in the choroid. Previous study indicated that scattering contributes minimally (less than 6%) to overall optical density of melanin [28]. Therefore, we assumed that ballistic photons dominated the illumination in the RPE and choroid and modeled the 1D optical fluence decay in the RCC by Beer-Lambert’s law.
Fig. 1 (a) Structure of RCC and light illumination. (b) Schematic of optical excitation and PA generation by RCC. (c) Simulated energy deposition profile along the depth direction. E. D.: energy deposition. (d) Simulated PA A-line with 110-MHz acoustic detection bandwidth. Norm. Amp.: normalized amplitude. (e) Simulated PA A-line with 70-MHz acoustic detection bandwidth. In PA simulations, RPE thickness was 10 µm; RPE melanin concentration was 20 mg/mL; choroidal melanin concentration was 60 mg/mL.
The initial PA amplitude can be defined by
where is the Grüneisen parameter and is the energy deposition. The energy deposition can be calculated as the product of the absorption coefficient and the optical fluence within tissue , given byIn ex vivo examination, melanin is the dominant optical absorber in both layers since choroidal blood is removed from tissue preparation [5, 14, 29, 30]. The absorption coefficient can be related to the concentration of the melanin by mass extinction coefficient , asBy combining Eq. (1)-(3), we havewhere is an overall coefficient that relates the initial PA amplitude to the melanin concentration. One should note that the equations above are simplified and aimed only for idea demonstration. Rigorously, the optical fluence within a slab of tissue is also affected by its absorption coefficient due to propagation loss. Though not represented in the equations, this is considered in the simulation below, which makes simulation result deviate slightly from the perfect linear relationship presented above.To simulate the PA pressure generated by the RCC, we assigned the melanin concentrations of RPE and choroid individually according to the literature [13]. The mass extinction coefficient of melanin was measured to be 45.36 ± 1.34 (Mean ± SD) cm−1/(mg/mL) at 532 nm by taking transmission through diluted melanin suspension in standard cuvette, which is detailed in later section. We calculated the absorption coefficients of the two layers and simulated optical fluence distribution following Beer-Lambert’s law. Optical scattering was neglected as discussed previously, assuming dominant ballistic photons illumination. We obtained energy deposition profile in the RCC by multiplying the optical fluence with the distribution of absorption coefficients within the two layers. We used the energy deposition profile to represent initial PA amplitude under constant. The energy deposition in RCC is shown in Fig. 1(c). Finally, to simulate actual PA A-line collected by acoustic detectors, we convolved the energy deposition profile in the RCC with Gaussian-shaped impulse responses at different bandwidths [27].
Figures 1(d) and 1(e) are the simulated PA A-lines and their signal envelopes obtained from Hilbert transform detected with 110-MHz (measured by full width at half maximum (FWHM), center frequency are identical to bandwidth in the simulation) and 70-MHz bandwidths. An acoustic detection bandwidth of 110-MHz provides sufficient axial resolution and the two separate peaks, corresponding to RPE and choroid respectively, can be recognized from the signal envelope. Though choroid is much thicker than the 10-µm RPE, significant optical energy only deposits in the superficial choroid due to strong attenuation by melanin. Therefore, the envelope peak width is comparable to that of the RPE. An acoustic detection bandwidth of 70-MHz is unable to distinguish separate RPE and choroid peaks due to its limited axial resolution. The two peaks in signal envelope merge into one and its amplitude is affected by both RPE and choroid. Theoretically, the PA signal amplitude of the RPE can be used as a measure of its melanin concentration, which, however, becomes no longer valid when the detection bandwidth is limited.
To determine the detection bandwidth required to distinguish the RPE layer in the PA A-line and, thus, accurately measure the RPE melanin concentration, we simulated the relationship between the measured PA amplitude of the RPE with varied RPE melanin concentrations and a constant choroidal melanin concentration under different acoustic detection bandwidths, as shown in Fig. 2(a). In order to make a fair comparison among the curves with different detection bandwidths, we normalized them to the point where the RPE melanin concentration equaled the preset choroidal melanin concentration (15 mg/mL). At this point, the boundary between the RPE and choroid vanishes and the RCC becomes a continuous layer. The ideal line (the magenta solid line in Fig. 2(a)) was obtained by applying infinite acoustic detection bandwidth, showing a perfect linear dependence of signal peak amplitude on the RPE melanin concentration. As the detection bandwidth decreases, such linear dependence no longer holds for both high and low RPE melanin concentrations. At high RPE melanin concentrations, the signal peak amplitude saturates, which starts to happen at lower melanin concentration for smaller detection bandwidth. This bandwidth-dependent saturation effect was discussed in previous publications [20, 31]. Briefly, the increase of melanin concentration leads to a larger absorption coefficient and a shorter optical penetration depth within the RPE. When the optical penetration depth is smaller than the axial resolution determined by detection bandwidth, the high frequency components in energy deposition profile cannot be collected by the detector and the actual amplitude of energy deposition becomes unrecoverable. This leads to saturation in PA signal at high melanin concentrations. At low RPE melanin concentrations, the PA peak amplitude ceases to increase monotonically with respect to the melanin concentration, and even decreases at lower RPE concentration for small detection bandwidth, for example, 70-MHz bandwidth and below. This is because when PA detection bandwidth is limited, only one peak from the RCC can be resolved. Therefore, at lower RPE melanin range, the choroidal melanin becomes the major PA source and determines the peak amplitude. The influence of choroidal melanin is reduced with 1) increase of RPE melanin concentration, making less optical fluence reaching choroid or 2) increase of detection bandwidth, making signal from RPE more apparent. As the RPE melanin concentration increases to a certain extent that the RPE replaces choroid to become the major PA source, the peak amplitude starts to increase monotonically with the RPE melanin concentration. The reported human RPE melanin concentration varies from 11 to 22 mg/mL [13]. When the detection bandwidth is sufficiently large, such as 110-MHz shown above, linear correlation between the melanin concentration and the detected RPE PA amplitude can be found at both high and low melanin concentrations.
Fig. 2 (a) Variation of normalized peak amplitudes of RPE (and choroid in cases where axial resolution cannot resolve the boundary between the two layers) PA A-lines with preset RPE melanin concentrations for different acoustic detection bandwidths. M. C.: melanin concentration; choroidal melanin concentration: 15 mg/mL. (b) Variation of relative mean square error with acoustic detection bandwidth if using RPE (and choroid in cases where axial resolution cannot resolve the boundary between the two layers) PA amplitude to estimate RPE melanin concentration. Conditions under different choroidal melanin concentrations were simulated. MSE: mean square error. The relative MSEs were calculated from preset RPE melanin concentration values between 10 mg/mL and 30 mg/mL, which range covers physiological RPE M. C. (c) Variation of choroid to RPE PA signal ratio with choroidal melanin concentrations. Conditions under different RPE melanin concentrations were simulated. Acoustic detection bandwidth: 110 MHz. RPE thickness was 10 µm in all simulations.
To quantitatively determine the bandwidth needed for accurately measuring RPE melanin concentrations, we used the ideal curve as the reference and calculated the deviation of different curves from it by relative mean square error (MSE). To adhere to physiological concentration values, we calculated relative MSEs only using curve segments between 10- and 30-mg/mL RPE melanin concentrations. We observed the decrease of MSE with acoustic detection bandwidth under different choroidal melanin concentrations, as shown in Fig. 2(b). Choroidal melanin significantly influences the PA amplitude-RPE melanin concentration curve when the bandwidth is low. Here we set the threshold at 100 MHz to tolerate simulation errors. Therefore, we believe that the relative MSE decreases with acoustic detection bandwidth and approaches zero for bandwidths above 100 MHz for all choroidal melanin concentrations. A detection bandwidth of at least 100 MHz should be sufficient to quantify the RPE melanin concentration.
We also investigated the retrieval of choroidal melanin concentration from the PA signal of the RCC under broadband detection. Unlike RPE, the energy deposition in choroid does not rely simply on the melanin concentration within itself; it is also significantly affected by the melanin concentration in the RPE. A higher RPE melanin concentration attenuates the probing light and reduces the local optical fluence and thus the energy deposition in choroid, leading to a reduced PA signal amplitude. To consider the influence of RPE melanin, we simulated the ratio between the peak amplitudes of choroid and RPE and its dependence on choroidal melanin concentration. The assumed acoustic detection bandwidth was 110 MHz. As shown in Fig. 2(c), the RPE melanin concentration does significantly affect the choroid-RPE signal ratio. However, given a certain RPE melanin concentration, the PA peak amplitude ratio increases monotonically with choroidal melanin concentration. As mentioned earlier, RPE melanin concentration can be accurately quantified without the influence of choroid under over 100-MHz detection bandwidth. Therefore, we can use the pre-calculated RPE melanin concentration to specify a calibration curve from Fig. 2(c). Though only three curves corresponding to three preset RPE melanin concentrations are shown, a calibration curve for any arbitrary RPE melanin concentration can be calculated.
Based on the simulation results above, we designed the experimental procedures to obtain melanin concentrations in the RPE and the choroid in broadband PAM system as follows. We first calibrated the system by imaging phantoms with known melanin concentrations to establish the linear relationship between the detected PA signal amplitude and the melanin concentration, whose slope was the coefficient a in Eq. (4). We then imaged RCC samples with broadband detection and identified two separate peaks from each PA A-line, corresponding to RPE and choroid. We converted the amplitude of the RPE peak to RPE melanin concentration by Eq. (4) using the predetermined coefficient a. Finally, we estimated the choroidal melanin concentration by comparing the RPE melanin concentration and the choroid-RPE amplitude ratio to the choroidal melanin calibration curve in Fig. 2(c). These processes were repeated for all PA A-lines to obtain a map of melanin concentration in both RPE and choroid.
It is notable that we assumed that the impulse response of the ultrasound detector was Gaussian [31, 32], which may not be entirely accurate for some optical detectors such as MRR. MRR has a short-pass ultrasonic spectral response where the ultrasound response does not decay in low frequency, and the bandwidth is determined by the cut-off frequency at the high frequency end. We also simulated such short-pass impulse response. The conclusions were identical with those from Gaussian impulse response.
2.2 PAM system
We experimentally demonstrated accurate quantification of RPE melanin concentration on a broadband PAM system [26]. The schematic of the experimental setup is shown in Fig. 3. We used a 532-nm Nd:YAG nanosecond pulsed laser (1-ns pulse duration, Elforlight) as the excitation light source. The laser beam was scanned by a pair of galvanometer mirrors (Nutfield Technology) and focused by an objective lens (NA 0.1, Olympus) onto the sample, which was placed on a coverslip with the side of interest (e.g. the RPE side of the RCC) facing downward. A homebuilt MRR ultrasonic detector (280-MHz bandwidth) [31] was placed between the sample and the objective lens. The laser from a narrow-band tunable source (765-781 nm, TLB-6712, New Focus) was coupled via an optical fiber to the MRR. The transmittance through MRR was modulated by the PA pressure wave generated from the sample, which shifted the resonance peaks of the MRR [31]. The fluctuation of transmittance resembled the PA wave oscillation and was recorded by an APD (APD210, 1 GHz bandwidth, Menlo Systems) through the optical fiber coupled at the outlet of the MRR. We carefully chose the wavelength of the tunable laser on the waist of a sharp resonance peak to ensure maximum detection sensitivity. A customized needle-shaped piezoelectric transducer (27-MHz bandwidth, 0.4-mm active-element), which was placed on the other side of the sample, was used to acquire the PA image simultaneously for direct comparison. Both detectors were stationary during imaging and covered a 0.3 mm by 0.3 mm field of view with good sensing uniformity. Water was applied between the sample and the detectors for ultrasound coupling. The PA signals detected by both detectors were amplified by an amplifier (500-MHz bandwidth, ZFL500NL + , Mini-circuits), digitized and recorded by an acquisition card (CobralMax, GaGe). The PA A-lines from the MRR detector were digitized at 3 GHz to capture high frequency components of the PA waveform. The PA A-lines acquired with piezoelectric transducer were digitized at 500 MHz. A computer was used to synchronize laser trigger, galvanometer mirrors scanning, and data acquisition. The imaging A-line rate was 3 kHz and it took 22 seconds to acquire a 256 by 256 image.
Fig. 3 PAM experimental system. Relay lenses are omitted in the dashed box. Piezoelectric transducer and MRR transducer were used to detect PA signal from both sides of the sample. MRR: micro-ring resonator; APD: avalanche photodiode.
2.3 Phantom preparation
Prior to quantitative RPE mapping, we calibrated the imaging system and determined the coefficient in Eq. (4) by correlating the detected PA signal amplitudes of RPE phantoms with their preset melanin concentrations. We first built RPE-mimicking phantoms with controlled melanin concentrations. We prepared the melanin suspension and characterized its extinction spectrum for phantom study. The melanin was extracted from squid ink (Alma Gourmet), which contained exclusively eumelanin, the same type of melanin as in RPE. We diluted the squid ink with distilled water at one-to-ten volume ratio and added a few drops of surfactant (Tween 20, Sigma-Aldrich). We sonicated the mixture for 30 minutes to obtain a homogeneous suspension. Then we centrifuged the suspension at 10000 RPM for two hours and removed the supernatant. We added water to solubilize the melanin pellets and repeated centrifuging two more times to further remove water-soluble substance. After that, we collected all the purified pellets and prepared a condensed melanin suspension. To determine the melanin concentration in the suspension, we sampled a small portion and measured its weight both before and after evaporating water solvent. The final concentration of the melanin suspension was calculated as 4.97 w%. The wavelength-dependent extinction coefficients of the melanin were obtained by measuring the optical attenuation through a 1-centimeter standard-sized cuvette of diluted suspension (2560 times dilution) using a commercial spectrometer (USB2000 + , Ocean Optics). Figure 4(a) shows the monotonically-decaying trend of the melanin extinction coefficient in the wavelength range between 400 nm and 800 nm. It is comparable to previous publication while slight discrepancies might be caused by different methods and different melanin samples used [33–35].
Fig. 4 (a) Melanin extinction coefficients spectrum. Shaded area, standard deviation. Ext. Coef.: extinction coefficient. (b) PA signal amplitude generated by phantoms with controlled melanin concentrations. Error bar: standard deviation.
We used agarose gel thin films containing different amount of melanin to mimic the RPE with different melanin concentrations. Agarose is a well-accepted tissue phantom in both ultrasound and PA imaging, which allows incorporation of additional ingredients to achieve desired optical and acoustic properties [36, 37]. To fabricate the agarose gel film, we first mixed different amount of the prepared melanin suspension with preheated homogeneous liquid-state water-agarose gel (2.5 w%), and dropped a tiny amount of the mixed gel on the center of a warm glass slide. We then applied two 12-µm thick small polyvinyl chloride (PVC) films as spacers on the both ends of the slide, and placed on top another glass slide. We waited 10 minutes for the agarose gel to cool down, and removed the spacers and the slide on the top. We wrapped carefully the agarose film and the bottom slide together with a PVC film to prevent water evaporation. The melanin concentration in the phantom was calculated from the initial quantities of melanin suspension, water and agarose, and confirmed by measuring the optical transmission attenuation. The melanin concentrations of RPE phantoms ranged from 0.75 mg/mL to 75 mg/mL which covered the possible range of physiological RPE melanin concentration.
2.4 Calibration of PAM system
We then calibrated the PA signal amplitude with varied melanin concentrations by imaging RPE tissue phantoms using the MRR detector. The depth of focus of the 0.1-NA objective was 49 µm, much larger than the RPE phantom thickness. As shown in Fig. 4(b), the peak amplitude of the RPE phantom linearly correlates with its melanin concentration until saturation begins to take place at 50 mg/mL [20], which agrees with our simulation result of a broadband PAM system in Fig. 2(a). Since the reported RPE melanin concentrations in both pig and human were much less than 50 mg/mL, we could convert the detected peak amplitude of RPE to melanin concentration in the following experiment based on the fitted dashed line.
2.5 RCC sample preparation
We prepared RCC samples for PAM imaging from pig and human eye (Caucasian, male, 72 y/o, healthy, Eversight Illinois) globes using the same procedure. The eye globes, preserved in paraformaldehyde (PFA) immediately after death, were divided into two chambers along coronal plane in experiment. We flattened the posterior chamber after cutting 5 incisions radially from optic disk. We removed the neural retina, rinsed the left tissue with phosphate buffered saline (PBS) and gently peeled a piece of RCC, 5 mm by 10 mm in size, off the sclera in the perifovea region. The RCC tissue was then carefully spread on a coverslip. After being air-dried, the samples adhered to the coverslip flatly. We applied nail polish on the edges for further fixation.
3. Results
We imaged porcine and human RCCs with the calibrated PAM system using both the broadband MRR detector and the narrowband piezoelectric transducer for comparison. As the two detectors were placed on opposite sides of the sample, we flipped the PA signal detected by piezoelectric transducer for comparison. Figures 5(a) and 5(b) show a typical temporal PA A-line and its corresponding Fourier transform acquired from the porcine RCC using the narrowband transducer. The bandwidth is 27 MHz, corresponding to an axial resolution of 48 μm. No separate peaks for RPE and choroid can be distinguished in the signal envelope obtained from Hilbert transform, which confirms that narrow bandwidth detection cannot separate RPE signal from choroid signal. Figure 5(c) shows the PA A-line of the porcine RCC acquired by the MRR detector. Two distinct peaks in the signal envelope are resolved, corresponding respectively to RPE and choroid. The RPE peak arrives earlier due to its closer proximity to the detector. The representation of the PA A-line in frequency domain is shown in Fig. 5(d). The approximate spectral response [31] of the transducer is modulated due to interference between PA signals generated at RPE and choroid. Figures 5(e) and 5(f) show the PA A-line of the human RCC in temporal and frequency domains. Despite demonstrating similar features to A-line from porcine RCC, the signal amplitude ratio between RPE and choroid of this human sample is much higher, indicating relatively lower choroidal melanin concentration.
Fig. 5 PA A-lines detected from RCC samples using different ultrasonic detectors. (a) PA A-line of a porcine RCC sample acquired by customized piezoelectric transducer. (b) Amplitude spectrum of the A-line signal in (a). (c) PA A-line of a porcine RCC sample acquired by broadband MRR detector. (d) Amplitude spectrum of the A-line signal in (c). (e) PA A-line of a human RCC sample acquired by broadband MRR detector. (f) Amplitude spectrum of the A-line signal in (e).
We raster scanned the illumination light on RCCs and acquired PAM images. We reconstructed en face images by performing maximum amplitude projection of PA A-lines. We compared the images acquired by narrowband piezoelectric transducer and broadband MRR detector, as shown in Fig. 6. The image of porcine RCC acquired by narrowband transducer is displayed in Fig. 6(a). Since RPE and choroid cannot be distinguished, both RPE and choroid PA signals contribute to the pixel intensity. The honeycomb-shaped RPE cells can be appreciated. The cell edges exhibit higher intensity than the cell centers, because lower melanin concentration in RPE cell edges allows less attenuated probing light reaching even more pigmented choroid underneath, which leads to a higher energy deposition and thus a stronger PA signal. This phenomenon agrees with our simulation result in Fig. 2(a), which illustrates that less melanin concentration in RPE can generate stronger PA signal due to higher energy deposition in choroid when acquisition bandwidth is limited. We acquired PA image of the same area with the broadband MRR transducer, which rendered sufficient axial resolution to distinguish RPE from choroid. We applied Hilbert transform on all the A-lines and segmented the two peaks from signal envelopes corresponding to RPE and choroid. Figure 6(b) shows the PA image of segmented porcine RPE. Higher pixel intensity indicates denser melanin distribution. Vanished signal on cell boundaries confirms our previous hypothesis that melanin concentrations are higher in the cell centers. Figure 6(c) shows the PA image of segmented porcine choroid. Unlike RPE, the pixel intensity is affected not only by melanin distribution within itself, but also the shadow pattern left by uneven melanin densities in the RPE. One should note that, even if the choroidal melanin concentration is comparable with the RPE melanin concentration, the peak amplitude corresponding to choroid will be much lower, due to the strong attenuation of the probing light in the RPE. However, in our case, the overall image intensity of choroid is still higher than that of the RPE, suggesting a much higher melanin concentration in the choroid.
Fig. 6 PA images of porcine and human RCC. (a) PA image of porcine RCC acquired by customized piezoelectric transducer. (b)-(c) PA image of porcine RCC acquired by broadband MRR detector. (b) Axially segmented porcine RPE. (c) Axially segmented porcine choroid. (a)-(c) are images of the same area on a single sample. (d) PA image of human RCC acquired by customized piezoelectric transducer. (e)-(f) PA image of human RCC acquired by homemade MRR detector. (b) Axially segmented human RPE. (c) Axially segmented human choroid. (d)-(f) are images of the same area on a single sample. Scale bar, 50 µm.
We imaged human RCC sample using PAM with different ultrasonic detection bandwidths. Figure 6(d) shows the PA image acquired by narrowband piezoelectric transducer. The RPE cells boundaries are much less obvious compared with Fig. 6(a) possibly due to cell senility [38, 39]. We acquired PA image on the same area using broadband MRR detector and did axial segmentation to separate RPE and choroid. Figure 6(e) shows the image of segmented human RPE. The cell boundaries become more apparent. The weaker contrast between cell centers and edges compared with porcine RPE may be attributed to aging and degeneration of cellular structure. Figure 6(f) shows the image of segmented human choroid. The comparison between human and porcine choroid images indicate a much lower choroidal melanin concentration in human. Figures 7(a) and (b) show the typical B-scans of pig and human RCC respectively, where the RPE and choroid are axially separated. The comparison between B-scans also confirms that the choroid of porcine sample has stronger signal than human sample in this particular study. One should note that the relative intensities of RPE and choroid on PA B-scans do not translate directly to relative melanin concentrations. Though the melanin concentrations in RPE and choroid are similar for the human sample as will be introduced later in Fig. 8, Fig. 7(b) shows much weaker PA signal from choroid than RPE due to optical fluence attenuation. Though differences in melanin concentrations in various ocular tissues across different species were observed previously [13], our current study does not have sufficient sample size to make comprehensive comparison between human and pig eyes.
Fig. 7 (a) PA B-scan of porcine RCC. (b) PA B-scan of human RCC. RPE cells are visualized on top of choroid tissue. Scale bar, 30 µm.
Fig. 8 Comparison of melanin concentration in RPE and choroid between porcine and human samples. M. C.: Melanin concentration. Error bar: standard deviation.
We calculated RPE and choroidal melanin concentrations in porcine and human based on PA image acquired by the MRR detector. We converted PA amplitude of RPE to melanin concentration according to the calibration curve in Fig. 4(b) and calculated the choroidal melanin concentration using RPE to choroid PA amplitude ratio based on our simulation results in Fig. 2(c). The comparison of melanin RPE and choroidal melanin distribution between our imaged porcine and human eyes is shown in Fig. 8. From these particular measurements, we found similar melanin concentrations (17 mg/mL and 14.7 mg/mL) in porcine and human RPEs. The measured melanin concentration in choroid was slightly less than that of RPE in human, while porcine choroid showed much higher melanin concentration than porcine RPE. From both porcine and human samples, the variance of measured choroidal melanin concentrations across all the pixels in the image are larger than the variance of measured RPE melanin. Two reasons might lead to the increased variance from choroid measurement. First, our calculation of choroidal melanin concentration requires a known RPE melanin concentration value, which, however, is also a calculated value. So the obtained choroidal melanin concentration contains more accumulative error. Second, PA signal from the choroid is affected by the fluence decay due to the RPE melanin absorption, leading to a lower signal to noise ratio (SNR).
4. Discussion and conclusion
We demonstrated a noninvasive, quantitative measurement of melanin concentrations in RPE using broadband, high-axial-resolution PAM. Due to the difficulties in separating the two adjacent layers, previous investigators only measured the total melanin concentration in the RCC across different species without differentiating RPE and choroid and found that RCC melanin concentrations differed among animals and were generally much higher than human [13]. This might lead to an assumption that animals have much more RPE melanin than human. However, our study suggests that there can be difference in relative distribution of RPE and choroidal melanin among individual samples and probably among species. Therefore, an overall quantification within entire RCC might not give representative information about the RPE or choroid specifically. Though the current study, focused on theory and technology validation, does not have sufficient sample size to support physiological findings, one hypothesis worth future investigation is that the previously detected higher melanin concentration in animal RCCs [13] may majorly attribute to choroidal melanin while the RPE melanin concentration are actually comparable.
Our study also confirms that choroidal melanin may preclude accurate measurement of RPE melanin concentration if choroidal tissue is not completely excluded on the prepared RPE sample. We have shown that the melanin concentration in choroid is not negligible and may even exceed that in the RPE. In occasions where only RPE concentration is of interest, inclusion of any choroidal tissue undermines the validity of the measurement and even leads to erroneous conclusions. Therefore, extra care needs to be taken when physically peeling the RPE off the choroid in non-imaging-based method. For imaging-based method, the depth resolving capability and sufficient axial resolution is critical.
Compared with other methods, quantifying the melanin concentration in the RPE by broadband PAM has several advantages. First, the sample preparation of RCCs is much simpler and no extensive chemical procedures are involved. Second, given the measurement is non-destructive, the RCC samples can be preserved and the measurement is reproducible. Third, a melanin distribution can be imaged by PAM with fine spatial resolution in 3D. In addition, the measurement system setup can be easily integrated with other parallel microscopic studies for comprehensive RPE investigation. For example, the RPE melanin concentration for each single cell can be easily extracted and correlated with its susceptibility to lethal photic stress [40]; a dual-modality system combining PAM and autofluorescence imaging can be developed to monitor the real-time variation of RPE intracellular melanin and lipofuscin, the two major indicators of RPE aging, in response to various physical and chemical stimulations [26, 39, 41].
Detection of broadband acoustic signal is critical for quantification of RPE melanin concentration by PAM. Besides using a broadband ultrasonic detector, like MRR, we also need to limit the attenuation of high frequency PA signal on its propagation path. In ex vivo experiments, detector can be placed tightly close to the sample. However, for in vivo experiments, PA signal must propagate through various ocular tissues, among which vitreous and lens can significantly attenuate high frequency PA signal [42]. In addition, the safety of focusing pulsed laser onto human retina to generate PA waves with sufficient SNR still needs thorough investigation even the illumination power can be designed to be within the ANSI laser safety limit. Although RPE melanin quantifications in living human subjects can be challenging, the technology will find immediate in vivo applications in small animals like mouse, whose small eye globe (~3 mm) induces much less acoustic attenuation as compared with that from humans.
In summary, we demonstrated quantitative measurement of melanin concentration in the RPE monolayer using broadband PAM. Both numerical simulation and ex vivo experiment confirmed that the high axial resolution provided by broadband ultrasonic detection precluded influence from choroidal melanin and guaranteed RPE-specific quantitative measurement, which was hardly achieved by previous studies even with extensive anatomical and chemical preparations. Exclusion of choroid was necessary for accurate RPE melanin quantifications. The reported method provides an accurate and non-destructive quantification of the RPE melanin concentration ex vivo which can be further applied to in vivo measurements on small animals. Our current work has focused on preliminary method development and validation. In future study, we will compare RPE melanin quantification results by PAM method and analytical chemistry. We will also increase sample size and involve more species.
Funding
National Institutes of Health (NIH) (DP3DK108248, R24EY022883, and R01EY026078); National Science Foundation (NSF) (CBET-1055379 and DBI-1353952); and Illinois Society for the Prevention of Blindness.
Acknowledgments
Hao F. Zhang and Cheng Sun have financial interests with Opticent Inc., which did not support this study.
References and links
1. O. Strauss, “The retinal pigment epithelium in visual function,” Physiol. Rev. 85(3), 845–881 (2005). [CrossRef] [PubMed]
2. Y. Liu, L. Hong, K. Wakamatsu, S. Ito, B. B. Adhyaru, C. Y. Cheng, C. R. Bowers, and J. D. Simon, “Comparisons of the structural and chemical properties of melanosomes isolated from retinal pigment epithelium, iris and choroid of newborn and mature bovine eyes,” Photochem. Photobiol. 81(3), 510–516 (2005). [CrossRef] [PubMed]
3. O. Chiarelli-Neto, A. S. Ferreira, W. K. Martins, C. Pavani, D. Severino, F. Faião-Flores, S. S. Maria-Engler, E. Aliprandini, G. R. Martinez, P. Di Mascio, M. H. Medeiros, and M. S. Baptista, “Melanin photosensitization and the effect of visible light on epithelial cells,” PLoS One 9(11), e113266 (2014). [CrossRef] [PubMed]
4. T. Sarna, J. M. Burke, W. Korytowski, M. Rózanowska, C. M. Skumatz, A. Zaręba, and M. Zaręba, “Loss of melanin from human RPE with aging: possible role of melanin photooxidation,” Exp. Eye Res. 76(1), 89–98 (2003). [CrossRef] [PubMed]
5. T. Sarna, “Properties and function of the ocular melanin--a photobiophysical view,” J. Photochem. Photobiol. B 12(3), 215–258 (1992). [CrossRef] [PubMed]
6. P. T. de Jong, “Age-related macular degeneration,” N. Engl. J. Med. 355(14), 1474–1485 (2006). [CrossRef] [PubMed]
7. S. A. Randolph, “Age-Related Macular Degeneration,” Workplace Health Saf. 62(8), 352(2014). [CrossRef] [PubMed]
8. B.-L. L. Seagle, K. A. Rezai, Y. Kobori, E. M. Gasyna, K. A. Rezaei, and J. R. Norris Jr., “Melanin photoprotection in the human retinal pigment epithelium and its correlation with light-induced cell apoptosis,” Proc. Natl. Acad. Sci. U.S.A. 102(25), 8978–8983 (2005). [CrossRef] [PubMed]
9. M. Boulton, “Ageing of the retina and retinal pigment epithelium,” in Age-Related Macular Degeneration (Springer, 2013), pp. 45–63.
10. L. Feeney-Burns, E. S. Hilderbrand, and S. Eldridge, “Aging human RPE: morphometric analysis of macular, equatorial, and peripheral cells,” Invest. Ophthalmol. Vis. Sci. 25(2), 195–200 (1984). [PubMed]
11. J. J. Weiter, F. C. Delori, G. L. Wing, and K. A. Fitch, “Retinal pigment epithelial lipofuscin and melanin and choroidal melanin in human eyes,” Invest. Ophthalmol. Vis. Sci. 27(2), 145–152 (1986). [PubMed]
12. S. Y. Schmidt and R. D. Peisch, “Melanin concentration in normal human retinal pigment epithelium. Regional variation and age-related reduction,” Invest. Ophthalmol. Vis. Sci. 27(7), 1063–1067 (1986). [PubMed]
13. C. Durairaj, J. E. Chastain, and U. B. Kompella, “Intraocular distribution of melanin in human, monkey, rabbit, minipig and dog eyes,” Exp. Eye Res. 98, 23–27 (2012). [CrossRef] [PubMed]
14. X. Shu, W. Liu, and H. F. Zhang, “Monte Carlo investigation on quantifying the retinal pigment epithelium melanin concentration by photoacoustic ophthalmoscopy,” J. Biomed. Opt. 20(10), 106005 (2015). [CrossRef] [PubMed]
15. D. L. Nickla and J. Wallman, “The multifunctional choroid,” Prog. Retin. Eye Res. 29(2), 144–168 (2010). [CrossRef] [PubMed]
16. J. van de Kraats, T. T. Berendschot, and D. van Norren, “The pathways of light measured in fundus reflectometry,” Vision Res. 36(15), 2229–2247 (1996). [CrossRef] [PubMed]
17. C. N. Keilhauer and F. C. Delori, “Near-infrared autofluorescence imaging of the fundus: visualization of ocular melanin,” Invest. Ophthalmol. Vis. Sci. 47(8), 3556–3564 (2006). [CrossRef] [PubMed]
18. B. Baumann, J. Schirmer, S. Rauscher, S. Fialová, M. Glösmann, M. Augustin, M. Pircher, M. Gröger, and C. K. Hitzenberger, “Melanin pigmentation in rat eyes: in vivo imaging by polarization-sensitive optical coherence tomography and comparison to histology,” Invest. Ophthalmol. Vis. Sci. 56(12), 7462–7472 (2015). [CrossRef] [PubMed]
19. B. Baumann, S. O. Baumann, T. Konegger, M. Pircher, E. Götzinger, F. Schlanitz, C. Schütze, H. Sattmann, M. Litschauer, U. Schmidt-Erfurth, and C. K. Hitzenberger, “Polarization sensitive optical coherence tomography of melanin provides intrinsic contrast based on depolarization,” Biomed. Opt. Express 3(7), 1670–1683 (2012). [CrossRef] [PubMed]
20. J. Wang, T. Liu, S. Jiao, R. Chen, Q. Zhou, K. K. Shung, L. V. Wang, and H. F. Zhang, “Saturation effect in functional photoacoustic imaging,” J. Biomed. Opt. 15(2), 021317 (2010). [CrossRef] [PubMed]
21. J. Yao and L. V. Wang, “Sensitivity of photoacoustic microscopy,” Photoacoustics 2(2), 87–101 (2014). [CrossRef] [PubMed]
22. L. V. Wang and S. Hu, “Photoacoustic tomography: in vivo imaging from organelles to organs,” Science 335(6075), 1458–1462 (2012). [CrossRef] [PubMed]
23. J. A. Viator, J. Komadina, L. O. Svaasand, G. Aguilar, B. Choi, and J. Stuart Nelson, “A comparative study of photoacoustic and reflectance methods for determination of epidermal melanin content,” J. Invest. Dermatol. 122(6), 1432–1439 (2004). [CrossRef] [PubMed]
24. J.-T. Oh, M.-L. Li, H. F. Zhang, K. Maslov, G. Stoica, and L. V. Wang, “Three-dimensional imaging of skin melanoma in vivo by dual-wavelength photoacoustic microscopy,” J. Biomed. Opt. 11(3), 034032 (2006). [CrossRef] [PubMed]
25. X. Zhang, H. F. Zhang, C. A. Puliafito, and S. Jiao, “Simultaneous in vivo imaging of melanin and lipofuscin in the retina with photoacoustic ophthalmoscopy and autofluorescence imaging,” J. Biomed. Opt. 16(8), 080504 (2011). [CrossRef] [PubMed]
26. B. Dong, H. Li, Z. Zhang, K. Zhang, S. Chen, C. Sun, and H. F. Zhang, “Isometric multimodal photoacoustic microscopy based on optically transparent micro-ring ultrasonic detection,” Optica 2(2), 169–176 (2015). [CrossRef]
27. G. J. Diebold, T. Sun, and M. I. Khan, “Photoacoustic monopole radiation in one, two, and three dimensions,” Phys. Rev. Lett. 67(24), 3384–3387 (1991). [CrossRef] [PubMed]
28. J. Riesz, J. Gilmore, and P. Meredith, “Quantitative scattering of melanin solutions,” Biophys. J. 90(11), 4137–4144 (2006). [CrossRef] [PubMed]
29. W. J. Geeraets, R. C. Williams, G. Chan, W. T. Ham Jr, D. Guerry 3rd, and F. H. Schmidt, “The relative absorption of thermal energy in retina and choroid,” Invest. Ophthalmol. 1, 340–347 (1962). [PubMed]
30. S. J. Preece and E. Claridge, “Monte Carlo modelling of the spectral reflectance of the human eye,” Phys. Med. Biol. 47(16), 2863–2877 (2002). [CrossRef] [PubMed]
31. H. Li, B. Dong, Z. Zhang, H. F. Zhang, and C. Sun, “A transparent broadband ultrasonic detector based on an optical micro-ring resonator for photoacoustic microscopy,” Sci. Rep. 4, 4496 (2014). [PubMed]
32. C. Y. Chao, S. Ashkenazi, S. W. Huang, M. O’Donnell, and L. J. Guo, “High-frequency ultrasound sensors using polymer microring resonators,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 54(5), 957–965 (2007). [CrossRef] [PubMed]
33. P. Crippa, V. Cristofoletti, and N. Romeo, “A band model for melanin deduced from optical absorption and photoconductivity experiments,” Biochim. Biophys. Acta-General Subjects 538(1), 164–170 (1978). [CrossRef]
34. S. Jacques, Optical Absorption of Melanin (Oregon Medical Laser Center, 1998).
35. T. Sarna and R. C. Sealy, “Photoinduced oxygen consumption in melanin systems. Action spectra and quantum yields for eumelanin and synthetic melanin,” Photochem. Photobiol. 39(1), 69–74 (1984). [CrossRef] [PubMed]
36. T. Berer, E. Leiss-Holzinger, A. Hochreiner, J. Bauer-Marschallinger, and A. Buchsbaum, “Multimodal noncontact photoacoustic and optical coherence tomography imaging using wavelength-division multiplexing,” J. Biomed. Opt. 20(4), 46013 (2015). [CrossRef] [PubMed]
37. M. O. Culjat, D. Goldenberg, P. Tewari, and R. S. Singh, “A review of tissue substitutes for ultrasound imaging,” Ultrasound Med. Biol. 36(6), 861–873 (2010). [CrossRef] [PubMed]
38. E. A. Rossi, P. Rangel-Fonseca, K. Parkins, W. Fischer, L. R. Latchney, M. A. Folwell, D. R. Williams, A. Dubra, and M. M. Chung, “In vivo imaging of retinal pigment epithelium cells in age related macular degeneration,” Biomed. Opt. Express 4(11), 2527–2539 (2013). [CrossRef] [PubMed]
39. T. Ach, C. Huisingh, G. McGwin Jr, J. D. Messinger, T. Zhang, M. J. Bentley, D. B. Gutierrez, Z. Ablonczy, R. T. Smith, K. R. Sloan, and C. A. Curcio, “Quantitative autofluorescence and cell density maps of the human retinal pigment epithelium,” Invest. Ophthalmol. Vis. Sci. 55(8), 4832–4841 (2014). [CrossRef] [PubMed]
40. M. Zareba, T. J. Sarna, and J. M. Burke, “The ratio of melanosome to lipofuscin granule number, not lipofuscin content alone, determines the susceptibility of individual RPE cells to lethal photic stress in vitro,” Invest. Ophthalmol. Vis. Sci. 53, 4273 (2012).
41. S. Ito, A. Pilat, W. Gerwat, C. M. Skumatz, M. Ito, A. Kiyono, A. Zadlo, Y. Nakanishi, L. Kolbe, J. M. Burke, T. Sarna, and K. Wakamatsu, “Photoaging of human retinal pigment epithelium is accompanied by oxidative modifications of its eumelanin,” Pigment Cell Melanoma Res. 26(3), 357–366 (2013). [CrossRef] [PubMed]
42. S. Ye, K. Harasiewicz, C. Pavlin, and F. Foster, “Ultrasound characterization of normal ocular tissue in the frequency range from 50 MHz to 100 MHz,” IEEE Trans. Ultrason. Ferroelectr. Freq. Control 42(1), 8–14 (1995). [CrossRef]
