Designing Gold Nanoparticles to Enhance Colorimetric Perception in Biomedicine

Abstract

Color perception conveys visual information as well as esthetic experience in personal, academic, and occupational settings. In biomedicine, colorimetric point-of-care devices offer rapid, low-cost diagnosis and health monitoring based on optical distinction of interacting biomarkers labeled by e.g., gold nanoparticles. Reliable evaluation and accurate interpretation of readouts from nanoparticle-based colorimetric assays depends on consistent perception of quantitative color attributes such as hue, chromaticity, brightness, and saturation. Yet color perception is highly subjective and varies widely as a result of physical features such as lighting, local environment, and extinction mode as well as biological factors that include genetics, health, and age. This chapter examines contributions of gold nanoparticle size and shape, illumination, sample environment, signal processing, and color vision deficit on quantitative perceptual color difference in order to coordinate a rational framework for design and implementation of gold nanoparticles in biomedical devices to enhance differentiation of analyte-induced changes in nanoparticle-supported color.

1. Introduction

Advances in portable, high-power computing and image acquisition and processing have fueled an explosion in point-of-use, colorimetric chemical/biosensors [1] and point-of-care biomedical in vitro analytics and diagnostics [2, 3]. Design and development of simple, effective, economic, portable biosensors requires attention to, e.g., illumination, labeling, sample environment, signal processing, and user perception. There is a broad range of possible specifications for each attribute. Proliferation of inexpensive, wavelength-tunable, high-intensity light emitting diodes (LED) offer ready alternatives to conventional fluorescent or daylight illumination. Optically-active ions, dyes, chromophores, fluorophores, plasmonic nanocrystals, and quantum dots offer various approaches to harness incident photons by optoelectronic labels. The target bioanalyte can be detected through its interaction with a selected label to produce a color change as a result of absorption and/or scattering modes of extinction. The resulting signal - an observable change in color in the visible spectrum - can be mathematically processed using one of many color spaces to obtain data by which perceptual color difference may be quantitated.

Perception of color differences varies widely. Seven to eight percent of the population (300 million people worldwide) are deficient in their ability to see color or discern differences in color, with deficits ranging from mild to severe [4]. Color vision deficit is largely inherited. Color perception also declines due to many degenerative diseases [5] and to aging [6]. Yet prior consideration of illumination, optoelectronic label, environment, and signal processing in concert with color vision deficit in a comprehensive manner to develop point-of-care, biomedical in vitro analytics and diagnostics appears absent.

The present study examined a novel, integrated approach to prospectively compare effects of selected illuminant, optoelectronic label, sample environment, and color vision deficit on the ability to distinguish perceptual color difference that resulted from optical absorbance in colorimetric point-of-care biomedical sensing using different color spaces. After considering a broadly-used chromaticity diagram, an internationally accepted uniform color space [7] validated with clinical data [8] was adapted, for the first time, to quantitatively process signal acquired from a colorimetric sensor at varying illuminations, nanoparticle geometries, and sample environments, for users with a range of color vision deficits. In this uniform color space, illumination by LED, incandescent, and daylight sources were compared. Labeling by plasmonic gold nanocrystals of various sizes and shapes was considered, due to enhanced intrinsic optical cross section [9] and near-field resolution [10, 11, 12] of plasmon labels. Optical absorption of sample environment was evaluated. Effects on perceptual color difference of color vision deficit that ranged from mild to severe protanomaly and deuteranomaly, respectively, were quantified. Interactive effects between retinal cone receptor, sensitivity, illuminant, label spectra and color vision deficit on perceptual color difference were examined.

This quantitative study of color space, nanoparticle geometry, illumination, sample environment, and color vision deficit resulted in five guidelines to inform selection and specification of components in order to enhance perceptual color difference in colorimetric sensing. Application of these guidelines to design an effective colorimetric nanoparticle sensor was illustrated. The present study, its resulting guidelines, and illustration of application of the guidelines support prospective design and comprehensive specification of key interacting features in colorimetric point-of-care biomedical analytics and diagnostics.

2. Methods

The effect of different plasmon resonant nanocrystal labels on changes in perceived color in order to support a quantitative readout from a colorimetric biomedical point-of-care analytic or diagnostic was examined. Perceived color in the International Commission on Illumination (CIE) XYZ 1931 color space was quantified using optical absorption from 400 to 700 nanometers (nm) for each label to determine chromaticity coordinates (x_i, y_i). The chromaticity coordinates were calculated from tristimulus values using color matching functions as reported by CIE [13]. Perceived color in the CIE Color Appearance Model (CAM) 02 uniform color space was quantified using optical absorption from 400 to 700 nm for each nanocrystalline label to determine color coordinates (a’_i, b’_i). The color coordinates were calculated from tristimulus values using color matching function as reported by Moroney et al. [14] using color-science 0.4.4 [15].

In order to quantify perceptual color difference, the psychophysical limit for a distinguishable change in color was determined in the CIE XYZ 1931 color space at each chromaticity coordinate as represented by a MacAdam ellipse. Each MacAdam ellipse was calculated followed CIE guidelines [16].

Perceptual color difference was used to quantify the difference in color needed to produce a distinguishable change in color to support a visible colorimetric biomedical readout. The change in color perceived from each plasmon resonant label resulted from changes in its optical absorption effected by altering its physicochemical environment. Physicochemical changes resulted in reversible decoupling of localized surface plasmons and decrease in incident phase velocity. One unit of perceptual color difference corresponded to a just noticeable difference in color. Perceptual color difference was calculated in CIE XYZ 1931 color space using Euclidian distance [17]. Perceptual color difference was calculated in CIECAM02 uniform color space using color-science 0.4.4 which was based on a prior report [18].

In order to determine the effect of illumination conditions on perceptual color difference, three representative standard illuminants were examined across the visible spectrum (400 to 700 nm). These illuminants were selected to represent conditions at which point-of-care biomedical analytics and diagnostics were likely to be visualized. The illuminants selected were noon daylight (D65) [19], warm white fluorescence (F4) [13], and full spectrum light emitting diode (LED) [20].

In order to evaluate effects on perceptual color difference of changes in nanocrystalline label size and environmental state, localized surface plasmon resonant absorption spectra were simulated using different illuminants across the visible spectrum (400 to 700 nm). The optical spectra were examined in two environmental states: neat aqueous solution (refractive index = 1.33) and aqueous adsorbate (refractive index = 1.46). The optical spectra were quantified for nanocrystalline label radii from 6 to 18 nm in neat and adsorbate conditions using an online calculator [21] which was based on Mie theory [22].

In order to determine the effect of color vision deficiency on perceptual color difference, the effects of normal and deficient retinal cone receptor sensitivities on perceptual color difference were simulated across the visible spectrum (400 to 700 nm). Color vision deficiency for mild, medium, and severe protanomaly and deuteranomaly was calculated as previously reported [23].

3. Results and discussion

3.1 Color spaces quantitated perceived color

Color perception is the result of complex interplay between quantitative optical and physical factors and subjective psychophysical phenomena. Color spaces use human observer data to quantitatively link spectra at specific optical frequencies to physiologically perceived colors. In colorimetric biomedical point-of-care analytics and diagnostics to date, red-green-blue (RGB) levels are frequently derived from an image or spectra of a sample. These RGB levels are processed to quantify perceived color by identifying the color coordinates in a defined color space. Then perceptual color difference has been measured by calculating the Euclidean distance between these color coordinates. The International Commission on Illumination (CIE) 1931 XYZ color space has been most often used [24]. This color space related visible wavelengths to perceived color based on experimental evaluation of color by seventeen trained observers. Figure 1 shows the chromaticity diagram in the CIE 1931 XYZ color space.

Figure 1.

CIE 1931 XYZ color space. MacAdam ellipses (gray, magnified ten times) define indistinguishable colors. Symbols show perceived color of gold nanorod-dye complexes at neutral pH (green circles) and basic pH (orange triangles).

The CIE 1931 XYZ color space, however, does not uniformly account for variations in color perception that occur across the optical spectrum. Figure 1 includes gray MacAdam ellipses (magnified ten times their actual size) within the chromaticity diagram [25]. The ellipses enclose colors that are indistinguishable to a typical human eye. Just noticeable differences (JND) in chromaticity to a standard observer are defined by boundaries of the ellipse. The area of the ellipses increases with their eccentricity, corresponding to a loss of visual acuity. Euclidian distance between any two points, (x₁, y₁) and (x₂, y₂), in this space, or between the corresponding XYZ chromaticity coordinates, does not represent visually perceived differences in chromaticity [16] .

3.2 Gold nanoparticles supported perceptual color differences

Metal nanoparticles support a localized surface plasmon resonance that can yield brilliant color due to excitation of confined free electrons in the visible spectrum via a frequency-dependent dielectric function [22]. Plasmon extinction cross-sections of metal nanoparticles are 2 to 6 orders of magnitude greater than semiconductor quantum dots, organic fluorophores, and atoms or ions, respectively [9]. Gold nanocrystals, in particular, have extinction cross sections 3 to 8 logs larger than absorption cross sections of organic fluorophores. Plasmonic metal nanocrystals support near-field resolution of 3–100 nm [26, 27, 28] which is up to nearly 10² better than near-field optical resolution of ~200 nm for high numerical apertures (1.4) at high-energy visible light (2.76 eV). These characteristics uniquely qualify gold nanoparticles to support perceptual color differences in colorimetric biomedical point-of-care analytics and diagnostics.

Consider the alternation of perceived color of gold (Au) nanorods (NRs) to which cyanine dye was bound from green to red as pH was cycled from neutral to basic [29]. The AuNRs (91 nm x 30 nm) were synthesized by encapsulation in cationic cetyltrimethylammonium bromide (CTAB) and coated with anionic poly(sodium 4-styrenesulfonate) (PSS) to which cationic, lipophilic 1,1′,3,3,3′,3′-hexamethylindoletricarbocyanine iodide (HITC) was bound. Cycling pH from neutral to basic reversibly decoupled localized surface plasmons on the PSS-coated AuNR (λ_max, = 732 nm) from conjugated double bonds on HITC (λ_max, = 736 nm) by attaching a hydroxyl group to HITC that disrupted the conjugated bonds. Alternating perceived colors of the AuNR-CTAB-PSS-HITC complex (hereafter referred to as AuNR-cyanine dye) from green (circles, neutral pH) to red (triangles, basic pH) for cycles 1, 2, and 3 were plotted in CIE 1931 XYZ color space in Figure 1.

The three pairs of color points (x_i, y_i) for each of three neutral (green) to basic (red) pH cycles were shown in the Figure 1 inset. These color pairs illustrated inaccuracies in use of Euclidian distance in the CIE 1931 XYZ color space to determine perceptual differences in chromaticity. Gray outlines of 1.18-step MacAdam ellipses surrounded each of the three neutral points as well as basic 1 point. The outlines identified JND thresholds at 50% probability. Each neutral (green) color was perceptually distinct. Conversely, all basic (red) colors were colorimetrically indistinguishable, as they fell within one JND ellipse perimeter. Although the Euclidian distance between basic 1 and basic 2 was comparable to that between neutral 1 and neutral 2, the former (red) colors were perceptually indistinguishable whereas the latter (green) colors were differentiable. This illustrated that Euclidian distance in CIE 1931 XYZ color space does not accurately represent perceptual color differences in colorimetric biomedical point-of-care analytics and diagnostics.

3.3 Uniform color spaces quantitated perceptual color differences

Technological advances in e.g., inks, illuminated displays, and digital recording necessitated development of improved, uniform color spaces (UCSs) to accurately quantify perceptual color differences. A UCS accounts for quantum catch, phototransduction, chromatic adaptation, and receptor opponency to yield a psychophysical color space validated by behavioral and clinical data [8, 30, 31]. Superiority of the UCS based on the CIE color appearance model (CAM) adopted by CIE in 2002 [14], known as the CIECAM02 UCS, has led to its broad adoption [18]. The present study examined quantitation of perceived color difference (ΔE’) in CIECAM02 uniform color space [18].

Figure 2 quantified perceived color difference (ΔE’) between pairs (N-B i) of AuNR-cyanine dye complexes at neutral (N) pH and basic (B) pH obtained in three (i) consecutive cycles. For comparison, values of ΔE’ were calculated using CIE XYZ 1931 color space (XYZ, green) and CIECAM02 uniform color space (UCS, orange). Calculated values of ΔE’ decreased from 8.2 to 4.5 to 2.4 in CIE XYZ 1931 color space and from 9.8 to 6.5 to 4.0 in CIECAM02 UCS in three consecutive cycles. These consecutive decreases in ΔE’ reflected the observation that each color pair (N-B i) became perceptually closer at consecutive cycles, as shown in Figure 1 inset. The relative changes in ΔE’ between the 1st, 2nd and 3rd cycles in CIECAM02 UCS, however, were smaller than in the Euclidian distance in CIE XYZ 1931 color space. This is a consequence of nonuniform representation of perceived color difference in CIE XYZ 1931 color space. Specific values for relative changes in ΔE’ were −45% between cycles 1 and 2, and −48% between cycles 2 and 3 in CIE XZY 1931 color space. Values for relative changes in ΔE’ were −34% between cycles 1 and 2 and −38% between cycles 2 and 3 in CIECAM02 UCS.

Figure 2.

Perceptual color differences between three pairs (N-B) of gold nanorod-dye complexes at alternating neutral (N) pH and basic (B) pH calculated using CIE XYZ 1931 color space (XYZ, green) and CIECAM02 uniform color space (UCS, orange).

Uniform representation of perceived color difference in CIECAM02 UCS was illustrated in Figure 3, for comparison to Figure 1 inset. Three pairs of color points (x_i, y_i) for each of three neutral (green) to basic (red) pH cycles were plotted. With reference to colors observed in the CIE XYZ 1931 chromaticity diagram, Cycle 1 alternated the color of AuNR-cyanine dye from deeper green (neutral 1, cyan diamond) to red (basic 1, blue diamond). Cycle 2 alternated the AuNR-cyanine dye color from a moderate green (neutral 2, lime triangle) to red (basic 2, green triangle. Cycle 3 alternated the AuNR-cyanine dye color from a light green (neutral 3, crimson circle) to red (basic 3, orange circle). In CIECAM02 UCS, colors of the three basic pH samples aligned less perpendicularly to neutral points 2 and 3 than in the CIE 1931 XYZ color space, where orientation of basic (red) colors was distorted alongside eccentric JND ellipses. This reduction in chromatic distortion in the CIECAM02 UCS resulted in smaller relative changes in ΔE’ between the 1st, 2nd and 3rd cycles. The abscissa (x-axis) in CIECAM02 UCS is the correlate for green-red (a’) which represents the departure from yellow. The ordinate axis is the correlate for cyan-yellow (b’) which represents the departure from red.

Figure 3.

CIECAM02 uniform color space. Symbols show perceived color of gold nanorod-cyanine dye complexes as pH was changed from neutral to basic in three consecutive cycles.

3.4 Illuminant affected perceptual color difference

The optical spectrum of illumination on a colorimetric biomedical point-of-care analytic or diagnostic sample has a fundamental effect on color differences perceived by an observer. Figure 4 showed optical distributions for CIE standard illuminants that represent noon daylight (D65) and warm white fluorescence (F4) and for a full spectrum light emitting diode (LED). The distributions were scaled to a value of 1 at λ = 650 nm. The correlated color temperatures (CCTs) of D65, F4, and LED were 6504, 2940, and 6500 K, respectively. D65 has been canonically recommended for colorimetric calculations that require representative daylight [32]. F4 represents a warm white calcium halophosphate fluorescent lamp. It was used to calibrate the CIE color rendering index (CRI), having a CRI value of 51 [33]. The full spectrum LED with a CCT of 6500 K, known as ‘cool white’, provides bright, bluish-white illumination that is useful for photography, display screens, and crisp indoor lighting.

Figure 4.

CIE standard illuminants for noon daylight (D65) and warm white fluorescence (F4) and full spectrum light emitting diode (L30).

The effect of illuminant on the color difference perceived by a normal observer for AuNR-cyanine dye was summarized in Figure 5. Color differences were calculated in CIECAM02 UCS between the three pairs of gold nanorod-cyanine dye complexes at neutral (N) pH and basic (B) pH using illuminants D65, F4, and LED. Noon daylight (D65, blue circles) provided the largest pairwise perceptual color differentiation for the neutral to basic AuNR-dye transition. The (unitless) values for ΔE’ under D65 illumination were 9.8, 6.5, and 4.0 for cycles 1, 2, and 3, respectively. Illumination by either the halophosphate fluorescent F4 source or the full spectrum LED source reduced the perceptual color difference of each N-B pH cycle.

Figure 5.

Perceptual color differences between three pairs of gold nanorod-dye complexes at alternating neutral (N) pH and basic (B) pH illuminated by noon daylight (D65), warm white fluorescence (F4), and full spectrum LED (LED). Perceptual color difference (ΔE’) in each case was calculated using CIECAM02 UCS.

Full spectrum LED illumination (green circles) resulted in one unit less perceptual color differentiation than D65, providing 8.9, 5.6, and 3.2 for cycles 1, 2, and 3, respectively. Warm white fluorescence illumination (orange circles) resulted in a reduction to almost one-third the perceptual color difference of D65, i.e., 3.4, 2.1, and 1.2 for cycles 1, 2, and 3, respectively. In warm white fluorescence light, the third green (neutral) to orange (basic) transition of AuNR-cyanine dye complex was just 0.2 units above the limit of perceptual color difference.

The reduction in perceptual color difference due to illumination was attributed to low illumination at optical wavelengths to which retinal cones were sensitive to changes in color. The sensitivities of retinal cones across the optical spectrum were plotted in Figure 6. A standard normal observer perceives color by use of short- (s, blue solid line), medium- (m, green solid line) and long-wavelength (l, red solid line) retinal cones.

Figure 6.

Spectral sensitivities of short (s, blue solid line), medium (m, green solid line) and long (l, red solid line) retinal cones for a standard observer. Change in absorbance (right-hand y-axis) for AuNR-cyanine dye from neutral (coral dashed line) to basic (gold dashed line) pH in cycle 3. Red-green color vision deficit occurs due to deuteranomalous red-shift of m-cones (olive dash-dot) or protanomalous blue-shift of l-cones (crimson dash-dot).

Figure 4 showed that there was minimal illumination by F4 at energies lower than 650 nm or higher than 550 nm. But retinal m- and s-cones are particularly sensitive to energies higher than 550 nm. In Figure 6, pH-induced changes in absorbance spectra from neutral (salmon dotted line) to basic (gold dotted line) for cycle 3 in AuNS-cyanine dye occurred primarily at energies higher than 450 nm or lower than 570 nm. pH-induced spectral changes in cycles 1 and 2 were similar, occurring primarily at wavelengths <450 nm and > 570 nm. Yet, illumination by warm white fluorescence was relatively low in both of these spectral regions. Conversely, LED illumination was relatively high >450 nm, and daylight was relatively uniform across the spectrum. Thus, LED and D65 illumination enabled larger perceptual color differences than F4.

3.5 Color vision deficit affected perceptual color difference

To date, perceptual color difference in analysis of colorimetric biomedical point-of-care analytics and diagnostics appears to have been examined only for a standard normal observer. Figure 6 illustrated retinal sensitivities of short, medium, and long-wavelength cones to incident light for a standard observer across the 400 to 700 nm optical spectrum as classified by the CIE action spectra. Meanwhile, more than 12 million people in the U.S. (300 million people worldwide) are deficient in their ability to see color or discern differences in color [4, 34]. This negatively impacts their quality of life for health, emotions, and especially careers [35, 36]. Color vision deficit (CVD) causes difficulty selecting and preparing food, driving, choosing, clothing, and taking medications [37]. Yet deleterious effects of CVD on perceptual color difference in colorimetric biomedical point-of-care analytics and diagnostics appears not to have been considered.

Red-green color vision deficit is the most common. About 75% of those with red-green CVD classify as deuteranomalous [38]. Figure 6 illustrated that deuteranomaly is due to red-shifting of the m-cone (olive dash-dot), while protanomaly is due to a blue-shifted l-cone (crimson dash-dot). Either condition increases overlap of m- and l-cones which results in increasing red-green color confusion. The severity of red-green CVD is quantifiable as the fraction of red-shift (to higher wavelengths/lower energies) or blue-shift (to lower wavelengths/higher energies) for deuteranomaly and protanomaly, respectively, between zero (normal) and full red-green colorblindness [23]. Red-shifts of 0.25, 0.5 and 0.75 correspond to mild, moderate, and severe deuteranomaly. Classification of blue-shifts of protanomaly are analogous.

Red-green CVD decreased perceptual color differences between the three pairs of AuNR-cyanine dye complexes at alternating neutral (N) and basic (B) pH. This resulted in part from the green color of AuNR-dye suspension at neutral pH and its red color at basic pH. Figure 7 summarized values of perceptual color difference, ΔE’, for each N-B pair. The ΔE’ values decreased with the severity of CVD from mild (0.25_j) to moderate (0.5_j) to severe (0.75_j) protanomaly (subscript p) and deuteranomaly (subscript d), respectively. Mild protanomaly and deuteranomaly decreased ΔE’ by comparable amounts. But moderate and severe protanomaly reduced ΔE’ by up to one JND unit more than moderate and severe protanomaly. Values for ΔE’ in Figure 7 were calculated in CIECAM02 UCS under noon daylight (D65) illumination.

Figure 7.

Red-green color vision deficit decreased perceptual color differences between three pairs of gold nanorod-dye complexes at alternating neutral (N) and basic (B) pH. ΔE’ decreased with the severity of protanomaly (subscript p) and deuteranomaly (subscript d). ΔE’ was calculated in CIECAM02 UCS at noon daylight illumination.

Red-green CVD reduced perceptual color difference relative to a normal observer in part because pH-induced changes in AuNS-cyanine dye color occurred outside the region of high red-green color sensitivity near 560 nm. Figure 6 showed the change in absorbance, (plotted on the right-hand y-axis) for AuNR-cyanine dye from neutral pH (coral dashed line) to basic pH (gold dashed line) pH in cycle 3. Changes of absorbance in cycles 1 and 2 occurred at similar spectral regions. These regions of absorbance change lie outside the region of highest red-green color sensitivity, near 560 nm. In the spectral region from 450 to 650 nm affected by red-green CVD, the change in absorbance between neutral 3 and basic 3 was larger above 600 nm where sensitivity of l-cones (red solid line) exceeded that of m-cones (green solid line). Protanomaly therefore accrued a larger decrease in ΔE’ than deuteranomaly. The lowest ΔE’ was for severe protanomaly, which decreased to slightly more than two JND units.

3.6 Effects on perceptual color differenced by CVD and illuminant were interactive

Color vision deficit can compound potentially deleterious effects of illuminant on perceptual color difference. When illuminated by warm white fluorescent light (F4), perceptual color differences between the three pairs of AuNR-cyanine dye complexes at alternating neutral (N) and basic (B) pH decreased below the JND threshold when perceived by an individual with red-green CVD. Figure 8 showed values of ΔE’ for each N-B pair decreased with the severity of CVD from mild (0.25_j) to moderate (0.5_j) to severe (0.75_j) protanomaly (subscript p) and deuteranomaly (subscript d), respectively. Values for ΔE’ were calculated in CIECAM02 UCS warm white fluorescent (F4) illumination. Protanomaly decreased ΔE’ more than protanomaly at all levels. Higher severity of red-green CVD enlarged the decrease. Perceptual color difference between the N-B pair at the third cycle was at or below the JND threshold for mild, moderate, and severe protanomaly. As observed in Figure 4, this was attributable to vanishing illumination by F4 at wavelengths shorter than 550 nm This nearly extinguished light to which m-cones are most sensitive, and decreased by about one-half the spectral frequencies to which m- and l-cones are sensitive.

Figure 8.

Fluorescent illumination exacerbated the reduction in perceptual color differences between three pairs of gold nanorod-dye complexes at alternating neutral (N) and basic (B) pH due to red-green color vision deficit. ΔE’ decreased with the severity of protanomaly (subscript p) and deuteranomaly (subscript d). ΔE’ was calculated in CIECAM02 UCS at warm white fluorescent illumination.

3.7 Sample environment affected perceptual color difference

Besides absorption by colorimetric labels like AuNR, optical absorption by other components of the sample and/or its environment in a biomedical point-of-care assay may affect perceptual color difference. Polydimethylsiloxane (PDMS) is widely used in lab-on-chips for analytic and diagnostic methods. Figure 9A showed that optical absorption by polydimethylsiloxane (PDMS) decreased ΔE’ for normal and CVD observers, relative to a transparent environment shown in Figure 7. The decrease in ΔE’ was 2.9, 3.1, and 3.0 percent for normal, protanomaly, and deuteranomaly, respectively. Figure 9B showed that relatively constant PDMS transmission across the visible spectrum between 90 and 94% resulted in nearly uniform reductions in ΔE’ for normal and CVD observers.

Figure 9.

Optical absorption by polydimethylsiloxane decreased perceptual color differences between three pairs of gold nanorod-dye complexes at alternating neutral (N) and basic (B) pH. ΔE’ decreased with the severity of protanomaly (subscript p) and deuteranomaly (subscript d). ΔE’ was calculated in CIECAM02 UCS at noon daylight illumination subject to PDMS absorption.

3.8 Designing gold nanoparticles to enhance colorimetric perception in biomedicine

Evaluation of the effects on quantitation of perceptual color difference that result from choice of color space, plasmonic labels, illuminant, and sample environment, as well as widely occurring color vision deficit suggested five considerations in the design of gold nanoparticle labels to enhance colorimetric perception in biomedical point-of-care analytics and diagnostics. The proposed design guidelines are:

1. Use a uniform color space to quantitate perceived color and perceptual color difference in order to account for visually perceived differences in chromaticity that occur across the visible spectrum.

2. Use metal nanoparticles with localized surface plasmon resonance at optical frequencies in order to enhance extinction cross section and near-field resolution relative to alternative labels such as organic fluorophores, semiconductor quantum dots, atoms, and or ions.

3. Illuminate the sample at optical wavelengths to which retinal cones are sensitive to targeted color changes in order to maximize perceptual color difference relative to induced color change.

4. Specify geometry (size, shape) and dielectric environment of a plasmon resonant label that exhibits maximal absorbance near 560 nm in order to increase m- and l-cone sensitivity and to reduce effects of red-green color vision deficit.

5. Reduce absorption by sample or its environment in a biomedical point-of-care assay in order to maximize perceptual color difference.

Use of these five guidelines was examined in the following design illustration that considered using a change in refractive index adjacent to gold nanosphere (AuNS) as a result of, e.g., macromolecular binding in order to achieve high perceptual color difference in a biomedical point-of-care analytic or diagnostic assay.

‘Design guideline 2’ indicated use of gold (Au) nanospheres (NS) ranging from 6 nm to 18 nm in radius. For AuNS with radii in this range, optical absorption exceeds extinction by ≥10-fold. Figure 10 compared Mie spectra from AuNS with radii of 6, 12, and 18 nm, respectively. Initially the AuNS were taken to be immersed in water (H₂O, refractive index, RI = 1.33). Spectra for these AuNS are shown at the left of Figure 10. Increasing the radius of AuNS in increments of 6 nm, from 6 to 12 to 18 nm, red-shifted each consecutive wavelength of the corresponding spectral maximum, λ_max, due to increased retardation of the localized surface plasmon resonance (LSPR) [39]. The magnitude of the redshift of λ_max as a result of increasing size occurred in increments of 2.6 and 2.9 nm, respectively, between 6, 12, and 18 nm AuNS in H₂O.

Figure 10.

Mie spectra of 6, 12, and 18 nm radius gold nanospheres immersed in water before (H₂O, left) and after adsorption of e.g., protein (Pr, right).

‘Design guideline 4’ indicated to target a λ_max near 560 nm. A large AuNS radius, 18 nm, was selected for which absorbance was ≥10 times scattering when, e.g., protein analyte (Pr, RI = 1.46) adsorbed onto the AuNS. Figure 10 showed that increased retardation at larger radii for Pr-covered AuNS red-shifted their LSPR λ_max in increments of 0.5 and 5.4 nm, respectively, between 6, 12, and 18 nm. In addition, the increase in RI from H₂O (1.33) to Pr (1.46) redshifted the spectral λ_max for each size of AuNS as the matching frequency of incident light shifted to longer wavelength (lower energy) as incident phase velocity decreased [40]. This Pr-induced redshift of λ_max due to the change in local dielectric environment was 11.5, 9,4, and 11.9 nm, respectively, for 6, 12, and 18 nm AuNS. The Pr-induced dielectric redshift averaged 10.9 nm.

‘Design guideline ‘3’ indicated to illuminate the AuNS using full spectrum LED, which had the highest relative illumination in the region of AuNS color change. The color change occurred in the region of maximum retinal red-green cone sensitivity.

‘Design guideline 5’ indicated to eliminate any optically absorbing components in the sample environment, e.g., PDMS. Retrospective analysis showed that inclusion of PDMS in the optical path reduced the average ΔE’ by 2.1 percent for a normal observer.

‘Design guideline 1’ indicated to quantitate perceptual color difference between non-adsorbed (H₂O) and protein-adsorbed (Pr) AuNS in this design for a biomedical point-of-care assay using CIECAM02 UCS [18]. Figure 11 showed that perceptual color differences, ΔE’, between the three AuNS before and after adsorbed analyte red-shifted the LSPR were 8.1, 8.4, and 9.1 for a normal (Norm) trichromatic observer for 6, 12, and 18 nm AuNS, respectively. Notably, the RI-induced perceptual color differences for 6, 12, and 18 nm AuNS were greater than the pH-induced perceptual color differences for AuNR-cyanine dye complexes in the 2nd and 3rd cycles and just one JND below that of cycle 1. Importantly, the RI-induced perceptual color differentiation in AuNS was comparable to (at cycle 1) or better than (after cycle 1) AuNR-cyanine dye without increasing AuNR size, encapsulation in toxic CTAB, coating with PSS, or complexation with cytotoxic HITC.

Figure 11.

Perceptual color differences, ΔE’, between three gold nanospheres of radius 6, 12, and 18 nm, respectively, before and after adsorbed analyte red-shifted the localized surface plasmon resonance. Red-green color vision deficit increased perceptual color differences relative to a normal observer (norm) as severity of protanomaly (subscript p) and deuteranomaly (subscript d) increased from 0.25 to 0.75. ΔE’ was calculated in CIECAM02 UCS at full spectrum LED illumination.

The relative changes in ΔE’ between the 6, 12 and 18 nm AuNS in CIECAM02 UCS were smaller than in the Euclidian distance in CIE XYZ 1931 color space. This was a consequence of nonuniform representation of perceived color difference in CIE XYZ 1931 color space. Specific values for relative changes in ΔE’ were 2.1% between 6 and 12 nm, and 5.0% between 12 and 18 nm in CIECAM02 UCS. Values for relative changes in ΔE’ were 5.6% between 6 and 12 nm, and 8.4% between 12 and 18 nm in CIE XYZ 1931 color space.

Red-green CVD increased perceptual color differences relative to a normal observer as the severity of protanomaly (subscript p) and deuteranomaly (subscript d) increased from mild (0.25) to severe (0.75). From data in Figure 11, Table 1 summarized the percent increase in perceptual color difference due to mild (0.25_j), medium (0.50_j), and severe (0.75_j) protanomaly (j = p) and deuteranomaly (j = d), respectively. The increase in perceptual color difference due to red-green CVD ranged from 7.6% for the color difference of 12 nm AuNS after protein adsorption perceived by a mildly protanomalous observer to 27.6% for the color difference of 12 nm AuNS after protein adsorption perceived by a severely protanomalous observer. Inclusion of PDMS in the optical path reduced the average ΔE’ observed by a protanomalous or deuteranomalous observer by 2.5%.

	6 nm	12 nm	18 nm
0.25p	7.6	7.5	7.1
0.50p	16.4	16.5	15.9
0.75p	26.8	27.5	27.3
0.25d	8.1	8.6	9.0
0.50d	14.5	15.9	17.7
0.75d	15.8	18.8	23.5

Table 1.

Increase in perceptual color difference relative to normal trichromatic vision due to color vision deficit (%).

The CVD-enabled increase in ΔE’ for AuNS after Pr adsorption was attributed to (1) a reduction in color confusion due to optical absorption of AuNS in the spectral region near 560 nm where m- and l-cone sensitivities most overlap; and (2) relatively higher illumination by broad spectrum LED in the region of RI-induced color change. Reduction of color confusion by AuNS has led to their use in lenses to manage color blindness [41]. Six and 20-nm radii AuNS were reported to manage color blindness more effectively than 40-nm AuNS.

From values of CIECAM02 UCS perceptual color difference, a sensitivity to change in refractive index that red-shifted LSPR spectra of these plasmonic AuNS was estimated. Sensitivity to refractive index change was calculated as ΔE’/ΔRI. Values for RI sensitivity of normal perception for 6, 12, and 18 nm AuNS illustrated in Figure 11 were 63, 66, and 71 RIU⁻¹, respectively. Sensitivity for 6, 12, and 18 nm AuNS increased to 86, 91, and 97 RIU⁻¹, respectively, for severe protanomalous perception. These RI sensitivities in broad spectrum LED illumination were up to 24% higher than RI sensitivities for RI-induced AuNP color change in daylight illumination.

3.9 Benefits of guidelines to design gold nanoparticles to enhance colorimetric perception in biomedicine

Evaluation of perceptual color difference based on variation of refractive index adjacent to gold nanospheres illustrated benefits for each guideline proposed to design colorimetric biomedical point-of-care analytics and diagnostics.

Uniform color space: Quantitative evaluation of RI-induced perceptual color differences due to analyte adsorption on AuNS illustrated that CIECAM02 UCS, a uniform color space, accounted for variations in perceived chromaticity across the visible spectrum. For AuNR-cyanine dye, this UCS gave smaller relative changes in perceptual color difference compared to CIE XYZ 1931 color space, a nonuniform color space.

Plasmonic nanoparticle geometry: Selection of a size and shape of nanoparticle to locate its localized surface plasmon resonance spectra near the color confusion region of the retinal cone sensitivities enhanced perceptual color difference. This avoided e.g., use of large nanoparticle, toxic encapsulation, chemical coating, and/or cytotoxic dye resonator to achieve comparable or lower perceptual color differences.

Illumination: Selection of illumination by broad spectrum LED enhanced perceptual color difference and its sensitivity to changes in refractive index relative to daylight or fluorescent illumination.

Relatively higher illumination by LED lighting in the region of red-green color confusion where RI-induced color changes occurred due to size/shape selection of AuNP resulted in increased perceptual color difference in protanomalous and deuteranomalous observers.

Component absorption: Elimination of absorption by components in the sample matrix or surrounding environment increased perceptual color difference by a few percent in the case of PDMS, for which transparency was 90–94% in the optical spectrum.

4. Conclusions

Five guidelines to design colorimetric biomedical analytics and diagnostics based on plasmonic nanoparticles were proposed. The guidelines were drawn from quantitative evaluation of perceptual color differences when spectra of gold nanoparticles of different sizes and shapes were changed by altering pH- or local refractive index. Effects of different color spaces, illuminants, and absorbing components were considered. Alteration of perceptual color difference due to color vision deficit was examined.

Use of the CIECAM02 uniform color space was shown to account for spectral variations in perceived color, thereby improving quantitation of perceptual color difference compared to e.g., CIE XYZ 1931 chromaticity diagram. The selection of nanoparticle size and shape to locate its LSPR spectra near the color confusion region of the retinal cone sensitivities was shown to enhance perceptual color difference for both normal and color vision deficit observers. Relatively intense illumination of the sample at optical wavelengths where retinal cones are sensitive to targeted color changes additionally increased perceptual color difference.

Co-locating intense illumination and LSPR spectra of nanoparticle labels with cone sensitivity was shown to increase perceptual color difference for color-vision-deficient observers with mild, moderate or severe protanomaly and/or deuteranomaly relative to normal observers. Minimizing optical absorption by sample components or environment improved perceptual color difference.

Based on a uniform, quantitative description of color perception and perceptual color difference, these guidelines enable selection of color space, nanoparticle type and geometry, illumination, and absorbing components to maximize perceptual color differences for both normal and color vision deficient observers in colorimetric biomedical point-of-care analytics and diagnostics.

Acknowledgments

This research work was supported by NIH R15 EY035066.

Conflict of interest

“The author declares no conflict of interest.”

Notes/thanks/other declarations

The author thanks Dr. Robert Koprowski, Dr. Luis Jesus Villarreal-Gomez, and Mr. Dominik Samardzija for the invitation to contribute a chapter to “Biotechnology - Biosensors, Biomaterials, and Tissue Engineering - Annual Volume 2024”, part of the “Biomedical Engineering” book series.

Nomenclature

a’	departure from yellow
Au	gold
B	basic
b’	departure from red
CAM	color appearance model
CCT	color correlated temperature
CIE	International Commission on Illumination
CRI	color rendering index
CTAB	cetyltrimethylammonium bromide
CVD	color vision deficit
D65	noon daylight illumination
ΔE’	perceptual color difference
F4	warm white fluorescence
HITC	1,1′,3,3,3′,3′-hexamethylindoletricarbocyanine iodide
JND	just noticeable distance
LED	light emitting diode
N	neutral
nm	nanometers
NS/NR	nanosphere/nanorod
PSS	poly(sodium 4-styrenesulfonate)
UCS	uniform color space