Paper The following article is Open access

Quantitative comparison of luminescence probes for biomedical applications

, , , , , and

Published 9 July 2021 © 2021 The Author(s). Published by IOP Publishing Ltd
, , Citation B Krajnik et al 2021 Methods Appl. Fluoresc. 9 045001 DOI 10.1088/2050-6120/ac10ae

2050-6120/9/4/045001

Abstract

Optical imaging holds great promise for the early-stage detection of diseases. It plays an important role in the process of protecting the patient's health. Most of the organic dyes suffer due to photobleaching, light scattering, short light penetration depth, and autofluorescence of specimen, thus, need to be replaced with alternative nanoprobes emitting light in the optical biological window (700–1350 nm). The group of candidates which can challenged described problems are colloidal quantum dots (e.g. CdSe and PbS) and upconverting nanocrystals (e.g. NaGdF4:Er, Yb). This paper presents comprehensive and systematic studies of the aforementioned probes, using specially designed tissue phantom, and custom-built wide-field fluorescence microscope. We investigated how the absorption and scattering of light at the water, hemoglobin, and intralipid may affect the intensity of luminescence probes and the quality of optical images. We propose a protocol, that could be easily implemented for investigating other nanoprobes that allow for comparison of their optical performance.

Export citation and abstract BibTeX RIS

Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

1. Introduction

Development of the biodegradable and biocompatible luminescence probes that could be functionalized and applied to living organisms is an important goal in biomedical research [1]. Most bioimaging research is related to in vitro imaging of cells and tissues or in vivo investigation of tissues close to the skin surface. That is why one of the most important parameters which need to be considered while choosing the optimal luminescence probe is light penetration depth. Scattering and absorption of light limit the applicability of most of the luminescence probes to the depth of ∼1 mm [2, 3]. To break that barrier, it is necessary to use nanoprobes optically active within the so-called 'biological window'. It is divided into three distinctive spectral regions: so-called first biological window (NIR-I, 700 nm − 950 nm), second biological window (NIR-II, 1000 nm −1350 nm), and the third biological window (NIR-III, 1550 nm − 1870 nm) as shown in figure 1 [47]. In these spectral regions, the absorption rates of water, hemoglobin, fats, and other common tissue components are the lowest. Tissue imaging with the use of nanoprobes optically active at the 'biological windows' increases not only the light penetration depth but also image contrast and reduces the effect of autofluorescence.

Figure 1.

Figure 1. (a) Normalized emission spectra of the investigated optical probes. Each luminescence spectra was labeled with the name of probe material and central wavelength of emission, where R6G is rhodamine 6G, CdSe - cadmium selenide QDs, PbS - lead sulfide QDs, UPC - up-converting nanoparticles of NaGdF4:Er, Yb emitting within the visible (UPC VIS) and near-infrared regions (UPC NIR). (b) Attenuation spectra of pure deionized water (DI) and solution of DI and various concentration of hemoglobin (H) and intralipid (I). Inset to this panel shows water attenuation peak versus 980 nm laser excitation wavelength.

Standard image High-resolution image

The first water-soluble D-A-D (Donor-Acceptor-Donor) NIR-II dye was CH1055, developed by Antaris and Chen, and co-workers. Despite low QY ∼0.5% PEGylated CH1055 allows for imaging of brain tumor at 4 mm depth [8]. The further optimization of molecular fluorophores led to the development of IR-FTAP with QY of 5.3% [9, 10]. Another group of fluorophores optically active at the 'biological windows' are nanoprobes emitting in near-infrared (NIR) e.g. colloidal QDs and upconverting nanocrystals (UPNCs). These probes have recently grown as a strong alternative to organic dyes [1114]. Quantum dots are semiconducting nanostructures with narrow and continuously tunable emission which changes with the QD size due to the quantum confinement effect. The typical diameter of QDs varies between 2 and 15 nm. QDs are composed of atoms from II-VI (CdSe, CdTe, CdS, ZnS, ZnSe), III-V (GaAs, GaN, InP, InAs), and IV-VI (PbS, PbSe) groups of the periodic table [15]. Recently, also other materials like organic-inorganic (Pb-based) perovskites [16] or non-stoichiometric compounds of Cu2-xSe [17] are used in QDs design.

In comparison to organic dyes used for biological applications, QDs have many advantages like low photobleaching, broad absorption spectra ranging from UV up to NIR region, and large effective Stokes shifts [18, 19]. In the past several years, there have been many scientific reports of successfully implementing QDs within the bioimaging area, which gave for example the possibility of using CdTe/ZnSe QDs in vitro imaging of animal tissue and embryo [20] or the PbS/CdS/ZnS QDs as a multifunctional platform for in vivo near-infrared fluorescence imaging [11]. Another promising luminescence probe emitting in the NIR-II region is Ag2S QDs. They feature high QY (5.8% and 15.5% when functionalized) and low toxicity. No significant toxic effect was found over the period of 2 months [21]. Also, the UPNCs show great potential for imaging and detection in both in vivo and in vitro applications [22]. This new generation of fluorophores has the possibility of converting infrared light (λexc = 980 nm) into visible spectra, which is called the process of upconversion. Unique luminescence properties of UPNCs such as large anti-Stokes shifts, narrow emission bands, and lack of photobleaching makes them a good candidate as a nanoprobe [20]. Due to these unique properties multifunctional UPNCs made of NaYF4:Yb3+, Er3+ core, and NaGdF4 shell were used to target cancer cell nuclei and also to deliver the anticancer drug directly to the nuclei region [23]. Zeng et al were able to construct a multi-functional nanoprobe consisting of PEG-modified BaGdF5:Yb3+, Er3+ UPNCs used as contrast agents for improving the detection of splenic diseases [24].

One of the major research directions in the field of biophotonics is the development of nanoprobes, that are optically active in the NIR region. They hold a promise of high-contrast imaging, deeper tissue penetration by excitation light, lower light scattering, higher luminescence intensity, and lack of autofluorescence [11, 25]. Nevertheless, the abovementioned features are defined mostly in a qualitative way. These properties should be measured with a well-defined protocol, giving the possibility to compare one probe with another. In many cases, nanoprobes are investigated in vivo on animals, however, most of that research is pass or fail tests. Other authors present experiments based on tissue phantoms, which mimic specific tissue properties like for example human skin. For the simulation of light scattering and absorption in tissues, scientists use agarose based solutions [23, 26], polystyrene microspheres [27], or intralipid [25, 28], black ink [28], or hemoglobin-based solutions [25]. In most cases, only one luminescence probe is studied. Because of specific experimental conditions in different laboratories and the use of various techniques, these results are not comparable. Benayas et al presented a series of experiments based on three different NIR emitting QDs which emission passed through chicken breast tissue of variable thickness [11]. Analyzing emission intensity as a function of tissue thickness, they determined the emission penetration depth of various nanoprobes. By approximation of the data with a single exponential function, they calculated the mean penetration depth. The best result was obtained for QD with an emission band centered at 1270 nm with a mean penetration depth of 5.2 mm. These probes were additionally tested by in vivo imaging on mice, during which the real-time biodistribution was recorded.

The most detailed research on emission properties of luminescence probes was done by Nayoun Won et al [12]. They investigated tissue samples and tissue phantoms in conditions, which simulated in vivo experiments. The recorded images were analyzed using a specific algorithm, which focused on both the intensity and the shape of the luminescence probe image. Moreover, the influences of photoluminescence quantum efficiency, nanoprobes concentration, and fluence rate were considered. Despite wide and insightful investigation only two samples of QDs emitting in the NIR region were studied, having luminescence peaks at 800 nm and 1300 nm. The mean penetration depths were calculated for both NIR emitting QDs to be 8.5 mm and 11.6 mm, respectively.

In this work, we performed a systematic overview of the optical properties of several popular luminescence probes, from the three material groups, in which emission spectra cover the range from 500 to 1600 nm. To properly determine the optical properties of nanoprobes and correctly estimate the penetration depth, we proposed a method of quantitative analysis of luminescence probes performance. We focused on the absorption of water, hemoglobin, and intralipid that mimic different aspects of the optical properties of living tissues. All the results are brought together and analyzed with a unified protocol that can be easily implemented and used for comparison of other luminescence probes of promising properties.

2. Results and discussion

2.1. Methodology

To choose the most suitable nanoprobe for bioimaging, the quantitative values of optical properties of fluorescence probes are needed. The fundamental parameter is the energy of the luminescence peak. In our study, we chose a set of probes having emissions from 500 nm to 1600 nm as shown in figure1.(a). These are rhodamine 6G (R6G, QY 95%), cadmium selenide (CdSe, QY 65%) QDs, three lead sulfide (PbS, QY 15%–20%) QDs with various sizes, and UPNCs (NaGdF4:Er, Yb, QY ∼1%). All samples, except UPNCs, were excited at 532 nm. In the case of UPNCs, the excitation wavelength was 980 nm and the luminescence of Er3+ ions is both in the visible (UPC VIS) and near-infrared (UPC NIR) spectral range. Every fluorescence probe was labeled with the following convention: material@luminescence_peak. The only exception is UPC VIS@535&660 nm, where both 'green' and 'red' luminescence is present.

We designed a dedicated tissue phantom based on deionized water (DI) solutions. In many other scientific reports, scientists chose the phosphate-buffered saline (PBS), which based on our preliminary experiments, gives similar results to DI (see Fig. S1 in supplementary materials (available online at stacks.iop.org/MAF/9/045001/mmedia)). We simulated tissue absorption and scattering properties using hemoglobin (H) and intralipid (I) dispersed in DI. The concentration of both varies from 0% to 2%. 2% concentration of hemoglobin in DI approaches the maximum content of hemoglobin in the blood, while the solution of 2% intralipid in DI scatters light more than human skin [12]. In our work, we use hemoglobin solutions as a purely absorbing model and intralipid solutions as a purely scattering model. For both types of phantoms, we prepared solutions of H and I of various concentrations 0.25%, 0.5%, 1%, and 2%. We assumed that DI phantom causes light attenuation mainly due to absorption. Because both absorption and scattering lead to emission intensity losses and spectra modification, we measured the light attenuation spectra of our phantom solutions, as shown in figure1(b). Light of wavelength shorter than 700 nm is strongly absorbed by hemoglobin and scattered by intralipid liquids. For wavelength longer than 1350 nm, strong absorption of water reduces the luminescence intensity. Inset in figure1(b) shows the DI absorption peak at 980 nm, which overlaps with the UPNCs excitation wavelength.

Measurements of nanoprobes emission losses were measured with a custom-built optical microscope operating in transmission mode. To determine the light penetration depth, we used a liquid tissue phantom in a form of a wedge, where the height of the solution layer h increases linearly along the base of the wedge. A schematic diagram of this experimental setup is shown in figure 2(a). To properly measure changes in fluorescence probe emission intensity, few important issues need to be addressed. Firstly, the surface of the tissue phantom needs to be clean and flat. The construction of the optical wedge cannot create any artifacts. In our experiment, the liquid tissue phantom was introduced into the wedge made of inert material with wells, which was secured from the bottom with a thin microscopic slide. The diameter of the wells has to be large enough to reduce the effect of wetting of container walls. Microscopic slides need to be clean, so the liquid layers can easily spread on the surface. The nanoprobes in a cuvette were excited with an expanded and collimated laser beam. Emission from probes was guided through the tissue phantom of various thicknesses and collected with the objective lens. Then, the light was directed to the monochromator equipped with two liquid-cooled detectors recording the spectra in VIS and NIR range.

Figure 2.

Figure 2. Schematic diagram of the experimental setup equipped with the optical wedge for spectral measurements. An expanded, collimated laser beam excites the sample. Luminescence light passes through the wedge sample of various thicknesses. Transmitted light is collected with the objective lens and guided to the detectors.

Standard image High-resolution image

To investigate the penetration depth of the emitted light, the concentration of the luminescence probes has to be set. We adjust the concentration of fluorescence probes to provide the equal absorbance of all the samples upon 532 nm (980 nm for UPNCs) excitation, equal to 0.06 over the 10 mm optical path. The concentration is relatively low to provide a constant laser power density on the entire excitation path in the cuvette. For the same reason, the excitation beam was expanded and collimated. In our experiment, the cuvette with nanoprobes was arranged horizontally, keeping the solution free of air bubbles. The imaging and spectral measurement were performed for various thicknesses of the tissue phantom. We measured the probe's emission intensity for the tissue phantom thickness between 0 mm and 6 mm (hmax) in a step Δh = 0.5 mm. To collect the emission intensity from all six nanoprobes, shown in figure 1(a), two detectors for VIS and NIR spectral range have been mounted on the monochromator providing an equal length of the detection path.

2.2. Spectral measurements

To investigate the influence of the emission spectrum on the penetration depth, we performed the measurements with two linear CCD cameras connected to the monochromator. When the nanoprobes emission spectra pass through the optical wedge of different thicknesses the light intensity, beam shape, and position of the luminescence peak are changed (figure 3(a)). It is the result of the variation of absorption and scattering of tissue phantom material for photons of different energy. Therefore, we integrated the emission intensity collected for different phantom thicknesses (figure 3(b)). To compare results obtained for all six nanoprobes with nine different tissue phantoms, we used two parameters, which are Light Transmission Factor (LTF) and tissue mean penetration depth value (pdepth). The idea behind the LTF parameter is to provide information about the overall emission intensity over the entire penetration depth. LTF represents the integrated luminescence spectrum that is normalized to the initial spectrum measured without the tissue phantom. In figure 3(b), a grey rectangle shows LTF equal to 100%. The area under the red curve shows the luminescence intensity changes of PbS@1350nm passing through 0.5% hemoglobin liquid phantom, results in the LTF of 44.3%. Comparison of the LTF for different phantom types and nanoprobes allows for the proper selection of the optimal luminescence probe for given tissue thickness. Despite the comparative potential of the LTF parameter, it does not provide information about the luminescence attenuation rate. Therefore, additional parameter need to be introduced (Fig. S2).

Figure 3.

Figure 3. (a) Emission spectra from PbS@1350nm, which passes through different thicknesses of 0.5% hemoglobin dispersed in DI located in the optical wedge. (b) The emission intensity integrals and LTF in a function of tissue phantom thickness. The area under these curves is used to calculate the Light Transmission Factor (LTF) (c) The emission intensity as a function of tissue phantom thickness fitted with the single exponential decay function. (d) The comparison of LTF and pdepth values versus hemoglobin concentration in DI (open points). For pdepth data, the cut-off thickness of 6 mm was marked with a dashed line.

Standard image High-resolution image

The attenuation of the emission intensity in our studies is similar to the results reported by other authors [6, 29] where the emission intensity decrease also shows an exponential decay (figure 3(c)). The emission intensity defined as the height of the luminescence peak is inaccurate, due to the wavelength-dependent absorption and scattering of the biological media. To take this effect into account, we propose an integral over the luminescence spectra as a measure of light attenuation. Mean penetration depth (pdepth) is the parameter of the single exponential decay function fitted to the normalized decrease of spectra integrals as a function of the phantom thickness (figure 3(c)). In some cases, especially at a low concentration of the scattering media, the calculated pdepth is larger than the maximum phantom thickness (6 mm, dashed line) (figure 3(d)). It is a consequence of the single exponential model, that goes beyond experimental conditions. LTF and pdepth describe different aspects of intensity emission decay, LTF provides information about overall intensity decrease while pdepth describes the decrease rate. Presented curves provide direct insight into the light attenuation process, which allows for the comparison of the emission intensity losses for different nanoprobes for every studied thickness of tissue phantom (figure 4).

Figure 4.

Figure 4. Normalized, integrated luminescence spectrum as a function of tissue phantom penetration thickness h for DI liquid (a), 2% intralipid dispersed in DI (b), and 2% hemoglobin dispersed in DI (c).

Standard image High-resolution image

Figure 4 shows results obtained for tissue phantom based on pure DI and its solutions containing 2% intralipid and 2% hemoglobin. Probes have been divided into two groups, emitting in the visible and near-infrared regions of the optical spectrum. VIS emitting group includes R6G@565nm, CdSe@600nm, and UPC VIS@535&660nm (open symbols) and NIR emitting group includes PbS@920nm, PbS@1215nm, PbS@1350nm, and UPC NIR@1540nm (solid symbols). All the results were normalized. Figure 4(a) shows that emission from VIS probes is only slightly attenuated by DI in comparison to NIR probes. R6G@565nm and CdSe@600nm lose only 20% of initial emission intensity after passing through 6 mm of DI liquid. In comparison, emission from UPC NIR@1540nm decreased by 90% after passing through a 2 mm layer of DI tissue phantom. As expected, NIR nanoprobes are more susceptible to attenuation by water. Intralipid significantly decreases the emission intensity of all investigated nanoprobes (figure 4(b)). The luminescence intensity of VIS nanoprobes after passing through 0.5 mm of a phantom containing 2% intralipid, decreases by a factor of 5. NIR probes are less susceptible to attenuation by intralipid. Their emission intensity decreases below 10% after passing through 3 mm of tissue thickness (2% of intralipid). Especially, the PbS@1350nm probe shows a 3-fold enhancement of LTF value and 7-fold higher pdepth in comparison to R6G@535nm or CdSe@600nm. The tissue phantom composed of a water dispersion of hemoglobin decreases the emission intensity of all probes stronger than pure DI but weaker than intralipid (figure 4(c)). Solution containing 2% of hemoglobin decreases light intensity of VIS emitting probes to LTF < 40% and pdepth < 2.5 mm. Emission from R6G@565nm or CdSe@600nm decreases below 20% of the initial value after passing through the 3 mm layer of tissue phantom. NIR emitting probes are less attenuated than VIS probes. In figure 5 we present data collected for all concentrations and described using LTF and light pdepth parameters. Even a small amount of intralipid (0.25%) decreases the LTF value for VIS probes by the factor of 5 and emission pdepth by the factor of 6 (figures 5(a), (c)). Results obtained for NIR emitting probes are better than for VIS probes - LTF decreased by a factor of 2 and emission pdepth by a factor of 3. A higher concentration of intralipid reduces the difference between both groups of probes. Hemoglobin concentration changes have a weaker influence on the performance of NIR probes, especially for PbS@1215nm and PbS@1350nm. They provide a stable luminescence intensity even after passing through tissue with variable hemoglobin concentration. The best least attenuated probes are PbS@920nm and PbS@1215nm, where more than 30% of initial light intensity passed through 6 mm of tissue phantom. In the case of UPNCs, the NIR-VIS channel is more efficient for biological applications than NIR-NIR. Our observations show that NIR probes are less affected by hemoglobin absorption than VIS probes. However, due to the strong absorption of water LTF and pdepth values for UPC NIR@1540nm are very low. The LTF for UPC NIR@1540nm is below 20% and pdepth is lower than 1 mm, making them the worst candidate from all studied nanoprobes in the case of hemoglobin-based tissue phantom. Spectral measurements show that VIS nanoprobes maintained higher intensity after passing through DI-based tissue phantom than NIR probes. It is mainly due to an increase of water absorption in the NIR spectral region. Even a small amount of intralipid dispersed in DI strongly attenuates emission from all investigated probes, but this effect is weaker for NIR nanoprobes. Hemoglobin absorption attenuates luminescence intensity weaker than the effect of intralipid scattering. NIR probes perform better and are less affected by the changes in hemoglobin concentration than VIS emitters. LTF and pdepth values analysis show that the UPC VIS@535&660nm channel is the best candidate from VIS emitting probes group. UPC NIR@1540nm channel is the worst of all investigated probes, mainly due to the strong absorption of water.

Figure 5.

Figure 5. lTF values for intralipid (a) and hemoglobin (b). Mean PL intensity depths pdepth for intralipid (c) and hemoglobin (d).

Standard image High-resolution image

3. Conclusions

In this work, we presented various fluorescent nanoprobes from the perspective of their application in bioimaging. The comprehensive and systematic results have been obtained by the investigation of six nanoprobes from three different material groups, covering emission spectra from 500 to 1600 nm. We designed a tissue phantom based on the hemoglobin and intralipid dispersions in water to investigate scattering and absorption processes. For comparison of the optical properties of luminescence probes, we proposed the measurement scheme for the investigation of the luminescence light penetration depth in tissues based on analysis of luminescence spectra.

The overall comparison of results obtained for spectral measurements is shown in figure 6. The highest emission intensity after passing through DI-based tissue phantom was obtained for VIS emitting probes, such as CdSe@600nm and R6G@565nm. In the NIR region, these values decrease significantly due to water absorption. This is especially visible for UPC NIR@1540nm. Intralipid added to DI strongly attenuates emission from all studied probes but this effect is weaker for NIR emitting probes. Hemoglobin absorption attenuates the emission to a lesser extent compared to the scattering of intralipid. NIR emitting probes perform better and are less affected by the changes in hemoglobin concentration than VIS emitters. However, the luminescence above 1350 nm is strongly attenuated due to water absorption. Therefore, from comprehensive considerations, the PbS@920nm and PbS@1215nm probes should be considered as the best candidates for the experiments where the total luminescence intensity passing through tissue has the greatest importance.

Figure 6.

Figure 6. Comparison of LTF and light pdepth values for all nanoprobes measured with tissue phantom based on DI, 2% intralipid dispersed in DI and 2% hemoglobin dispersed in DI.

Standard image High-resolution image

4. Methods

4.1. Fluorescence probes

The solution of R6G@565nm is a commercially available FluoSpheres™ (Carboxylate-modified Microspheres, 2% solids, size 0.110 μm) provided by ThermoFisher. Other nanoprobes were synthesized. UPNCs (NaGdF4:Yb3+/Er3+) were synthesized according to already published protocol [30]. CdSe quantum dots were obtained in a one-step synthesis according to the procedure proposed for CdS QDs [31], however, the Se powder (100 mesh) was used as a sulfur precursor. Oleate capped PbS quantum dots were synthesized according to modified protocol [32].

Intralipid solutions have been prepared using Lipofundin MCT/LCT 20% from B. Braun Melsungen AG, which was diluted within DI to solutions containing 0.25%, 0.5%, 1%, 2%. Hemoglobin solution has been prepared from the dried pork blood containing 99% of hemoglobin, which was diluted within DI to solutions containing 0.25%, 0.5%, 1%, 2%.

4.2. Experimental setup

DI, intralipid, and hemoglobin attenuation spectra were measured using JASCO V570 UV–vis-NIR spectrometer. To measure PL spectra a custom-build transmission microscope has been used containing Horiba iHR550 monochromator connected with two spectral cameras for VIS spectrum Symphony II 1024 × 256 Back Illuminated Deep Depletion and for NIR spectrum Symphony Linear IGA 1700. All nanoprobes, except UPNCs, were excited using a continuous wave (CW) 532 nm laser line with 5 mW laser power (power density ∼2.6 W cm−2) from JDS Uniphase μGREEN-SLM LASER. In the case of UPNCs, due to the upconversion mechanism and small cross-section for the absorption of the Yb3+ ions, we used a 980 nm CW laser from Changchun New Industries Optoelectronics Technology Co., Ltd MDL-III-980 with 100 mW optical power (optical density ∼51 W cm−2). The excitation conditions for UPNCs were different than for other nanoprobes but focusing on the relative changes of the emission intensity makes them comparable with other probes.

Acknowledgments

This study was supported by a grant from the Polish National Science Center, FUGA program No. UMO 2016/20/S/ST3/00277 and Sonata Bis 3 Project No. UMO-2013/10/E/ST5/00651.

Data availability statement

The data generated and/or analysed during the current study are not publicly available for legal/ethical reasons but are available from the corresponding author on reasonable request.

Conflict Of Interests

The authors have declared that no conflicting interests exist

Please wait… references are loading.
10.1088/2050-6120/ac10ae