Low-cost compact optical spectroscopy and novel spectroscopic algorithm for point-of-care real-time monitoring of nanoparticle delivery in biological tissue models

Abstract

Objective:

Real-time monitoring of nanoparticle delivery in biological models is essential to optimize nanoparticle-mediated therapies. However, few techniques are available for convenient real-time monitoring of nanoparticle concentrations in tissue samples. This work reported novel optical spectroscopic approaches for low-cost point-of-care real-time quantification of nanoparticle concentrations in biological tissue samples.

Methods:

Fiber probe measured diffuse reflectance can be described with a simple analytical model by introducing an explicit dependence on the reduced scattering coefficient. Relying on this, the changes on the inverse of diffuse reflectance are proportional to absorption change when the scattering perturbation is negligible. We developed this model with proper wavelength pairs and implemented it with both a standard optical spectroscopy platform and a low-cost compact spectroscopy device for near real-time quantification of nanoparticle concentrations in biological tissue models.

Results:

Both tissue-mimicking phantom and ex vivo tissue sample studies showed that our optical spectroscopic techniques could quantify nanoparticle concentrations in near real-time with high accuracies (less than 5% error) using only a pair of narrow wavelengths (530 nm and 630 nm).

Conclusion:

Novel low-cost point-of-care optical spectroscopic techniques were demonstrated for rapid accurate quantification of nanoparticle concentrations in tissue-mimicking medium and ex vivo tissue samples using optical signals measured at a pair of narrow wavelengths.

Significance:

Our methods will potentially facilitate real-time monitoring of nanoparticle delivery in biological models using low-cost point-of-care optical spectroscopy platforms, which will significantly advance nanomedicine in cancer research.

I. Introduction

Nanoparticles are tiny materials having sizes ranging from 1 to 100 nm [1]. Because of their superior biocompatibility, high optical absorption properties, and capability of molecular-specific targeting of cancer cells [2], nanoparticles have been extensively exploited for many biomedical applications such as targeted drug/gene delivery [3] and cancer thermal therapy [4]. Real-time monitoring of nanoparticle delivery on in vivo biological models is essential to optimize nanoparticle-mediated therapies and better understand molecular interactions of nanoparticles with biological tissues [5–7]. For example, real-time longitudinal monitoring of nanoparticle delivery kinetics in tumors is essential to optimize the nanoparticle dose and laser induction protocol to maximize the enhanced permeability and retention effect, thereby providing maximized effectiveness in cancer thermal therapy [6, 7].

Several techniques have been explored to measure nanoparticle concentrations on in vivo biological models but with practical limitations. For example, magnetic resonance imaging (MRI) has been used to report nanoparticle concentrations where the nanoparticles were used as special imaging contrast [8]. However, MRI can only quantify certain types of nanoparticles with high costs. X-ray fluorescence-based imaging/tomography has been explored for nanoparticle concentration measurements [9]. However, X-ray techniques have relatively low sensitivity for nanoparticle concentration imaging with additional radiation safety concerns. Optical techniques have been explored as new ways for the non-invasive quantification of nanoparticle concentrations in biological tissues. For example, two-photon luminescence microscopy has been reported to measure gold nanoparticle concentration in animal tissues [10]. Fluorescence microscopy has been used to report nanoparticle concentrations in tissue samples where nanoparticles are used as fluorescence imaging probes [11]. Either luminescence or fluorescence-based techniques require the nanoparticles to be luminescent or fluorescent, the measured signals from both techniques will be distorted by tissue background signals. Photoacoustic imaging techniques have been also explored to monitor nanoparticle delivery in animals in vivo [5]. However, the photoacoustic imaging approaches typically require expensive equipment and complicated data processing. Though these techniques have been explored for in vivo monitoring of nanoparticle concentrations in tissue, most of them are: (1) housed in core facilities that require transporting subjects to their site; (2) costly in equipment, and (3) not in real-time due to complicated data processing to achieve reasonable accuracy. These factors all limit their access for high-frequency measurements or longitudinal monitoring of nanoparticle concentrations on in vivo biological models.

Diffuse optical spectroscopy techniques [6, 7] have been explored as a cost-effective approach to measuring nanoparticle concentrations on animal models by examining the optical absorption of nanoparticles. Diffusion approximation has been utilized to analyze the optical spectra [6, 7] measured on small animals thereby providing offline estimations of nanoparticle concentrations with moderate accuracy. To enable convenient monitoring of nanoparticle delivery in biological models for translational nanomedicine applications, it is highly significant to develop real-time techniques with point-of-care and low-cost footprints to quantity nanoparticle concentrations in biological tissues. Here we report novel optical spectroscopic approaches for low-cost point-of-care real-time quantification of nanoparticle concentrations in tissue samples with high accuracy. Specifically, we developed: (1) a novel optical spectroscopic model for rapid quantification of nanoparticle concentrations in tissues by quantifying the absorption changes induced by nanoparticles using only a pair of special wavelengths; (2) a low-cost compact diffuse reflectance spectroscopy device for point-of-care real-time optical measurements on biological models. To demonstrate the proof-of-concept of our point-of-care real-time optical approaches, we performed both tissue-mimicking phantom studies and ex vivo tissue sample studies for rapid quantification of nanoparticle concentrations. Our data showed that our techniques could quantify nanoparticle concentrations in real-time with high accuracies (less than 5% percent error on average) using only a pair of narrow wavelengths (530 nm and 630 nm). Due to the inherent simplicity of the proposed technique, our methods will potentially facilitate real-time monitoring of nanoparticle delivery in biological models using low-cost point-of-care spectroscopy platforms, which will significantly advance nanomedicine in cancer research.

II. Materials and Methods

A. Diffuse reflectance spectroscopic model

Optical fiber probe measured diffuse reflectance can be described with a simple analytical model by introducing an explicit dependence on the reduced scattering coefficient [12] as shown in equation (1), where R represents the measured diffuse reflectance, μ_a and ${\mu }_s^{\prime }$ represent absorption and reduced scattering coefficients respectively. K₁ and K₂ are fiber probe dependent constant parameters.

$$R=\frac{{\text{μ}}_s^{\prime }}{k_1+k_2{\text{μ}}_a}$$ (1)

Relying on equation (1), the inverse of diffuse reflectance measured on tissues before the addition of nanoparticles (define as R_baseline) can be expressed by equation (2).

$$\frac{1}{R_{baseline}}=\frac{k_2{\text{μ}}_{a_{\_}baseline}}{{\text{μ}}_{s_{\_}baseline}^{\prime }}+\frac{k_1}{{\text{μ}}_{s_{\_}baseline}^{\prime }}$$ (2)

When nanoparticles at biologically low doses are introduced into biological tissues, the changes in the reduced scattering coefficients caused by nanoparticles could be negligible given the reduced scattering coefficients of tissues are generally high [13]. Then the inverse of diffuse reflectance measured on biological tissues post the addition of nanoparticles (define as R_nano) can be approximated by equation (3).

$$\frac{1}{R_{nano}}=\frac{k_2({\text{μ}}_{a_{\_nano}}+{\text{μ}}_{a_{\_baseline})}}{{\text{μ}}_{s_{\_}baseline}^{\prime }}+\frac{k_1}{{\text{μ}}_{s_{\_}baseline}^{\prime }}$$ (3)

By subtracting equations (3) and (2), one can get equation (4) to describe the relationship between the absorption coefficient of nanoparticles and the corresponding optical signal changes due to the addition of nanoparticles.

$$\frac{1}{R_{nano}}-\frac{1}{R_{baseline}}=k_2\frac{{\text{μ}}_{a_{\_nano}}}{{\text{μ}}_{s_{\_}baseline}^{\prime }}$$ (4)

Based on the beer-lambert law, the absorption coefficients of nanoparticles are linearly proportional to their concentrations given their molar extinction coefficients are fixed [14]. Therefore, equation (4) can be further modified as equation (5).

$$\frac{1}{R_{nano}}-\frac{1}{R_{baseline}}=k\frac{C_{nano}}{{\text{μ}}_{s_{\_}baseline}^{\prime }}$$ (5)

The reduced scattering levels of tissue samples can be generally indicated by diffuse reflectance intensities at scattering dominant wavelengths bands (where μ_a << ${\mu }_s^{\prime }$) based on equation (1). Then equation (5) can be further developed into equation (6), which can be potentially used to monitor the nanoparticle concentrations in real-time using optically measurable signals at proper wavelength bands, where λs represents optical absorption dominant band while λs represents the scattering dominant band for biological tissue samples, and k is a system dependent constant factor.

$$k\ast C_{nano}=\left(\frac{1}{R_{nano}(\lambda a)}-\frac{1}{R_{baseline}(\lambda a)}\right)\ast R_{baseline}(\lambda s)$$ (6)

B. Optical spectroscopy platforms

To evaluate the novel diffuse reflectance spectroscopic model for nanoparticle concentration measurements, both a standard optical spectroscopy platform and a low-cost compact spectroscopy device were utilized. The standard optical spectroscopy system was built based on a high-power Solis™ white LED (SOLIS-3C, Thorlabs) and a compact spectrometer (FLAME-T-VIS-NIR, Ocean Optics) using the layout in Figure 1. A commercially available fiber probe (BF19Y2HS02, Thorlabs) with 10 illumination fibers and 9 collection fibers was used for light delivery and collection. The use of a Solis™ white LED provided a broad illumination band (Figure 1 bottom left), which allows us to explore all potential wavelengths pairs for nanoparticle concentration quantification using equation (6). Isosbestic wavelengths for hemoglobin (500 nm, 530 nm, 545 nm, 570 nm, and 580 nm) were explored for the absorption band wavelengths, while five wavelengths from 600 nm to 700 nm (every 25 nm) were explored for scattering band wavelengths. The total cost for the major components used in our standard spectroscopy system was around $7500.

Figure 1.

Design of a standard diffuse reflectance spectroscopy platform and its actual layout in a suitcase .

The low-cost optical spectroscopy system has a similar design (Figure 2) to the standard spectroscopy platform but uses ultra-low-cost LEDs (Luxeon LEDs) along with heating sinks. Two colors of LEDs (Luxeon, 530 nm (07040-PM000-L) and 630 nm (J035-L1C1RED)) were used to build a low-cost illuminator (less than $100). A two-to-one fiber was used to deliver the two colors of light to the main fiber probe for sample illumination and collection. Due to the use of low-cost compact optical components (Figure 2, right), the total cost for the major components used in our low-cost spectroscopy device is reduced to ~$3500 and its total weight is minimized to be ~3 lbs. Note that this system cost can be further minimized to ~$1500 by replacing the spectrometer with a low-cost CCD three-color camera (CS165CU, Thorlabs) in our future plan.

Figure 2.

Low-cost compact diffuse reflectance spectroscopy platform

C. Nanoparticle and Tissue-Mimicking Phantoms

To demonstrate the principle of concept of our spectroscopic model [equation (6)] for rapid and accurate monitoring of nanoparticle concentrations in tissue samples, the iron oxide nanoparticles reported previously [15] and fabricated by our team were used in tissue-mimicking phantom and ex vivo tissue studies. Figure 3 (A) showed the image of the iron oxide nanoparticles acquired by a transmission electron microscope (FEI Talos F200X), while Figure 3 (B) showed the optical absorbance of the nanoparticles at biologically relevant concentrations (0 to 20 ug/ml).

Figure. 3.

(A). Transmission electron microscopy of nanoparticles. (B). The absorbance of the nanoparticles at biologically relevant concentrations.

The iron oxide nanoparticles were added to liquid tissue-mimicking phantoms with biological relevant concentrations to mimic the nanoparticle distributed in biological tissues. Tissue mimicking phantoms were prepared based on the formerly reported optical properties of mice subcutaneous tumor models [16]. Specifically, these phantoms had the following average absorption coefficients and reduced scattering coefficients (400-600 nm): μ_a= [1.5, 3.0, 4.5] cm⁻¹ and ${\mu }_s^{\prime }$= [4.5, 9.0, 13.5] cm⁻¹. Dehydrated human hemoglobin powder (H0267, Sigma-Aldrich) was used as a major absorber in the phantoms. 20% intralipid (Sigma-Aldrich) was used to mimic tissue scattering. Phosphate Buffered Saline (PBS) 1x (Fisher Scientific) was used to suspend the intralipid and hemoglobin for the liquid phantoms. All baseline phantoms (no nanoparticle) had the same starting volume of 5 mL, then a stock of 25 μL nanoparticle (1 mg/ml) was added sequentially to create the phantoms with the designated nanoparticle concentrations (from 0 to 20 ug/ml by every 5 ug/ml).

D. Ex vivo tissue sample studies

To further evaluate both our optical spectroscopic model and our low-cost compact spectroscopy device for rapid and accurate measurement of nanoparticle concentrations in real tissue samples, fresh ex vivo chicken tissue samples purchased from a local grocery store were also utilized for our experiments. As illustrated in Figure 4, fresh chicken tissue samples were fixed on the measurement stage using syringe needles. The iron oxide nanoparticles with four different biological relevant injection concentrations (0.25 to 1 mg/ml by every 0.25 mg/ml) were injected to different sites of tissue samples using a syringe with a 30 G needle. At each concentration, 50 uL nanoparticle solution was used and the injection was done gently and slowly to ensure there was no leakage around the injection site. An optical fiber probe was placed on the tissue surface with gentle contact with the help of an adjustable probe holder for actual optical measurements (Figure 4 right).

Figure 4.

Nanoparticle injection and optical measurements on fresh chicken tissue sample.

E. Optical Measurements and Data Analysis

Diffuse reflectance on all sets of phantoms prior and post to the addition of nanoparticles over a wide wavelength range (500-700 nm) was measured with an integration time of 75 ms using our standard optical spectroscopy platform (Figure 1) only. Diffuse reflectance on fresh ex vivo tissue samples prior and 10 minutes post to the injection of nanoparticles was measured using both our standard optical spectroscopy and our low-cost compact optical spectroscopy device. After all optical measurements on the phantoms or ex vivo tissues were completed, a reference spectrum on a diffuse reflectance standard puck (20%, Spectralon, Labsphere) was collected for data calibration. All diffuse reflectance spectra were calibrated using a 20% reflectance standard (Spectralon. Labsphere) for the wavelength-dependent response by normalizing them to the diffuse reflectance spectrum measured on the reflectance standard puck. The calibrated optical data before and post addition of nanoparticles were then processed using our model to quantify the nanoparticles concentrations. The novel optical spectroscopic model was implemented in Matlab (Mathworks, Natick, Massachusetts) for data processing. The mean percent errors were calculated for the comparison between the estimated nanoparticle concentrations with their corresponding true values by using Pearson’s test. The Bland-Altman plots were also used to show the statistical confidence intervals.

III. Results

A. Diffuse reflectance spectra of tissue-mimicking phantoms

Both absorption spectra of hemoglobin and scattering spectra of intralipid at biologically relevant concentrations used in this study were provided by us previously [17]. Figure 5 (A) showed the corresponding diffuse reflectance spectra for all nine sets of baseline tissue-mimicking phantoms with optical properties covering both normal and tumorous tissues in mice subcutaneous tumor models. The spectral data in Figure 5 (A) further confirmed that the reduced scattering levels of tissue samples can be generally indicated by diffuse reflectance intensities at scattering dominant wavelength bands (650-750 nm). Figure 5 (B) showed the representative diffuse reflectance spectra measured on tissue-mimicking phantoms before and post the additions of nanoparticles for phantoms with middle scattering levels. The spectra showed that there was a decrease in the diffuse reflectance once the nanoparticles were added, which confirmed our assumption that the scattering changes caused by low doses of nanoparticles might be negligible.

Figure. 5.

(A) Diffuse reflectance spectra of nine groups of tissue-mimicking phantoms before the addition of nanoparticles (baseline); (B) Representative reflectance spectra before and post the additions of nanoparticles for phantoms with μ_a=1.5 cm⁻¹ and ${\mu }_s^{\prime }$= 9.0 cm⁻¹.

B. Rapid estimation of nanoparticle concentration from the turbid medium using reflectance intensities at a pair of narrow wavelengths

Figure 6 showed the performance of our proposed diffuse reflectance spectroscopic model for quantification of nanoparticle concentrations from turbid tissue-mimicking phantoms using a pair of wavelengths (λa=530 nm, and λs=650 nm). The 530 nm was chosen as the absorption band wavelength for data processing because it is one of the isosbestic wavelength points for hemoglobin, which will minimize the effect caused by tissue oxygen saturation. The 650 nm was chosen as the scattering band for baseline scattering correction as the diffuse reflectance at a longer wavelength can represent the scattering levels well. Figure 6 (A) showed representative diffuse reflectance spectra measured on tissue-mimicking phantoms before and post the additions of nanoparticles for phantoms with low, middle, and high scattering levels. All spectra showed that there was a decrease in the diffuse reflectance once the nanoparticles were added, which confirmed our assumption that the scattering changes caused by low doses of nanoparticles might be negligible. Figure 6 (B) showed the nanoparticle concentrations estimation using equation (5) which contained the distortions caused by tissue baseline scattering. Strong correlation and linear ship between nanoparticle concentrations and the corresponding changes on the inverse of diffuse reflectance were achieved using equation (5), while the tissue baseline scattering significantly affected the nanoparticle concentrations estimations as evidenced by the different slopes of these linear fit curves. Figure 6 (C) showed the nanoparticle concentrations estimation using equation (6) which effectively corrected the tissue baseline scattering distortions. With the baseline scattering correction model, the optical indicators can accurately report the nanoparticle concentration regardless of the baseline optical absorption and scatter levels. As shown in the top panel of Figure 6 (D), a linear fit to the optical indicators extracted by our model versus true nanoparticle concentrations (for all 36 phantoms) yields a coefficient of determination R² of 0.98 and p-value less than 0.02, indicating the high performance of the novel diffuse reflectance spectroscopic model for nanoparticle concentration quantification. Figure 6 (D) bottom panel showed the Bland Altman plot for all the optical indicators shown in the top panel of Figure 6 (D), further indicating the high performance of our model for nanoparticle concentration quantification (less than 5% error).

Figure. 6.

(A). Representative reflectance spectra before and post the additions of nanoparticles. (B). Nanoparticle concentrations estimation without baseline scattering correction using equation (5). (C). Nanoparticle concentrations estimation with baseline scattering correction using equation (6). (D). Top: Correlation between the optical indicators obtained using equation (6) and true nanoparticle concentrations for all phantoms. Bottom: Bland Altman Plot for the data shown in the top panel. LoA represents limits of agreement (defined as the mean difference ± 1.96 SD of differences). Mean diff represents the mean difference. λa and λs represent absorption and scattering bands. μ_a and ${\mu }_s^{\prime }$ represent absorption and reduced scattering coefficients respectively.

C. Proper wavelength bands and bandwidths for the use of the optical spectroscopic model

To further popularize the general use of our novel diffuse reflectance spectroscopic model for nanoparticle concentration measurements, we also explored the potentially available wavelength bands and wavelength bandwidths based on the commonly used optical bandpass filters. Specifically, all isosbestic wavelength points for hemoglobin (500 nm, 530 nm, 545 nm, 570 nm, and 580 nm) were explored for the absorption band wavelengths, while five wavelengths from 600 nm to 700 nm (every 25 nm) were explored for scattering band wavelengths. Figure 7 (A) showed the performances of our proposed diffuse reflectance spectroscopic model for quantification of nanoparticle concentrations using different combinations of these wavelengths when the bandwidth was set to 1 nm. Figure 7 (A) showed that the best absorption band wavelengths were 530 nm, 545 nm, and 570 nm, regardless of the choices of scattering band wavelengths, as evidenced by the high R² values (always higher than 0.96). However, the use of two isosbestic wavelengths, 500 nm and 580 nm, should be avoided due to their relatively worse performances. Figure 7 (B) showed the effect of wavelength bandwidths on the performance of our proposed model. Though the R² values decreased as expected when the wavelength bandwidths were increased, this change was almost negligible as evidenced by the data shown in Figure 7 (B). A larger bandwidth of the optical filters typically would provide enhanced optical signals. Thus, a tradeoff between the signal strengths and the model accuracy needs to be considered for future use.

Figure. 7.

(A) Coefficient of determination (R²) value for different wavelength pairs. (B) Relation between R² values and wavelength bandwidths.

E. Rapid estimation of nanoparticle concentration from ex vivo tissue sample using low-cost compact optical spectroscopy

To further evaluate both our optical spectroscopic model and our diffuse reflectance spectroscopy platforms (Figures 1 and 2) for rapid and accurate measurement of nanoparticle concentrations in real tissue samples using only a pair of wavelengths, fresh ex vivo chicken tissue sample studies were also conducted. Both a standard spectroscopy system and our compact low-cost spectroscopy platform were used for optical measurements on ex vivo tissues before and 10 minutes post nanoparticle injections. The light at 530 nm and 630 nm were chosen as the absorption and scattering band wavelength for nanoparticle concentration measurements based on the commonly available low-cost LEDs colors (Luxeon LEDs). Figure 8 showed the performances of both our standard optical spectroscopy platform and low-cost compact spectroscopy device for the quantification of nanoparticle concentrations from fresh ex vivo tissue samples.

Figure. 8.

Optical measurements of nanoparticle concentrations in ex vivo chicken tissue using the standard optical spectroscopy (A) and our two-color LEDs based low-cost optical spectroscopy (B).

The top panel in Figure (8) showed the typical raw optical spectra or signals measured using a standard spectroscopy and the two-color LEDs-based low-cost spectroscopy platform. The optical signals showed that there was a decrease in the diffuse reflectance once the nanoparticles were added as expected. The middle panel in Figure 8 showed the estimation of nanoparticle concentrations from ex vivo tissues using equation (6). The linear fits to the optical indicators extracted by optical platforms versus injection nanoparticle concentrations indicate the high performance of both the standard spectroscopy and low-cost spectroscopy systems as evidenced by high R² values (over 0.98) and low RMSE (less than 0.01). The bottom panel of Figure 8 showed the Bland Altman plot for all the optical indicators shown in the middle panel of Figure 8, further indicating the high performance of our optical techniques for nanoparticle concentration quantification from ex vivo tissue samples (less than 5% error). The performance of our low-cost optical spectroscopy is almost identical to our standard optical spectroscopy with slightly larger limits of agreement (LoA) in the Bland Altman plot.

IV. Discussion

In nanoparticle-mediated thermal therapies, the intravenous administration of particles leads to nanoparticle accumulation within the tumors, and the illumination of these areas with proper light leads to hyperthermia. To achieve the best cancer thermal therapy effectiveness, it is essential to optimize the nanoparticle dose and laser induction protocol which requires real-time longitudinal monitoring of nanoparticle delivery kinetics in tumors [6, 7]. Though several techniques including MRI, X-ray, and fluorescence imaging techniques have been explored to quantify the nanoparticle concentrations in small animals, none of them has been adapted for longitudinal monitoring of nanoparticle delivery due to their practical limitations. Recently, optical spectroscopy techniques [6, 7] have been explored as a non-invasive approach to measuring nanoparticle concentrations in animals by examining the optical absorption of nanoparticles. Mireles et al reported a dual-model optical platform that combined diffuse reflectance spectroscopy and diffuse correlation spectroscopy to monitor both nanoparticle concentrations and hemodynamics in small animals to capture the gold nanoparticle delivery kinetics in tissues [6]. Zaman et al reported a custom-designed diffuse reflectance spectroscopy platform to quantify gold nanoshells concentrations on small animals and an accuracy of 12.6% percent error was achieved [7]. In both the two studies [6, 7], the diffusion approximation model was used for nanoparticle concentration estimation in which offline data processing was needed and the accuracy was limited by the approximation. In this study, we have demonstrated: (1) a novel diffuse optical spectroscopic model for rapid estimation of nanoparticles on tissue-mimicking turbid medium using reflectance signals at only a pair of narrow wavelengths; (2) a low-cost portable diffuse reflectance spectroscopy platform for real-time optical measurements, both will potentially facilitate point-of-care real-time in vivo monitoring of nanoparticle delivery in biological models, which will advance translational cancer research using nanoparticles.

Both our tissue-mimicking phantom studies and actual tissue sample studies showed that our technique could quantify nanoparticle concentrations with high accuracies (less than 5% error) using only two narrow wavelengths (λa and λs). The requirement of light at only two narrow wavelengths could further significantly reduce the equipment cost and system size by using low-cost LEDs for illumination instead of a high-power white LED light source as demonstrated by us in Figure 2. The system cost can be further minimized by replacing the spectrometer with a low-cost CCD three-color camera (CS165CU, Thorlabs) in our future plan. The green channel of the CCD camera can be used to collect signals for green light (i.e. 530 nm) while the red channel of the CCD camera can be used to detect signals from red light (i.e. 630 nm). Due to the inherent simplicity of the proposed approach, as shown in equation (6), our novel techniques have the potential to enable online real-time nanoparticle concentration monitoring on small animals in vivo using optical diffuse reflectance spectroscopy platforms.

In this study, a low-cost commercially available fiber probe with an average source-detector separation of 1 mm was used. This fiber with a relatively small source-detector separation was used due to our special interest in oral cancer investigations [18]. It should be noted that a fiber probe with a larger source-detector distance should be used for other tumor models such as flank tumor models or mammary pad breast cancer models to ensure sufficient optical sensing depth [19]. A former study [12] reported the fiber probe-based diffuse reflectance model [equation (1)] will be valid when the source-detector distance was increased from 0.5 mm to 10 mm. We envision that our spectroscopic model for nanoparticle concentration estimation will be still valid when a different fiber probe with a large source-detector distance (<10 mm) is used. Our recently published data [19] showed that a fiber probe with source-detector distances in the range of ~2 to 3 mm will provide sufficient sensing depth to probe murine subcutaneous tumors given the tumor depth is around ~ 1 mm beneath the tissue surface [16]. To monitor nanoparticle delivery in subcutaneous tumors, a new fiber probe with source-detector distances of ~3 mm should be considered. It is commonly known that fiber probe based optical spectroscopy lacks the capability to capture heterogeneity information as it usually gets an overview of tissue status through probing a tissue volume. Therefore, performing optical spectroscopy measurements at multiple sites of a sample may be considered to help capture the heterogeneity of nanoparticles distribution in tissue.

To implement the general use of the diffuse optical spectroscopic model with existing commercial optical components, we evaluated the effect of the central wavelength bands and their bandwidths on the accuracy of nanoparticle concentration quantification. Generally, we found that the bandwidth had minimal effect on the performance of our technique for nanoparticle concentration measurement when proper wavelength bands were selected for data processing as shown in Fig. 7 (B). Based on this, a larger bandwidth of the optical filters (if a white illumination source is used) or relatively broad LEDs (two-color LEDs are used ) are recommended in the design of an optical system as they typically yield stronger optical signals, which may potentially provide faster data acquisition as minimal integration time may be used. However, we did observe that the selection of central wavelength bands may significantly affect the performance of the spectroscopic method for nanoparticle concentration measurement based on the data shown in Fig. 7 (A). Proper selection of these wavelength bands will be necessary to enable high accuracy of the method for data processing. Two low-cost monochrome LEDs (530 nm and 630 nm) with a full bandwidth of 50 nm have been implemented with our spectroscopy device for accurate measurements of nanoparticle concentrations in tissue samples. Our data in Figure 7 showed that the LEDs bandwidth had minimal effect on the performance of our model, while the future use of the monochrome LEDs and a color camera may have decreased performance as camera has low spectral resolution. Bandpass filters may be used along with the LEDs to narrow the illumination light bandwidth, thereby improving the performance of color camera based device for nanoparticle concentration measurement.

In this study, we conducted both tissue-mimicking phantom and ex vivo tissue studies with iron oxide nanoparticles to demonstrate the proof-of-concept of our approaches due to the popular use of iron oxide nanoparticles in cancer research. We envision our methods will be generally applicable to other types of nanoparticles concentrations measurements as long as the assumption that the scattering caused by nanoparticles is negligible to background tissue scattering levels can be met. The demonstration of our custom-designed low-cost portable spectroscopy along with our novel spectroscopic model opens new possibilities for studying the dynamics of nanoparticle accumulation in tumors in a real-time and point-of-care manner. Optical spectroscopy is well suited for tissue functional characterization in vivo and has been extensively explored to measure tissue oxygenation and metabolism as demonstrated by our former studies [20]. Because optical spectroscopy can measure tissue physiology (i.e., blood hemoglobin and oxygen saturation) [20], it can be used to simultaneously monitor tumor physiology (oxygen saturation and hemoglobin concentration) in addition to nanoparticle delivery, and treatment responses during photothermal therapy.

Normal and tumorous tissues have reduced scattering coefficient levels in the range of 5-15 cm⁻¹ [16]. The reduced scattering coefficient of nanoparticles is determined by both their extinction scattering coefficient and concentration. The extinction scattering coefficient of commonly used nanoparticles is small [21] and their in vivo use dose is very low as described previously, therefore the reduced scattering level of nanoparticles at biological relevant concentrations in tissues is extremely small compared to tissue background scattering, which meet the assumption we made in our model, i.e., the addition of nanoparticle causes negligible changes in the sample scattering properties. However, some nanoparticles may have efficient scattering. To explore if our approach may still work for the measurement of nanoparticles with non-negligible scattering levels, we assume ${\text{μ}}_{s_{\_}nano}^{\prime }=a\ast {\text{μ}}_{s_{\_}baseline}^{\prime }(a>0)$, then one can get equation (7) to describe the relationship between the nanoparticle absorption coefficient and the corresponding optical signal changes due to the addition of nanoparticles with non-negligible scattering levels.

$$\frac{1+a}{R_{nano}}-\frac{1}{R_{baseline}}=k_2\frac{{\text{μ}}_{a_{\_nano}}}{{\text{μ}}_{s_{\_baseline}}^{\prime }}$$ (7)

The equation (7) suggested that our model may also potentially work for the measurement of nanoparticles with efficient scattering level if the a value can be estimated. The a value may be estimated by examining the diffuse reflectance intensities on tissue samples before and post the use of nanoparticles at scattering dominant wavelengths, which will be explored and verified in our future work.

We conducted both tissue-mimicking phantom and ex vivo studies to demonstrate the proof-of-concept of our techniques. It will be straightforward to implement our spectroscopy platform and diffuse optical spectroscopic model for in vivo animal studies as demonstrated by others before [6, 7]. We will further develop and apply the reported empirical method for in vivo preclinical data processing in our future study plans. Our phantom study demonstrated that the empirical method can rapidly extract nanoparticle concentrations from diffuse reflectance intensities at only a pair of narrow wavelengths with high accuracy (less than 5% error). Due to the simple nature of our model for data processing, our technique will potentially facilitate real-time monitoring of nanoparticle delivery of tumor models in vivo using optical spectroscopy techniques, which will advance nanomedicine for translational cancer research. Our methodology reported here will also be generally applicable to other nanoparticles and other spectroscopy systems, and potentially imaging platforms.

V. Conclusion

This work reported: (1) a novel optical spectroscopic model for rapid quantification of nanoparticle concentrations in biological tissues by quantifying the absorption changes induced by the use of nanoparticles; (2) a cost-effective point-of-care diffuse reflectance spectroscopy platform for real-time optical measurements on biological models. The proposed diffuse optical spectroscopic model and the point-of-care spectroscopy platform will potentially enable real-time in vivo monitoring of nanoparticle delivery in biological models using low-cost point-of-care spectroscopy/imaging platforms, which will significantly advance nanomedicine in translational cancer research.

Biography

Mr. Md Zahid Hasan, received the B.S. degree in biomdical engineering from Khulna University Of Engineering & Technology, Khulna, Bangladesh, in 2020. He is currently pursuing the Ph.D. degree in biomedical engineering at University of Kentucky, Lexington, KY. His research interest includes the development of point-of-care optical spectroscopy technique for biomedical application including cancer research.

Ms. Jing Yan, received the B.S. degree in Biotechnology from Southern Medical University, Guang Don, China, in 2016. She is currently pursuing the Ph.D. degree in biomedical engineering at University of Kentucky, Lexington, KY. Her research interest includes the development of non-destructive optical imaging techniques for cancer applications.

Mr. Zhongchao Yi, received the B.S. degree in Functional Materials from Chongqing University of Science and Technology, Chongqing, China in 2018. He recied the M.S. Degree in biomedical eingineering from Rice University, Houston, Texas, in 2019. He is currently pursuing the Ph.D. degree in biomedical engineering at University of Kentucky, Lexington, KY. His research interest includes the development of nanoparticle for cancer research.

Madison Korfhage, is an undergraduate student and currently pursuing the B.S. degree in biomedical engineering at University of Kentucky, Lexington, KY. Her research interest includes the development of low-cost optical techniques for biomedical application including cancer research.

Dr. Sheng Tong, received the B.S. degree in Mechanical Engineering from University of Science and Technology of China, Hefei, China, in 1995. He received his M.S. in Mechanical Engineering from Peking University, Beijing, China, in 1998. Dr. Tong received his Ph.D. degree in biomedical engineering from Duke University, Durham, NC, in 2003. He is currently an associate professor from the department of biomedical engineering at University of Kentucky, Lexington, KY. His research interest includes nanobiotechnology, genome editing, immunoengineering, and cancer therapy

Dr. Caigang Zhu, received the B.S. degree in Biomedical Engineering from Huazhong University of Science and Technology, Wuhan, China, in 2008. He received his Ph.D. degree in biomedical engineering from Nanyang Technological University, Singapore, in 2014 He is currently an assisitant professor from the department of biomedical engineering at University of Kentucky, Lexington, KY. His research interest is focused on novel point-of-care optical spectroscopy and imaging technologies for cancer research.