Research ArticleOpen Access

Diffuse transmitted spectroscopy in conjunction with spectral peak averaging as a potential tool for noninvasive creatinine screening

Zvi Hai Barnea (Burbea)

Nephrology Department, Edith Wolfson Hospital, Holon 5822012, Israel

Sackler Faculty of Medicine, Tel Aviv University, P. O. Box 39040, Tel Aviv 6997801, Israel

E-mail Address: barnea@wmc.gov.il

Search for more papers by this author

Shmuel Zimlichman

Department of Electrical and Electronics Engineering, Ariel University, Ariel 4070000, Israel

Search for more papers by this author

, and

David Abookasis

Department of Electrical and Electronics Engineering, Ariel University, Ariel 4070000, Israel

E-mail Address: davida@ariel.ac.il

Corresponding author.

Search for more papers by this author

https://doi.org/10.1142/S1793545820500285Cited by:0 (Source: Crossref)

Abstract

Creatinine (Cr) is a biochemical waste molecule generated from muscle metabolism and primarily cleared from the bloodstream by the kidneys. If kidney function declines, Cr levels in the blood tend to increase. Therefore, Cr serves as an indicator of kidney function. In this work, we present a simple method for the rapid screening for impaired renal function based on the subject’s Cr concentration. In our setup, broadband white light is delivered to a finger clamp through a fiber-optic cable to illuminate the patient’s finger. The light is transmitted through the finger and collected by a second optical fiber coupled to a visible–near-infrared (VisNIR) spectrometer which covers the spectral range from 400nm to 1100nm. During the calibration process, the transmitted spectra acquired from 60 patients were measured. An average was calculated using the peak level of the transmitted, diffused intensity at three different wavelengths to create a “Cr intensity index”. Patients were divided into five groups according to their Cr concentration levels, ranging from 1mg/dL to 13mg/dL. Our observations indicated that each group featured a unique spectral fingerprint. Next, we tested the index on 20 patients not included in the calibration procedure (unknown samples). We were able to classify patients into groups according to their Cr level with moderate prediction accuracy ( $R^{2} = 0.55$ ) and mean screening error of up to 16%. Future efforts will evaluate the accuracy of this approach with larger patient populations representing a broad range of Cr concentration. Still, this preliminary work is an essential step toward developing this useful noninvasive Cr screening platform using NIR light spectroscopy.

Keywords:

1. Introduction

Creatinine (Cr) is a by-product of the breakdown of phosphocreatine in muscle tissue and is removed from the bloodstream by the kidneys.^1,2 Under normal conditions, Cr is produced at a constant rate. Elevated levels of Cr in the blood ( $> 2$ mg/dL) can indicate reduced kidney function among other conditions. Typically, normal values of Cr range from 0.5mg/dL to 1mg/dL in females and from 0.6mg/dL to 1.2mg/dL in males. Since Cr undergoes passive filtration, it serves as a key indicator of kidney function and is approximately correlated to the rate of glomerular filtration by the kidneys.³ The importance of Cr monitoring to aid diagnosis of renal and muscular dysfunctions is widely recognized. The standard techniques utilized include the colorimetry-based Jaffe’s reaction, chromatography, enzymatic analysis, and biosensing, among others.^4,5,6 Several additional methods have been suggested over the years, including light spectroscopy.^7,8

Visible and infrared light spectroscopies offer promising platforms for Cr monitoring as they do not require any special preparation or handling during measurement.^9,10,11 Optical spectroscopy is an established tool, widely used in many fields including pharmacology, food quality control, biomedicine, and more.^12,13,15 A typical system configuration involves a single broadband light source, a portable spectrometer to record the attenuated spectrum, and a pair of optical fibers for the delivery and collection of light during sample measurement. The collected, diffused light contains valuable, quantitative information about sample structure and composition which can be extracted using inverse photon transport modeling.^16,17 In the clinical setting, many diseases can be simply diagnosed in real-time by monitoring the spectra derived from spectroscopy measurements to recognize the unique spectroscopic fingerprints of the disease.^18,19,20 Spectroscopic platforms are portable, relatively cheap, utilize nonionizing radiation, and have the potential to continuously monitor parameters such as Cr.^8,21,22 These systems have exceptional potential at points of care which do not house a biochemistry laboratory. Thus, the medical community has taken a significant interest in the potential for the use of light spectroscopy in a variety of medical fields.

In this paper, we demonstrate the use of light spectroscopy (between 400nm and 1000nm) followed by peak averaging to screen the Cr level of human patients. A peak average is calculated from the intensity of the recorded diffused transmitted light which passes through the subject’s finger across the visible–near-infrared (VisNIR) spectral region. Specifically, our experiments revealed three salient wavelengths at 551, 615, and 815nm where the intensity reaches its maximum. During the calibration process, a Cr index was generated based on the averaging of intensity values obtained from the fingers of 60 patients with varied Cr range and compared to Cr values obtained by a blood test. Then, the index was tested on 20 patients, not included during calibration, for validation. A summary of patients’ distributions with regard to their Cr concentration group is given in Table 1. We emphasize that our goal was not to measure an exact quantitative value for Cr concentration, as its prediction from the collected diffuse spectral information is not straightforward, rather to screen and classify patients into categories defined by the ranges of Cr value. Rapid screening is an important tool in the overall management of diseased patients and led to decreased mortality rates. Overall, experimental results demonstrate our ability to classify patients into their Cr level group with a moderate prediction accuracy of $R^{2} = 0.55$ and a mean absolute error of 16%. We believe strongly in the utility and applicability of this method, and therefore, future work will focus on increasing the accuracy in order to reliably distinguish between Cr concentrations, as well as further validation in human subjects. We propose this method as a triage approach that aims to inform medical service providers of the range in which a patient falls. This initial discrimination may be helpful in a variety of environments such as in examination units, elderly citizens’ home, medical clinics, imaging institutes, and even a patient’s home. This, to our knowledge, distinguishes our work in relation to the methods described in the literature for Cr monitoring.

Table 1. Distributions of participants in the calibration and validation analyses based on their creatinine group classification. Overall, 60 patients were involved in the calibration and 20 in the validation.
Group	Calibration	Validation
Cr 1	3
Cr 5–6	20	8
Cr 7–8	18	6
Cr 9–10	10	4
Cr 11–13	9	2
TOTAL →	60	20

The rest of this paper is organized as follows: Cr sample collection, system setup, and data processing are described in Sec. 2. Results of the experiments are presented and interpreted afterward in Sec. 3. Finally, conclusions are drawn in Sec. 4.

2. Material and Methods

2.1. Study protocol

The protocol was approved by the Institutional Review Board at the Edith Wolfson Hospital, Holon, Israel, and all participants signed an informed consent form. A total of 80 patients (aged 30–80 years with a mean age of 55 years) were recruited for participation, with Cr concentrations ranging from 0.96mg/dL to 12.5mg/dL. The Cr values collected by the assay were analyzed and compared to spectra and concentration values obtained from the spectrometry system on the morning of the same day. Based on the blood test, patients were classified into the group’s level (Cr group). For example, group 9–10 represents all patients with Cr values between 9mg/dL and 10mg/dL. Sixteen of the patients were undergoing chronic hemodialysis (HD) treatment and seven others were patients in the chronic kidney disease (CKD) outpatient clinic. The remaining subjects were considered healthy, with normal kidney function. The subjects were randomly assigned, regardless of their ethnicity, age, sex, or health status, to either the calibration ( $n = 60$ ) or validation ( $n = 20$ ) group. Sample collection from HD patients was done as a part of their monthly routine blood testing and CKD patients donated blood samples during their periodic visits to the CKD outpatient clinic at Edith Wolfson Hospital’s Department of Nephrology. Intensity values obtained from the fingers of 60 patients (calibration group) were averaged, compared to Cr blood test results, and used to generate a Cr intensity index. Then, the index was tested on 20 novel patients (validation group), and again compared to the blood test results, for validation.

2.2. Apparatus

A sketch of the compact, portable fiber-optic spectroscopy setup used in this work is given in Fig. 1(a). A photograph is shown in Fig. 1(b). The system is equipped with a broadband quartz–tungsten–halogen light source (HL-2000-FHSA; Ocean Optics, FL) for illumination, a portable monolithic miniature spectrometer (MMS1; Carl Zeiss) to record the diffuse transmitted spectra from the finger, optical fibers for light delivery and collection, a cuvette holder (Ocean Optics, FL) where the patient’s finger is inserted for measurement [Fig. 1(c)], and a laptop computer. The spectrometer’s spectral range is 300–1150nm with a wavelength accuracy of 0.3nm and a spectral resolution of 10nm. A photograph of the entire setup in place in the nephrology department is shown in Fig. 1(d). Spectra from the spectrometer were acquired with in-house scripts written using Matlab software (MathWorks, Natick, MA) and the spectrometer development kit. Each dataset was comprised of replicate spectra co-added to a single average spectrum to increase the effective signal-to-noise ratio. The spectrometer is controlled by a laptop computer running the Windows operating system and is connected via a USB 2.0 port.

2.3. Measurement workflow

Spectroscopy measurements in transmission mode were taken from the index finger of 80 human subjects. The diffuse transmitted intensity $I$ measured from the finger was normalized to account for spectral nonuniformity of the light source, fiber attenuation, detector response, and the influence of the dark current by the following equation :

T=I−DW−D<math display="block" altimg="eq-00026.gif"><mi>T</mi><mo>=</mo><mfrac><mrow><mi>I</mi><mo>−</mo><mi>D</mi></mrow><mrow><mi>W</mi><mo>−</mo><mi>D</mi></mrow></mfrac></math>(1)

where

$T$ is the corrected diffuse transmission spectrum,

$W$ is the spectrum obtained when the holder is empty (no finger inside), and

$D$ is the transmission acquired in the absence of any light (dark response of the spectrometer). To ensure luminous temporal stability, spectra were acquired at least 15 min after the source was turned on. Data was acquired every 2

s along the 1-min experiment for each subject, for a total of 30 spectra. During measurement, consecutive spectra were co-added together into a single, averaged spectrum thus reducing the noise by a factor of

$\sqrt n$ . Spectra were exported as an Excel file, transferred to Matlab software, and stored in the computer for later processing. After each measurement, the holder was cleaned with alcohol and then wiped dry using lint-free tissue paper. Reference measurements were made using a whole-blood analyzer located in the hospital biochemical laboratory. The spectra of each sample were compared to the blood concentration value obtained from the analyzer. The range of concentrations of all participants was 0.96–12.5-mg/dL Cr. Data collection was performed at a constant room temperature of

$\sim 2 5^{\circ}$ C.

2.4. Data processing

During measurements, we observed that Cr has several peaks along the VisNIR spectral region with a maximum value at three wavelengths of 551, 615, and 815nm. These peaks motivated us to define a simple Cr index based on averaging the maximum intensity of the diffusely transmitted light at those wavelengths. Table 2 presents the index for each Cr group obtained from 60 patients during the calibration processing (classification model). For example, group 5–6 represents all patients with Cr values between 5mg/dL and 6mg/dL, according to the blood test. For each group, the mean ( $μ$ ) $\pm$ standard error (SE) is presented and its interval value range between $μ$ −SE and $μ$ +SE is also given. As seen down the table, each group is characterized by a different index, which allows us to discriminate between the groups. However, the overlapping zones’ axis (derived from the table) shown indicates an overlap between group 5–6 and group 7–8 which may reduce the power of prediction. This limitation is addressed and solved in Sec. 3. The performance of the index was then evaluated, and the prediction was considered successful if the predicted value fell in the range of $μ$ $\pm$ SE for each group. The distribution of Cr into groups is not trivial. During this study we have tested multiple strategies for group divisions and the one presented in the paper was found to be optimal during calibration processing. For this reason, for instance, there is no group of Cr 8–9. We are aware of the difficulty in finding the true division of groups that will increase our classification accuracy. In our future work, we hope to overcome this limitation. In addition to the method of peak intensity averaging presented here for the generation of the Cr index, we attempted to use additional approaches including geometrical averaging, harmonic averaging, intensity ratio, normalized intensity, and others, but determined each to be a less successful method of prediction than the one we present here. To demonstrate the power of our Cr index to predict the Cr value of a new sample, not included in the calibration, the Cr index was tested on 20 subjects, as will be outlined in the next section. The datasets in the validation group included Cr concentrations ranging from 5mg/dL to 13mg/dL. All processing and mathematical calculations were carried out using Matlab.

Table 2. Summary of creatinine indices generated during calibration. (Left) Indices per group and per wavelength; (right) $μ$ $\pm$ SE. The ranges of indices from low to high are also given. As shown, each group is characterized by a different index value. For visualization, an illustration of the range of each Cr index group is shown below. The overlap between groups 5–6 and 7–8 can be seen.

3. Results and Discussion

Figure 2(a) shows a representative example of nine diffuse transmittance intensity spectrum graphs in the VisNIR region of group 11–13 (11mg/dL $\leq$ Cr $\leq$ 13mg/dL). Each spectrum is the average of 30 spectra, corresponding to 1-min measurements. The average spectrum obtained from eight spectra is seen by the broken-line graph and denoted by the arrow on the plot. The overall appearance of the spectrum reveals few peaks, with dominant peaks apparent at 551, 615, and 815nm. At 551nm and 615nm, the absorption of deoxyhemoglobin is higher than that of oxyhemoglobin and therefore at these wavelengths the measurement is more sensitive to variation in deoxyhemoglobin. However, at 815nm, both oxy- and deoxy-hemoglobin absorb light equally (isosbestic point) and therefore changes in intensity are proportional to blood volume or total hemoglobin concentration.^23,24 In Fig. 2(b), the average spectra of all Cr groups are given. The overall shapes of all spectra are similar, with noticeable variations in intensities between the groups. Overall, clear spectral variations are observed among different Cr levels. In the visible wavelength region, oxyhemoglobin absorbance displays characteristic maxima at 542nm and 577nm while deoxyhemoglobin absorbance exhibits a single maximum at 555nm. These peaks can be observed in Fig. 2(b), however, they do not appear as expected since most of the samples in this study were taken from HD patients, with the remaining samples taken from patients with CKD. In these patients the level of hemoglobin is maintained by weekly intravenous (IV) administration of the protein hormone (Erythropoietin). IV iron is also given if needed. This is true for CKD patients as well. From these reasons we believe that the hemoglobin spectrum is modified and the typical double peaks are not clearly visible or, alternatively, masked. As evidence by directions of change in intensity, the spectra vary by Cr group while maintaining the three dominant peaks. Actually, each group has its own unique diffuse intensity spectra that discriminate it from others. Thus, we conclude that the intensity profile of the spectra can aid in estimating the Cr level. Furthermore, we observed a 26% difference in intensity magnitude between the peaks ( $\sim 2700 \to \sim 2000$ ) of the three wavelengths across the groups which may serve as an additional factor of examination. A spectrum was considered an outlier, and excluded during spectral averaging, if its mean value was greater than $3 \times$ SE away from the average value of the mean spectra. Outlier detection was applied to ensure the quality of the data used to construct the calibration model.²⁵ We consider the first step toward obtaining a high-quality Cr index to be the detection and exclusion of outliers.

Fig. 2. (a) Representative graphs of nine measured spectra in the group 11–13 are shown. Each spectrum is the mean of 30 spectra. The dotted line indicated by the arrow is the average spectrum obtained from eight spectra (one outlier spectrum was omitted). Three dominant peaks are visible at 551, 615, and 815nm. (b) The average spectra of all patients across all Cr groups. The overall shapes of all spectra are similar to each other, with variations in intensities.

To gauge the ability of our index to predict the Cr concentration of a novel patient, we carried out an experiment using 20 subjects with varying Cr concentrations. Here we present two representative examples. Figure 3(a) shows the first example, the intensity spectrum graph of a male patient with a Cr level of 11.8mg/dL as determined by the blood test. The labeled ( $x$ , $y$ ) coordinates at the three wavelengths are denoted by the arrows on the plot. The value calculated by the Cr index is 2239 [(2451+2247+2020)/3] and the $y$ -axis of the graph is between 1500 and 2500. Based on Table 2 and the plot summarized in Fig. 2(b), we concluded that the Cr is in the range of 11–13mg/dL which is consistent with the true value. A second experiment, with a female subject with a Cr value of 5.97mg/dL, was performed and the intensity spectrum measured is displayed in Fig. 3(b). The Cr index for this case was determined to be 2186 [(2000+2468+2090)/3]. In this case, we conclude that the Cr level is in the range of 11–13mg/dL which is, unfortunately, not consistent with the actual value.

Fig. 3. (a) Intensity spectrum graph of a male patient with Cr = 11.8 mg/dL. The labeled (x,y) coordinates at the three wavelengths are denoted by the arrows. (b) Same as in (a) for a female patient with Cr = 5.97 mg/dL.

Figure 4 shows a plot of Cr predictions versus actual (reference) concentrations over 20 patients. As can be seen from the figure, only eight predictive values (marked by circles) were found to be well fitting in the range of the corresponded group. To further evaluate the predictive ability of our index, the root-mean-square error of prediction (RMSEP) metric, Eq. (2), was used and revealed a small value of 1.85mg/dL,

RMSEP=√∑ni=1(Cp−Ca)2n<math display="block" altimg="eq-00070.gif"><mstyle><mtext mathvariant="normal">RMSEP</mtext></mstyle><mo>=</mo><msqrt><mrow><mfrac><mrow><msubsup><mrow><mo>∑</mo></mrow><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>n</mi></mrow></msubsup><msup><mrow><mfenced separators="" open="(" close=")"><mrow><msub><mrow><mi>C</mi></mrow><mrow><mi>p</mi></mrow></msub><mo>−</mo><msub><mrow><mi>C</mi></mrow><mrow><mi>a</mi></mrow></msub></mrow></mfenced></mrow><mrow><mn>2</mn></mrow></msup></mrow><mrow><mi>n</mi></mrow></mfrac></mrow></msqrt></math>(2)

Fig. 4. Scatter plot of measured versus predicted concentrations of creatinine over 20 patients ( $n$ = 20). Eight predictive values, marked by circles, are found to be well fitting in the range of the corresponded group. Validation statistics are presented.

where $C_{a}$ denotes the actual concentration, $C_{p}$ denotes the predicted concentration, and $n$ represents the number of predictions for a given concentration. The mean absolute error (MAE) in the validated data was also calculated, indicating about 16% error and a positive correlation coefficient, $R^{2}$ , of 0.55, showing the fair capability of the Cr index to discriminate between Cr groups. MAE and $R^{2}$ were calculated as follows :

MAE=1nn∑i=1|Cp−CaCa|<math display="block" altimg="eq-00077.gif"><mstyle><mtext mathvariant="normal">MAE</mtext></mstyle><mo>=</mo><mfrac><mrow><mn>1</mn></mrow><mrow><mi>n</mi></mrow></mfrac><munderover accentunder="true" accent="true"><mrow><mo>∑</mo></mrow><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>n</mi></mrow></munderover><mfenced separators="" open="|" close="|"><mrow><mfrac><mrow><msub><mrow><mi>C</mi></mrow><mrow><mi>p</mi></mrow></msub><mo>−</mo><msub><mrow><mi>C</mi></mrow><mrow><mi>a</mi></mrow></msub></mrow><mrow><msub><mrow><mi>C</mi></mrow><mrow><mi>a</mi></mrow></msub></mrow></mfrac></mrow></mfenced></math>(3)

R2=∑ni=1(Cp−C)2∑ni=1(Ca−C)2<math display="block" altimg="eq-00078.gif"><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup><mo>=</mo><mfrac><mrow><msubsup><mrow><mo>∑</mo></mrow><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>n</mi></mrow></msubsup><msup><mrow><mo stretchy="false">(</mo><msub><mrow><mi>C</mi></mrow><mrow><mi>p</mi></mrow></msub><mo>−</mo><mi>C</mi><mo stretchy="false">)</mo></mrow><mrow><mn>2</mn></mrow></msup></mrow><mrow><msubsup><mrow><mo>∑</mo></mrow><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>n</mi></mrow></msubsup><msup><mrow><mo stretchy="false">(</mo><msub><mrow><mi>C</mi></mrow><mrow><mi>a</mi></mrow></msub><mo>−</mo><mi>C</mi><mo stretchy="false">)</mo></mrow><mrow><mn>2</mn></mrow></msup></mrow></mfrac></math>(4)

where $C$ is the mean of actual concentration in the prediction set. A summary of validation statistics is shown in the figure which demonstrates that, in general, our method performs moderately well. One of the causes of this relatively small level of accuracy may be the small number of patients utilized in the calibration process. Therefore, increasing the number of subjects with significantly varied values may improve the prediction accuracy. Other factors which may have contributed to the low level of accuracy include a variety of biological parameters such as blood type, hydration level, skin pigmentation, and others, as well as variation in physiological conditions affecting the overall spectral quality. Other factors that may affect the predictive capability of our index include interference from the background, such as light source drift and temperature change. Furthermore, finger thickness (light penetration depth) and patient movement during measurement could potentially affect the results. Future work will aim to neutralize these effects.

4. Summary

In summary, this study has demonstrated the feasibility of using NIR spectroscopy in the diffuse transmission mode combined with spectrum peak intensity averaging to rapidly screen patients for their Cr level. Measurements were noninvasively obtained from a total of 80 subjects. Among the 80 subjects, 60 were selected randomly for the calibration set, and the remaining 20 subjects were included in the validation set. Creatinine is one of the main components of human urine and is a valuable clinical indicator of kidney dysfunction. Therefore, real-time estimation of Cr level is important to indicate organ functional status especially for chronic HD patients and those who suffer from CKD. We emphasize that the goal of this work was not to quantitatively determine the Cr level but rather to rapidly assign patients to a Cr range group in a noninvasive manner. The overall results indicate that Cr can be screened by this technique and that the prediction accuracy was fair with a mean screening error of 16% ( $R^{2} = 0.55$ ). Results obtained using our system were compared to the results obtained by the standard biochemical assay (Cobas 8000 modular analyzer) used to measure Cr concentration in the hospital laboratory. Because of the extensive variation in the patients’ health condition (patient-to-patient variability) and the relatively small number of participants (hospital restrictions), it is emphasized that further work must be performed to improve the robustness and accuracy and validate the method described here. A more accurate prediction can be expected when the calibration dataset includes a variable interference (which affects the overall spectra quality), a wider range of Cr levels, and a larger number of subjects to ensure a compositionally wide enough dataset. Thus, future work will be focused on applying these calibration strategies for more conclusive results. Nevertheless, we believe that the concept presented here has the potential to discriminate and classify patients based on their Cr level. This method is fast, easy to implement, and carries a relatively low instrumentation cost, and therefore has the potential to be a useful addition to existing medical triage and routine patient supervision procedures.

Conflicts of Interest

The authors declare that there is no conflict of interest relevant to this article.

References

1. R. Cooper, F. Naclerio, J. Allgrove, A. Jimenez, “Creatine supplementation with specific view to exercise/sports performance: An update,” J. Int. Soc. Sports. Nutr. 9(1), 33 (2012). Web of Science, Google Scholar
2. G. Winnett, L. Cranfield, M. Almond, “Apparent renal disease due to elevated creatinine levels associated with the use of boldenone,” Nephrol. Dial. Transplant. 26(2), 744–747 (2011). Web of Science, Google Scholar
3. I. H. Perkins, C. M. Chapman, Creatinine: Production, Diagnostic Uses and Role in Renal Disease, Nova Science Publishers, New York (2012). Google Scholar
4. M. Peake, M. Whiting, “Measurement of serum creatinine — current status and future goals,” Clin. Biochem. Rev. 27(4), 173–184 (2006). Google Scholar
5. E. Cholongitas, L. Marelli, A. Kerry, M. Senzolo, D. W. Goodier, D. Nair, M. Thomas, D. Patch, A. K. Burroughs, “Different methods of creatinine measurement significantly affect MELD scores,” Liver Transpl. 13(4), 523–529 (2007). Web of Science, Google Scholar
6. D. Tambaru, R. H. Rupilu, F. Nitti, I. Gauru, S. Suwari, “Development of paper-based sensor coupled with smartphone detector for simple creatinine determination,” AIP Conf. Proc. 1823, 020095 (2017). Google Scholar
7. R. A. Shaw, S. Kotowich, H. H. Mantsch, M. Leroux, “Quantitation of protein, creatinine, and urea in urine by near-infrared spectroscopy,” Clin. Biochem. 29(1), 11–19 (1996). Web of Science, Google Scholar
8. J. L. Pezzaniti, T. W. Jeng, L. McDowell, G. M. Oosta, “Preliminary investigation of near-infrared spectroscopic measurements of urea, creatinine, glucose, protein, and ketone in urine,” Clin. Biochem. 34(3), 239–246 (2001). Web of Science, Google Scholar
9. M. L. de Almeida, C. J. Saatkamp, A. B. Fernandes, A. L. B. Pinheiro, L. Silveira, Jr., “Estimating the concentration of urea and creatinine in the human serum of normal and dialysis patients through Raman spectroscopy,” Lasers Med. Sci. 31(7), 1415–1423 (2016). Web of Science, Google Scholar
10. D. P. Jerônimo, R. A. de Souza, F. F. da Silva, G. L. Camargo, H. L. Miranda, M. Xavier, K. K. Sakane, W. Ribeiro, “Detection of creatine in rat muscle by FTIR spectroscopy,” Ann. Biomed. Eng. 40(9), 2069–2077 (2012). Web of Science, Google Scholar
11. D. Gangopadhyay, P. Sharma, R. Nandi, M. Das, S. Ghosh, R. K. Singh, “In vitro concentration dependent detection of creatinine: A surface enhanced Raman scattering and fluorescence study,” RSC Adv. 6, 112562–112567 (2016). Web of Science, Google Scholar
12. Y.-S. Li, J. S. Church, “Raman spectroscopy in the analysis of food and pharmaceutical nanomaterials,” J. Food Drug Anal. 22(1), 29–48 (2014). Web of Science, Google Scholar
13. N. Lewen, “The use of atomic spectroscopy in the pharmaceutical industry for the determination of trace elements in pharmaceuticals,” J. Pharm. Biomed. Anal. 55(4), 653–661 (2011). Web of Science, Google Scholar
14. E. W. Ciurczak, “Growth of near-infrared spectroscopy in pharmaceutical and medical sciences,” Proc. SPIE 4626, 116–125 (2002). Google Scholar
15. D. A. Boas, C. Pitris, N. Ramanujam (Eds.), Handbook of Biomedical Optics, 1st Edition, CRC Press, Boca Raton (2011). Google Scholar
16. C.-Y. Wang, T.-C. Kao, Y.-F. Chen, W.-W. Su, H.-J. Shen, K.-B. Sung, “Validation of an inverse fitting method of diffuse reflectance spectroscopy to quantify multi-layered skin optical properties,” Photonics 6(2), 61 (2019). Web of Science, Google Scholar
17. D. C. Aiken, S. Ramsey, T. Mayo, J. Bellemare, S. G. Lambrakos, J. Peak, “Inverse analysis of near infrared transmission spectra for triarylamine, tetraaryldiamine, nickel dithiolene and indolium-iodide dyes,” J. Near Infrared Spectrosc. 23(3), 123–132 (2015). Web of Science, Google Scholar
18. J. T. Motz, S. J. Gandhi, O. R. Scepanovic, A. S. Haka, J. R. Kramer, Jr., R. R. Dasari, M. S. Feld, “Real-time Raman system for in vivo disease diagnosis,” J. Biomed. Opt. 10(3), 031113 (2007). Web of Science, Google Scholar
19. A. Y. Khaled, S. A. Aziz, S. K. Bejo, N. M. Nawi, I. A. Seman, D. I. Onwude, “Early detection of diseases in plant tissue using spectroscopy — applications and limitations,” Appl. Spectrosc. Rev. 53(1), 36–64 (2018). Web of Science, Google Scholar
20. A. E. Cerussi, N. S. Shah, D. Hsiang, A. Durkin, J. A. Butler, B. J. Tromberg, “In vivo absorption, scattering, and physiologic properties of 58 malignant breast tumors determined by broadband diffuse optical spectroscopy,” J. Biomed. Opt. 11(4), 044005 (2006). Web of Science, Google Scholar
21. C. J. Saatkamp, M. L. de Almeida, J. A. M. Bispo, A. L. B. Pinheiro, A. B. Fernandes, L. Silveira, Jr., “Quantifying creatinine and urea in human urine through Raman spectroscopy aiming at diagnosis of kidney disease,” J. Biomed. Opt. 21(3), 037001 (2016). Web of Science, Google Scholar
22. Z. H. Barnea, D. Abookasis, “Determination of creatinine level in patient blood samples by Fourier NIR spectroscopy and multivariate analysis in comparison with biochemical assay,” J. Innov. Opt. Health Sci. 12(6), 1950015 (2019). Link, Web of Science, Google Scholar
23. I. J. Bigio, S. Fantini, Overview of tissue optical properties, Quantitative Biomedical Optics: Theory, Methods, and Applications, pp. 19–59, Cambridge University Press, Cambridge (2016). Google Scholar
24. R. Nachabé, B. H. W. Hendriks, M. van der Voort, A. E. Desjardins, H. J. C. M. Sterenborg, “Estimation of biological chromophores using diffuse optical spectroscopy: Benefit of extending the UV-VIS wavelength range to include 1000 to 1600nm,” Opt. Express 18(24), 1432–1442 (2010). Web of Science, Google Scholar
25. P. J. Rousseeuw, A. M. Leroy, Robust Regression and Outlier Detection, John Wiley & Sons, Inc., New York (1987). Google Scholar

Vol. 13, No. 06

Metrics

Downloaded 785 times

History

Received 10 May 2020

Accepted 30 August 2020

Published: 1 October 2020

Information

This is an Open Access article. It is distributed under the terms of the Creative Commons Attribution 4.0 (CC-BY) License. Further distribution of this work is permitted, provided the original work is properly cited.

Keywords

PDF download