1The authors are with The Institute of Optics, University of Rochester, Rochester, New York 14627. D. Qi is also with the Department of Physics and Astronomy, University of Rochester, Rochester, New York 14627. A. J. Berger's e-mail address is ajberger@optics.rochester.edu.
Dahu Qi and Andrew J. Berger, "Chemical concentration measurement in blood serum and urine samples using liquid-core optical fiber Raman spectroscopy," Appl. Opt. 46, 1726-1734 (2007)
We report measurements of chemical concentrations in clinical blood serum and urine samples using liquid-core optical fiber (LCOF) Raman spectroscopy to increase the collected signal strength. Both Raman and absorption spectra were acquired in the near-infrared region using the LCOF geometry.
Spectra of 71 blood serum and 61 urine samples were regressed via partial least squares against reference analyzer values. Significant correlation was found between predicted and reference concentrations for 13 chemicals. Using absorption data to normalize the LCOF enhancement made the results more accurate. The experimental geometry is well suited for high-volume and automated chemical analysis of clear biofluids.
Annika M. K. Enejder, Tae-Woong Koo, Jeankun Oh, Martin Hunter, Slobodan Sasic, Michael S. Feld, and Gary L. Horowitz Opt. Lett. 27(22) 2004-2006 (2002)
You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.
The last column gives the reported reproducibility error at a representative concentration.
Globulin concentration is a derived quantity (TP − albumin), not measured directly.
TB concentration accuracy is limited by the equipment's digitization noise (0.1 mg∕dl).
Table 2
Comparison of Prediction Errors Using Corrected Raman Spectra RMSECVc with Reference Analyzer Errors Eref and Prediction Errors Using Direct Raman Spectra RMSECVd
Chemical
RMSECVc (mg∕dl)
gc
Eref (mg∕dl)
RMSECVd (mg∕dl)
Δg
Spectral Range (cm−1)
Cholesterol
4.0(15)
10.5
4.1
×
9.6(10)
Y
6.1
510–1500
TP
145(7)
3.5
148
×
182(9)
Y
0.7
610–1725
Albumin
83(15)
4.7
87
×
123(17)
Y
2.2
910–1800
CO2 (mEq∕l)
2.3(12)
1.9
1.9
×
2.2(10)
0
900–1800
BUN
1.5(15)
3.3
1.6
×
1.7(14)
0.7
510–1800
TB
0.09(16)
2.8
0.1
×
0.13(15)
Y
0.9
960–1600
UUN
40.0(4)
7.0
42.4
×
53.0(3)
Y
1.8
550–1800
Creatinine
4.3(17)
9.3
5.1
×
6.8(9)
Y
3.4
550–1800
Globulin
110(11)
4.1
—
—
130(11)
0.6
1000–1800
Glucose
8.8(14)
5.3
3.8
15.4(12)
Y
2.2
510–1320
Triglyceride
11.5(14)
6.1
8.0
18.0(11)
Y
2.2
510–1800
HDL*
11.9(13)
1.5
3.4
12.1(10)
0
510–1200
LDL*
12.4(12)
2.7
7.5
11.2(11)
−0.3
510–1200
Differences in error levels were assessed by an f test at the 95% confidence level. “×” indicates analytes whose RMSECVc reached the reference error limit . “Y” indicates analytes whose prediction error was significantly reduced by the correction methods . None was significantly increased. Correction also tended to increase g, as shown in the Δg column, ranging from 0.9 to 6.1. Ranks of the PLS cross validation are shown together with the RMSECV values in parentheses. Asterisks indicate analytes whose concentrations probably changed substantially between the reference and the Raman measurements.
Table 3
Characteristic Time tc at Which Shot Noise Causes About Half of the Total Prediction Error
Samples
Chemicals
tc (s)
RMSECVc(tc) (mg∕dl)
RMSECVc(2tc) (mg∕dl)
RMSECVc(tmax) (mg∕dl)
Blood serum (tmax = 150 s)
Cholesterol
36
5.6
4.5
4.0
Glucose
23
11.9
10.3
8.8
Triglyceride
9
18.6
14.9
11.5
Albumin
3
141
126
83
TB
9
0.16
0.13
0.09
BUN
9
2.2
1.9
1.5
TP
3
157
151
145
Globulin
3
132
127
110
LDL
6
16.9
14.3
12.4
CO2 (mEq∕l)
3
3.1
2.8
2.3
HDL
3
14.6
14.4
11.9
Urine (tmax = 64 s)
UUN
2
39.5
39.0
40.0
Creatinine
2
6.2
5.8
4.3
The last three columns indicate how RMSECV increases as t is raised from tc to tmax.
Tables (3)
Table 1
Statistical Distribution of Analyte Concentration Values in the Blood and Urine Specimen Groups
The last column gives the reported reproducibility error at a representative concentration.
Globulin concentration is a derived quantity (TP − albumin), not measured directly.
TB concentration accuracy is limited by the equipment's digitization noise (0.1 mg∕dl).
Table 2
Comparison of Prediction Errors Using Corrected Raman Spectra RMSECVc with Reference Analyzer Errors Eref and Prediction Errors Using Direct Raman Spectra RMSECVd
Chemical
RMSECVc (mg∕dl)
gc
Eref (mg∕dl)
RMSECVd (mg∕dl)
Δg
Spectral Range (cm−1)
Cholesterol
4.0(15)
10.5
4.1
×
9.6(10)
Y
6.1
510–1500
TP
145(7)
3.5
148
×
182(9)
Y
0.7
610–1725
Albumin
83(15)
4.7
87
×
123(17)
Y
2.2
910–1800
CO2 (mEq∕l)
2.3(12)
1.9
1.9
×
2.2(10)
0
900–1800
BUN
1.5(15)
3.3
1.6
×
1.7(14)
0.7
510–1800
TB
0.09(16)
2.8
0.1
×
0.13(15)
Y
0.9
960–1600
UUN
40.0(4)
7.0
42.4
×
53.0(3)
Y
1.8
550–1800
Creatinine
4.3(17)
9.3
5.1
×
6.8(9)
Y
3.4
550–1800
Globulin
110(11)
4.1
—
—
130(11)
0.6
1000–1800
Glucose
8.8(14)
5.3
3.8
15.4(12)
Y
2.2
510–1320
Triglyceride
11.5(14)
6.1
8.0
18.0(11)
Y
2.2
510–1800
HDL*
11.9(13)
1.5
3.4
12.1(10)
0
510–1200
LDL*
12.4(12)
2.7
7.5
11.2(11)
−0.3
510–1200
Differences in error levels were assessed by an f test at the 95% confidence level. “×” indicates analytes whose RMSECVc reached the reference error limit . “Y” indicates analytes whose prediction error was significantly reduced by the correction methods . None was significantly increased. Correction also tended to increase g, as shown in the Δg column, ranging from 0.9 to 6.1. Ranks of the PLS cross validation are shown together with the RMSECV values in parentheses. Asterisks indicate analytes whose concentrations probably changed substantially between the reference and the Raman measurements.
Table 3
Characteristic Time tc at Which Shot Noise Causes About Half of the Total Prediction Error
Samples
Chemicals
tc (s)
RMSECVc(tc) (mg∕dl)
RMSECVc(2tc) (mg∕dl)
RMSECVc(tmax) (mg∕dl)
Blood serum (tmax = 150 s)
Cholesterol
36
5.6
4.5
4.0
Glucose
23
11.9
10.3
8.8
Triglyceride
9
18.6
14.9
11.5
Albumin
3
141
126
83
TB
9
0.16
0.13
0.09
BUN
9
2.2
1.9
1.5
TP
3
157
151
145
Globulin
3
132
127
110
LDL
6
16.9
14.3
12.4
CO2 (mEq∕l)
3
3.1
2.8
2.3
HDL
3
14.6
14.4
11.9
Urine (tmax = 64 s)
UUN
2
39.5
39.0
40.0
Creatinine
2
6.2
5.8
4.3
The last three columns indicate how RMSECV increases as t is raised from tc to tmax.