Detection of Bilirubin using Raman Spectroscopy in a Neonatal SkullRoarke Horstmeyer, Bob Guenther, Hyun-Joong Kim, Scott McCain

Duke Imaging and Spectroscopy Program, Duke University, Durham NC 27708

IntroductionBilirubin is a naturally occurring chemical in the human body, whose main function

is thought to be as cellular antioxidant1. In high enough concentrations, though,

especially in areas around the brain, its can lead to jaundice, neurological defects,

seizures and even death. This accumulation, known as hyperbilirubinemia, is

mainly a symptom of newborn infants, who lack the intestinal bacteria that

facilitate the breakdown of bilirubin into a soluble substance. Thus, it is of extreme

importance for doctors to be able to effectively measure bilirubin concentrations in

the neonatal skull.

Currently, blood samples drawn

from an infant‟s heel or visual

indications from skin coloration

are used to determine bilirubin

levels. When found in excess,

phototherapy is used to help reduce

levels. Bilirubin is extremely sensitive to light, and isomerizes into an excretable

substance (lumirubin) when exposed to blue light 3.

The aim of this project was to develop a non-invasive method of detecting bilirubin

concentrations that takes advantage of this property of photoisomerization. Raman

spectroscopy was used to obtain optical spectra of bilirubin at different stages of

isomerization. When near-infrared (NIR) excitation is used, Raman spectroscopy

has the potential ability to penetrate through the skull of a neonate to provide useful

information concerning intracranial fluids4. To investigate this possibility, a

phantom skull was created to mimic the optical behavior of human bone. Our

findings show the feasibility of measuring the photoisomerization of bilirubin non-

invasively and initial results on modelling the optical properties of a human skull.

ResultsBilirubin has several major peaks in the Raman spectrum that

can be used as probes6 (fig. 4). For this experiment, peaks at

wave numbers shown in Table 1 were chosen. They can be

categorized into 3 different groups:

Group A: C=C double bonds stretching in a five member ring

Group B: Bending motion in CH2 and umbrella motion in CH3

Group C: Bending of both C-OH and C-CH2

Over 2 hours of exposure, the bilirubin concentration in our sample decreased from 24.4 mg/dL to 17.5 mg/dL. Changes in the Raman

spectrum can be seen in Figure 5. A partial least square regression vector in Fig. 6 shows the correlation between spectral channels and

bilirubin concentration. Note the decrease in intensity for group B peaks and increase in group A and C peaks, as expected with a

change from bilirubin into lumirubin. These changes in peak intensity highlight the possibility of quantitative determination of

bilirubin concentration.

Creation of a Phantom SkullMRI images from a patient were processed to extract skull segments and combined to create

a 3D image. The image was then reconstructed on a Solidworks machine to produce a

plastic model “phantom” skull.

The thickness of this phantom skull can be altered to mimic the optical properties of bone.

The intensity of light passing through a thin, dense material obeys the equation7,

I = Io e – µs d

Where µs is the scattering coefficient of

the material and d is its thickness

(see fig. 8).

A phantom infant skull with an average

thickness of approximately 3.41 mm is

the best optical model for this project.

Methods and MaterialsTwo types of Bilirubin samples were prepared with a 20mg/dL concentration, the lethal threshold for neonates5:

- Mixed with 1M NaOH, 5ml H20, and 75 µl Hcl

- Mixed with 2% Intra Lipid solution (to simulate in vivo scattering properties)

The samples were placed in a closed box (fig. 3) and exposed to blue light emitted from 8 LED‟s (at 800 µW/cm2 to simulate typical

clinical lighting equipment3) for 4 hours.

The Raman spectra of the samples were measured

every 30 minutes using HAMMER (Hadamard

Aperture Mask Multi-Excitation Raman) system (fig. 2).

The changes in bilirubin concentration were measured

and fluctuations in the Raman spectrum peak height

were analyzed using a partial least squares regression.

Frequency Number

(cm-1

)

Group Description

491.5 C Mixing of C-OH and C-CH2 bending

1339.4 CCH bending in CH3

1450 C-C and C-N mixed stretching Umbrella

motion in CH3 group

1494

B

C=C stretching in ring and C-C stretching between ring and CH=CH2

1611.4 A C=C stretching in five-membered ring

References1. D Baranano, M Rao, C Ferris, S Snyder. Biliverdin reductase: a major physiologic cytoprotectant. Proc Natl Acad

Sci USA v.99(25) Dec. 10, 2002

2. AF McDonaugh, DA Lightner. „Like a Shrivelled Blood Orange‟ – Bilirubin, Jaundice, and Phototherapy,

Pediatrics 1985; 75; 443-455

3. H Rose, A Rosen, D Rosen, B Onaral, M Hiatt. Use of a Light Emitting Diode (LED) Array for Bilirubin

Phototransformation.

4. R Aslin, M Jacques. Near-infrared spectroscopy for functional studies of brain activity in

human infants: promise, prospects, and challenges. Biomedical Optics V.10 J/F 2005

5.

6. KR Amareshwar, SB Rai, DK Rai, VB Singh. Spectrostopic studies and normal coordinate analysis of bilirubin,

Spectrochimica Acta Part A 58 (2002) 2145–2152

7. N Ugryumova, S Matcher, D Attenburrow. Measurement of bone mineral density via light scattering. Phys. Med.

Biol. 49 (2004) 469-483

8. Y Ogoshi, E Okada. Analysis of light propagation in a realistic head model by a hybrid method for optical brain

function measurement. Optical Review vol. 12 3 (2004) 264-269

9. J Ruan, P Prisad. The Effects of Skull Thickness Variations on Human Head Dynamic Impact Responses. SAE

International no. 2001-22-0018, 2001

10. P Inveen, T Reilly, J Tan. Device to simultaneously cut and coagulate infant skull during endoscopic

craniosynostosis operation. Univ. Wisconsin Madison Proj. May 2001

11. S. T. McCain, M. E. Gehm, Y. Wang, N. P. Pitsianis, D. J. Brady. Multimodal multiplex Raman spectroscopy

optimized for in vivo chemometrics, Proc. SPIE Int. Soc. Opt. Eng. 6093, 60930P (2006)

12. S. T. McCain, M. E. Gehm, Y. Wang, N. P. Pitsianis, D. J. Brady. Coded Aperture Raman Spectroscopy for

Quantitative Measurements of Ethanol in a Tissue Phantom, Appl. Spectrosc. *60*, 663-671 (2006)

ConclusionsOur hypothesis was supported: Raman spectroscopy can non-invasively measure the

decrease in bilirubin concentration during phototherapy.

•Changes in key peaks of bilirubin's spectrum due to photoisomerization can lead to

quantitative evaluation of bilirubin concentration.

•Further research into possibility of non-invasive detection through a neonatal skull is

possible with the creation of a phantom skull that mimics the optical properties of human

bone, as is research into other intracranial phenomena.

•Future research: Non-invasive detection through phantom skull, creation and use of

substance to mimic properties of intracranial fluids, quantitative analysis of Raman peak

changes to determine bilirubin concentration.

AcknowledgmentsWe would like to thank Dr. David Brady and Paul Vosburgh, both members of DISP, for

their help and support. Additionally, thanks to Dr. Robert. Pearlstein for providing MRI

images and helping with various medical-related questions, and Dr. David Tanaka for

advice and assistance on medical matters.

Figure 1: Diagram of the transformation of bilirubin (left) into

lumirubin as a result of photoisomerization2

Figure 2: HAMMER Raman spectrograph apparatus Figure 3: Photoisomerization box setup

Table 1: Bilirubin spectral peaks and associated physical cause6

Figure 4: Spectral fingerprint of 98% bilirubin

powder sample taken by HAMMER

Figure 6: Partial least square regression vector of figure 5.

Groups A and C-related peaks exhibit positive correlation, while

Group B-related peaks show negative correlation, as expected.

Figure 7: Image of phantom skull created on Solidworks

machine, scaled down ~50% from original size to mimic

optical properties of a neonatal skull

Scattering coefficient of human bone: 1.6mm-1(8)

Average adult skull thickness: 5mm 9

Average infant skull thickness: 1.75mm 10

From equation above,-ln(I/ Io) = µsd = 1.6*1.75 = 2.8

From best fit equation,d = (2.8+1.3)/1.2 = 3.41mm

Figure 8: Plot of experimental measurements of transmitted

infrared laser intensity vs. plastic thickness, with expected

intensity ratios for human skull. Followed by equations to find

optimal phantom skull thickness

Wave-number (cm-1)Wave-number (cm-1) Wave-number (cm-1)

Inte

nsity (

a.u

.)

Figure 5: Raman spectra of 20 mg/dL bilirubin sample at

different stages of photoisomerization. Exposures taken

every 30 min. for 4 hours.

Inte

nsity (

a.u

.)

Inte

nsity (

a.u

.)