Attenuation of near-IR light through dentin at wavelengths from 1300-1650-nm •
Andrew C. Chan, Cynthia L. Darling, Kenneth H.Chan and Daniel Fried
University of California, San Francisco, San Francisco, CA 94143-0758
ABSTRACT Light scattering in dental enamel decreases markedly from the UV to the near-IR and recent studies employing near-IR transillumination and reflectance imaging including optical coherence tomography indicate that this wavelength region is ideally suited for imaging dental caries due to the high transparency of enamel. The opacity of dentin is an important factor in optimizing the contrast of demineralization in reflectance measurements. It also influences the contrast of occlusal lesions in transillumination. Light scattering in dentin is an order of magnitude larger than in enamel, it is highly anisotropic and has a different spectral light scattering dependence than enamel. The objective of this study was to measure the optical attenuation of near-IR light through dentin at near-IR wavelengths from 1300-1650-nm. In this study the collimated transmission of near-IR light through polished thin sections of dentin of 0.05 to 0.6 mm thickness was measured. Beer-Lambert plots show that the attenuation coefficients range in magnitude from 20 to 40 cm-1. Attenuation increased significantly with increasing wavelength and the increases were not entirely consistent with increased water absorption. Keywords: light scattering, dentin, near-IR imaging
1. INTRODUCTION Several in vitro and in vivo studies have shown that the near-IR region is highly promising for imaging dental caries due to the high transparency of sound enamel 1-8. Upon demineralization, the scattering coefficient increases by almost two orders of magnitude which provides high contrast for imaging in both transillumination and reflectance 9. Optical coherence tomography has also been most successful at 1310nm for imaging teeth due to the high transparency of enamel 10-12. Light scattering in enamel appears to be dominated by Rayleigh scattering and the magnitude of Rayleigh scattering is inversely proportional to the fourth power of the wavelength13. Therefore, the magitude of light scattering in enamel decreases markedly from the UV to the near-IR. Light scattering in dentin is primarily due to the dentinal tubules which behave as cylindrical Mie scatterers. The wavelength dependence is more complex and measurements from 500-1050-nm do not show the marked reduction is scattering that has been observed for enamel 13-16. The amount of light reflected (backscattered) from the underlying dentin depends on the ratio of scattering to absorption. Assuming absorption is solely due to water in the near-IR, we postulate that longer near-IR wavelengths where scattering decreases and absorption increases will yield the highest contrast in reflectance. Hyperspectral reflectance measurements of Zakian 8 show that the dentin gets darker with increasing wavelength even when water absorption is not increasing. Previous angularly resolved optical measurements of sound dentin carried out at 543, 632 and 1050-nm indicated that the measured scattering and absorption coefficients of dentin are both almost an order of magnitude larger than for enamel13. In contrast with enamel, the scattering and absorption coefficients did not change significantly with •
Contact Author: [email protected]
Lasers in Dentistry XX, edited by Peter Rechmann, Daniel Fried, Proc. of SPIE Vol. 8929, 89290M · © 2014 SPIE · CCC code: 1605-7422/14/$18 · doi: 10.1117/12.2045629
Proc. of SPIE Vol. 8929 89290M-1 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 03/14/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
wavelength, which is consistentt with measurrements usingg an integratiing sphere, values v did varry signifiicantly with th he method of measurement, μs = 250–2880 cm-1 for anngularly resolvved goniometeer measuurements and 130–180 cm-1 for f integrating sphere measurrements. Valuues of the scatteering anisotroppy (g) alsso vary consid derably, Fried et al. 13report a g value of 0.93 that waas determined using angularlly resolvved goniometer measuremennts while Zijpp and ten Bosch report values v of 0.400 calculated by b integraating the measu ured scatteringg distributions of o 20-μm-thickk dentin samplees 15, 16. In ordder to accurateely measure thhe internally optical o properties it is extrem mely importannt to reduce thhe influennce of surfacee scattering. In I previous stuudies of dentall enamel, sam mples were placced in an indeex matchhing bath of th he same refracttive index, nam mely carbon disulfide d or a cargille c liquid 1, 13. Howeveer, dentinn poses an add ditional challennge since it is a highly poroous structure composed c of tuubule structurees whichh are typically filled with fluid f and it is the scatteringg of the tubulles that dominnate the optical properrties. Thereforre, if dentin waas immersed inn an index matcching fluid thee internal scatteering would alsso be greeatly reduced and a would not represent r the optical o behavioor expected clinnically. We thherefore chose to t either place the denttin sections in water or mainttain a layer of water on the surface s to best represent the in situ opptical environm ment. This appproach was takken in a prior study s involvinng highly porouus caries lesionns and demineralized d enamel wheree index matchhing agents would likely im mbibe into thee lesion greatlly reduciing the internaal scattering inn the lesion9. We were unabble to measuree all the samplles immersed in i water due to the ex xcessive attenuuation at some of the near-IR R wavelengthss where there is strong water absorpption, e.g. 1450 0-nm, thereforee we measuredd these samples immediately after immersioon in water, i.ee., with a water layer. The T study desiign emphasizess measurementt of relative diffferences in optical attenuatioon at nearr-IR wavelengtths from 1300--1650-nm.
2. MATERIA ALS AND METHODS M
A
I I
E
F
B
Fig. 1. The experimenttal setup for meassurement of the opptical attenuaation through thin n sections of denttin placed in a cuvvette. (A) Tungsten–halogen T lamp with (B B) filter wheel, (C) collimaating and focusin ng lenses, optical chopper (D), saample cuvettee (E) and (F) detecctor are shown.
Plano--parallel sectioons ranging from f 50-µm to t 600-µm thick were cut c from twentty-five extracteed molarss/premolars annd serially poolished to a 0.3 µm diamond grit finnish. Samples were stored inn a 1% thhymol solution to prevent deccay. Light from m an Ocean Optics (Dunedin, Fl) fiber-coupleed tungstten-halogen laamp (Model HL-2000-FHSA H A) with a filter wheel with w bandpass filters f (FBxxxxx12-FW WHM 12-nm)) Thorlabs (Newton, NJ) N centerred at 50-nm intervals i from m 1300-1650-nm m was ussed to illuminaate specific areas of the sample with a spot size off approximatelyy 100 µm. Thhe spot siize was confirm med through knife-edge k beam m profiliing techniques. Phase senssitive detectioon was employed e usingg a lock-in am mplifier (Moddel SR 8550) and an opptical chopperr from Stanforrd Researrch Systems (Stanford, ( CA)). A large-areea Ge phhotoreceiver Model M 2033 from New Focuus (Santaa Clara, CA)) equipped with w a variable apertuure was used foor detection. The T setup for thhe
attenuuation measurem ments is shownn in Fig. 1. Samplles were measu ured either fullly immersed inn a cuvette (1-ccm sq.) filled with water (n= =1.3) or wet. In I order to t measure thee samples with a wet surface the t sample wass wetted beforee each measureement by raisinng the cuuvette filled wiith water to weet the surface. This proceduure ensured that the positionn of the incidennt collim mated light beaam was fixed while w cycling through the filters. fi Fivee measurementts of collimateed transm mission were reecorded at varioous sections within w each sam mple. The collim mated signal (II) at the detectoor was coompared to thee initial intensitty of the beam m (Io). Using thhis ratio and thee thickness of the t samples (T T), the atttenuation coeffficient (μt) wass calculated usiing Beer's law and Beer-Lamb mbert plots:
Proc. of SPIE Vol. 8929 89290M-2 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 03/14/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
µt = - ln(I//Io)/ T
(11) Beeer's law statess that the inttensity of lighht passsing throughh a sampple decreasees expoonentially as thickness inncreases. Thhis overrall decrease is known as attenuation or o extinction, which is caused by suurface scatterinng s and diffuse d reflectiion at front annd or specular rearr surfaces, and a internal scattering annd absoorption. T The aperture was w sufficientlly smaall in front of thhe detector to ensure that lighht scatttered at smalll angles did not contribuute signnificantly to thee collimated traansmission.
.
3. RESULTS S
0.04
Tlìickì IC:m) Fig. 2. Beer-Lambert plot at 1300 nm forr dentin with a th hin layer of watter applied to the surface. Colliimated transmisssion for varying g sample thicknesss ranging from 50 5 to 600-μm are plotted. Each point p represents the t mean attenuaation of each of the five measurrements on each saample.
Collimated Transmission (In I/lo)
Figuures 2 and 3 contain c Beer-L Lambert plots at 1300-nm for sam mples with a water w layer annd fullyy immersed inn water. Thee values for thhe attennuation coefficcients were 23 ± 4.8 and 35 ± Correlation coefficients 10 respectively. r c (R R2) werre 0.49 and 0.34 and botth slopes werre nt from zero. The T slopes for all the Beer-Laambert plots foor the samples with the applieed signifiicantly differen water layer were sig gnificantly diffferent from zerro. This indicates that surface scattering does d not prevennt determ mination of th he internal scaattering and absorption a even though it is approximateely an order of o magniitude higher. Surface scatterinng is relativelyy independent of sample thicckness, therefore if attenuatioon is com mpletely domin nated by surfacce scattering, than t there wouuld be only a minimal m changge in attenuatioon with increasing i sam mple thickness.. This has been observed o for prrevious measurrements for enameel in the near-IR where thhe internal scatterring was extrem mely low and attenuation was dominated d by surface scatteering if the samplees were nott placed in an index matchhing fluid 1. Only the sloope for the 1300-nnm measurem ments was siignificantly differeent from zero for the denttin samples fully immersed i in water. w This wass due to the extrem mely low signaal intensity cauused by the absorpption of water in the cuvettess. Figure 4 contains a plot of th he attenuation coefficients c for eaach wavelengtth along with a plot of 0.04 water absorption taaken from thee paper of Ihlclkness (cm) 17 Hale and a Querry . The attenuaation values Fig. 3. Beer-L Lambert plot at 1300 nm for denttin immersed in lie beetween 20 and d 40 cm-1 annd increase water. Collimatted transmission for varying saample thickness ranging from 500 to 600-μm are plotted. p Each point represents the signifiicantly with increasing wavelength. w mean attenuation n of each of the fivve measurements on o each sample. Comparison of the mean attenuattion values with ANOVA ind dicates that there t is a signifiicant linear inccrease in attenuuation with increaasing waveleng gth (P
Andrew C. Chan, Cynthia L. Darling, Kenneth H.Chan and Daniel Fried
University of California, San Francisco, San Francisco, CA 94143-0758
ABSTRACT Light scattering in dental enamel decreases markedly from the UV to the near-IR and recent studies employing near-IR transillumination and reflectance imaging including optical coherence tomography indicate that this wavelength region is ideally suited for imaging dental caries due to the high transparency of enamel. The opacity of dentin is an important factor in optimizing the contrast of demineralization in reflectance measurements. It also influences the contrast of occlusal lesions in transillumination. Light scattering in dentin is an order of magnitude larger than in enamel, it is highly anisotropic and has a different spectral light scattering dependence than enamel. The objective of this study was to measure the optical attenuation of near-IR light through dentin at near-IR wavelengths from 1300-1650-nm. In this study the collimated transmission of near-IR light through polished thin sections of dentin of 0.05 to 0.6 mm thickness was measured. Beer-Lambert plots show that the attenuation coefficients range in magnitude from 20 to 40 cm-1. Attenuation increased significantly with increasing wavelength and the increases were not entirely consistent with increased water absorption. Keywords: light scattering, dentin, near-IR imaging
1. INTRODUCTION Several in vitro and in vivo studies have shown that the near-IR region is highly promising for imaging dental caries due to the high transparency of sound enamel 1-8. Upon demineralization, the scattering coefficient increases by almost two orders of magnitude which provides high contrast for imaging in both transillumination and reflectance 9. Optical coherence tomography has also been most successful at 1310nm for imaging teeth due to the high transparency of enamel 10-12. Light scattering in enamel appears to be dominated by Rayleigh scattering and the magnitude of Rayleigh scattering is inversely proportional to the fourth power of the wavelength13. Therefore, the magitude of light scattering in enamel decreases markedly from the UV to the near-IR. Light scattering in dentin is primarily due to the dentinal tubules which behave as cylindrical Mie scatterers. The wavelength dependence is more complex and measurements from 500-1050-nm do not show the marked reduction is scattering that has been observed for enamel 13-16. The amount of light reflected (backscattered) from the underlying dentin depends on the ratio of scattering to absorption. Assuming absorption is solely due to water in the near-IR, we postulate that longer near-IR wavelengths where scattering decreases and absorption increases will yield the highest contrast in reflectance. Hyperspectral reflectance measurements of Zakian 8 show that the dentin gets darker with increasing wavelength even when water absorption is not increasing. Previous angularly resolved optical measurements of sound dentin carried out at 543, 632 and 1050-nm indicated that the measured scattering and absorption coefficients of dentin are both almost an order of magnitude larger than for enamel13. In contrast with enamel, the scattering and absorption coefficients did not change significantly with •
Contact Author: [email protected]
Lasers in Dentistry XX, edited by Peter Rechmann, Daniel Fried, Proc. of SPIE Vol. 8929, 89290M · © 2014 SPIE · CCC code: 1605-7422/14/$18 · doi: 10.1117/12.2045629
Proc. of SPIE Vol. 8929 89290M-1 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 03/14/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
wavelength, which is consistentt with measurrements usingg an integratiing sphere, values v did varry signifiicantly with th he method of measurement, μs = 250–2880 cm-1 for anngularly resolvved goniometeer measuurements and 130–180 cm-1 for f integrating sphere measurrements. Valuues of the scatteering anisotroppy (g) alsso vary consid derably, Fried et al. 13report a g value of 0.93 that waas determined using angularlly resolvved goniometer measuremennts while Zijpp and ten Bosch report values v of 0.400 calculated by b integraating the measu ured scatteringg distributions of o 20-μm-thickk dentin samplees 15, 16. In ordder to accurateely measure thhe internally optical o properties it is extrem mely importannt to reduce thhe influennce of surfacee scattering. In I previous stuudies of dentall enamel, sam mples were placced in an indeex matchhing bath of th he same refracttive index, nam mely carbon disulfide d or a cargille c liquid 1, 13. Howeveer, dentinn poses an add ditional challennge since it is a highly poroous structure composed c of tuubule structurees whichh are typically filled with fluid f and it is the scatteringg of the tubulles that dominnate the optical properrties. Thereforre, if dentin waas immersed inn an index matcching fluid thee internal scatteering would alsso be greeatly reduced and a would not represent r the optical o behavioor expected clinnically. We thherefore chose to t either place the denttin sections in water or mainttain a layer of water on the surface s to best represent the in situ opptical environm ment. This appproach was takken in a prior study s involvinng highly porouus caries lesionns and demineralized d enamel wheree index matchhing agents would likely im mbibe into thee lesion greatlly reduciing the internaal scattering inn the lesion9. We were unabble to measuree all the samplles immersed in i water due to the ex xcessive attenuuation at some of the near-IR R wavelengthss where there is strong water absorpption, e.g. 1450 0-nm, thereforee we measuredd these samples immediately after immersioon in water, i.ee., with a water layer. The T study desiign emphasizess measurementt of relative diffferences in optical attenuatioon at nearr-IR wavelengtths from 1300--1650-nm.
2. MATERIA ALS AND METHODS M
A
I I
E
F
B
Fig. 1. The experimenttal setup for meassurement of the opptical attenuaation through thin n sections of denttin placed in a cuvvette. (A) Tungsten–halogen T lamp with (B B) filter wheel, (C) collimaating and focusin ng lenses, optical chopper (D), saample cuvettee (E) and (F) detecctor are shown.
Plano--parallel sectioons ranging from f 50-µm to t 600-µm thick were cut c from twentty-five extracteed molarss/premolars annd serially poolished to a 0.3 µm diamond grit finnish. Samples were stored inn a 1% thhymol solution to prevent deccay. Light from m an Ocean Optics (Dunedin, Fl) fiber-coupleed tungstten-halogen laamp (Model HL-2000-FHSA H A) with a filter wheel with w bandpass filters f (FBxxxxx12-FW WHM 12-nm)) Thorlabs (Newton, NJ) N centerred at 50-nm intervals i from m 1300-1650-nm m was ussed to illuminaate specific areas of the sample with a spot size off approximatelyy 100 µm. Thhe spot siize was confirm med through knife-edge k beam m profiliing techniques. Phase senssitive detectioon was employed e usingg a lock-in am mplifier (Moddel SR 8550) and an opptical chopperr from Stanforrd Researrch Systems (Stanford, ( CA)). A large-areea Ge phhotoreceiver Model M 2033 from New Focuus (Santaa Clara, CA)) equipped with w a variable apertuure was used foor detection. The T setup for thhe
attenuuation measurem ments is shownn in Fig. 1. Samplles were measu ured either fullly immersed inn a cuvette (1-ccm sq.) filled with water (n= =1.3) or wet. In I order to t measure thee samples with a wet surface the t sample wass wetted beforee each measureement by raisinng the cuuvette filled wiith water to weet the surface. This proceduure ensured that the positionn of the incidennt collim mated light beaam was fixed while w cycling through the filters. fi Fivee measurementts of collimateed transm mission were reecorded at varioous sections within w each sam mple. The collim mated signal (II) at the detectoor was coompared to thee initial intensitty of the beam m (Io). Using thhis ratio and thee thickness of the t samples (T T), the atttenuation coeffficient (μt) wass calculated usiing Beer's law and Beer-Lamb mbert plots:
Proc. of SPIE Vol. 8929 89290M-2 Downloaded From: http://proceedings.spiedigitallibrary.org/ on 03/14/2016 Terms of Use: http://spiedigitallibrary.org/ss/TermsOfUse.aspx
µt = - ln(I//Io)/ T
(11) Beeer's law statess that the inttensity of lighht passsing throughh a sampple decreasees expoonentially as thickness inncreases. Thhis overrall decrease is known as attenuation or o extinction, which is caused by suurface scatterinng s and diffuse d reflectiion at front annd or specular rearr surfaces, and a internal scattering annd absoorption. T The aperture was w sufficientlly smaall in front of thhe detector to ensure that lighht scatttered at smalll angles did not contribuute signnificantly to thee collimated traansmission.
.
3. RESULTS S
0.04
Tlìickì IC:m) Fig. 2. Beer-Lambert plot at 1300 nm forr dentin with a th hin layer of watter applied to the surface. Colliimated transmisssion for varying g sample thicknesss ranging from 50 5 to 600-μm are plotted. Each point p represents the t mean attenuaation of each of the five measurrements on each saample.
Collimated Transmission (In I/lo)
Figuures 2 and 3 contain c Beer-L Lambert plots at 1300-nm for sam mples with a water w layer annd fullyy immersed inn water. Thee values for thhe attennuation coefficcients were 23 ± 4.8 and 35 ± Correlation coefficients 10 respectively. r c (R R2) werre 0.49 and 0.34 and botth slopes werre nt from zero. The T slopes for all the Beer-Laambert plots foor the samples with the applieed signifiicantly differen water layer were sig gnificantly diffferent from zerro. This indicates that surface scattering does d not prevennt determ mination of th he internal scaattering and absorption a even though it is approximateely an order of o magniitude higher. Surface scatterinng is relativelyy independent of sample thicckness, therefore if attenuatioon is com mpletely domin nated by surfacce scattering, than t there wouuld be only a minimal m changge in attenuatioon with increasing i sam mple thickness.. This has been observed o for prrevious measurrements for enameel in the near-IR where thhe internal scatterring was extrem mely low and attenuation was dominated d by surface scatteering if the samplees were nott placed in an index matchhing fluid 1. Only the sloope for the 1300-nnm measurem ments was siignificantly differeent from zero for the denttin samples fully immersed i in water. w This wass due to the extrem mely low signaal intensity cauused by the absorpption of water in the cuvettess. Figure 4 contains a plot of th he attenuation coefficients c for eaach wavelengtth along with a plot of 0.04 water absorption taaken from thee paper of Ihlclkness (cm) 17 Hale and a Querry . The attenuaation values Fig. 3. Beer-L Lambert plot at 1300 nm for denttin immersed in lie beetween 20 and d 40 cm-1 annd increase water. Collimatted transmission for varying saample thickness ranging from 500 to 600-μm are plotted. p Each point represents the signifiicantly with increasing wavelength. w mean attenuation n of each of the fivve measurements on o each sample. Comparison of the mean attenuattion values with ANOVA ind dicates that there t is a signifiicant linear inccrease in attenuuation with increaasing waveleng gth (P
