Eur J Oral Sci 2005; 113: 494 498 Printed in Singapore. All rights reserved Copyright Ó Eur J Oral Sci 2005 European Journal of Oral Sciences Detection of initial caries lesions on smooth surfaces by quantitative light-induced fluorescence and visual examination: an in vivo comparison Roswitha Heinrich-Weltzien 1, Jan Kühnisch 2, Susanne Ifland 1, Sofia Tranæus 3, Birgit Angmar-Månsson 3, Lutz Stçßer 1 1 Department of Preventive Dentistry, University of Jena, Jena, Germany; 2 Department of Conservative Dentistry and Periodontology, Ludwig-Maximilians-University of Munich, Munich, Germany; 3 Department of Cariology and Endodontology, Institute of Odontology, Karolinska Institutet, Huddinge, Sweden Heinrich-Weltzien R, Ku hnisch J, Ifland S, Tranæus S, Angmar-Ma nsson B, Sto ßer L. Detection of initial caries lesions on smooth surfaces by quantitative light-induced fluorescence and visual examination: an in vivo comparison. Eur J Oral Sci 2005; 113: 494 498. Ó Eur J Oral Sci, 2005 The aim of this clinical study was to compare the outcome of quantitative laser/lightinduced fluorescence () and visual inspection (VI) for the detection of initial caries lesions on all maxillary and mandibular smooth surfaces in caries-risk adolescents. The subjects were 34 students, mean age 15 yr. A total of 879 buccal and 882 lingual surfaces were air-dried and visually examined at a magnification of 3.5. Fluorescence images of each smooth surface were captured with \clin equipment, and software 2.00 was used to display, store, and analyse the images. Fluorescence loss (DF; %) and area of the lesion (A; mm 2 ), and fluorescence loss integrated over the lesion area (DQ; DF A; % mm 2 ), were determined. The presence or absence of initial caries lesions was scored using both VI and. A total of 87.2% of all smooth surfaces were scored as sound or initially carious when assessed by VI+ in combination: 4.9% were detected by VI alone and 7.9% by alone. The parameters DF, A, and DQ differed significantly between lesions registered with VI+ and alone. It was concluded that (i) seems to be a sensitive method that is suitable for the detection of visually undetected initial caries lesions; and (ii) that the clinical use of is limited by several confounding factors in cariesrisk adolescents. Prof. Dr R. Heinrich-Weltzien, Department of Preventive Dentistry, University of Jena, An der Alten Post 4, D-07740 Jena, Germany Telefax: +49 3641 934573 E-mail: roswitha.heinrich-weltzien@med. uni-jena.de Key words: caries-risk adolescents; clinical study; initial caries lesion; ; visual inspection Accepted for publication August 2005 Visual inspection (VI) is a non-invasive method of caries detection on accessible smooth surfaces. For clinical examination, VI is simple, quick, and cost-effective. However, the method has considerable limitations. Lesions are usually detectable first at the white spot stage, but may be more advanced; mineral loss in enamel lesions is not quantifiable, and changes in mineral content loss or gain as a result of de- or remineralization cannot be monitored. It is suggested that objective methods would enhance conventional intraoral examination and improve the potential for correct diagnosis, especially in caries-risk patients (1). The quantitative laser/light-induced fluorescence method () is reported to be valuable tool for the detection and quantification of initial caries lesions and for monitoring de- or remineralization on smooth surfaces (2, 3). The first in vivo applications of (4, 5) were followed by clinical trials (1, 6 9). These studies disclosed an excellent repeatability and reproducibility for permanent teeth. The histological validation of clinically detected smooth surface lesions with performed on deciduous teeth revealed a sensitivity of 79% and a specificity of 75% (10). Moreover, histological studies verified the fluorescence loss of initial caries lesions measured by as an excellent parameter for the expression of the mineral loss of the lesion. The amount of fluorescence radiance validated with the use of transversal and longitudinal microradiography is closely correlated (r 2 ¼ 0.97) to the mineral loss in artificial and natural initial caries lesions (11 13). To date there are only limited data collected in the Indiana study comparing the method with visual and visual-tactile examination for the detection of caries lesions on smooth surfaces (8). The method of comparison in that study was based on a split-mouth design for clinical examination and assessment of buccal and lingual surfaces in maxillary teeth and of buccal surfaces of mandibular teeth (8). As the presence or absence of a lesion was determined by placing a marker on the live image indicating the lesion, the Indiana study design did not focus on an independent comparison of both methods. The aim of the present clinical study was therefore to assess the outcome of initial caries lesion detection on all maxillary and mandibular smooth surfaces by and meticulous VI, with the hypothesis that a meticulous VI would show a similar performance
Visual inspection vs. 495 as. Furthermore, the potential of for clinical use in high caries-risk subjects should be evaluated. Material and methods Study subjects The study comprised 34 healthy subjects aged 14 15 yr. All subjects were participants in a longitudinal caries riskassessment study (1993 99), and were accustomed to annual clinical, microbiological, and biochemical examinations (14). At baseline, 8-yr-old-children (n ¼ 189), attending the second grade, were randomly selected from all schools of a non-fluoridated city (Erfurt). At a later follow-up examination, the remaining participants (n ¼ 99) were invited to take part in an additional investigation of their teeth, followed by a preventive programme, including professional tooth cleaning, 3-monthly application of a fluoride varnish, and oral health instructions. The subjects dental care was provided by general dental practitioners. Informed proxy consent was obtained from the subjects and their parents. The study was approved by the Ethics Committee of the Friedrich-Schiller-University of Jena. Clinical examination Prior to clinical examination and professional tooth cleaning, the subjects plaque scores were estimated using the proximal plaque index (PPI) (15) and gingival status was estimated using the papillary bleeding index (PBI) (16). The clinical investigation was performed by a dentist (R.H.-W.) with considerable experience in clinical assessment of caries lesions. World Health Organization criteria (17) were used for classification of the tooth surface status. According to the caries diagnostic system proposed by Nyvad et al. (18), white spots on smooth surfaces and discoloured fissures were recorded separately as initial caries lesions, but were not included in the decayed, missing or filled tooth surfaces (DMFS) index. All examinations were conducted under standardized conditions using a dental unit equipped with compressed air and evacuation facilities. After prolonged air-drying (> 5 s) of the tooth surfaces, caries were recorded with the aid of a dental magnifying glass ( 3.5). The buccal and lingual surfaces of all permanent teeth were classified according to one of the following criteria: 0, sound (normal enamel translucency and texture); 1, initial caries lesion (surface of the enamel is whitish, yellowish or brownish opaque with/without normal enamel translucency); 2, localized surface defect (enamel/dentin cavity); 3, filled surface with intact filling; 4, filled with marginal/ recurrent decay, filled surface with intact filling and a separate lesion. Cavitated and filled surfaces were routinely excluded from the methodological comparison (39 buccal and 35 lingual surfaces). The additional presence or absence of developmental disorders (fluorosis, mineralization disturbances) was noted. Because of the similar appearance of the hypoplastic/fluorotic enamel and initial caries lesions in the fluorescence images, the detection of any enamel developmental disorder was the decisive criterion for exclusion of the surface from further image analysis. If no clear differentiation could be made, the surface was excluded (n ¼ 1). Intra-examiner reliability of VI was assessed by a second examination of 10 randomly selected subjects 1 wk after their first clinical examination. measurements A portable device (\clin) was used, equipped with a xenon microdischarge arc lamp as the light source and an optical filter system producing blue light with a maximum wavelength of 370 nm, conducted by a liquid filled guide (Inspektor Research Systems, Amsterdam, the Netherlands). The re-emitted fluorescence was collected with a micro-ccd-video camera (Panasonic WV-KS 152; Matsushita Electric Industrial Co., Ltd, Osaka, Japan) equipped with a yellow high-pass filter (k > 520 nm) to exclude any excitation or ambient light from reaching the detector. To avoid interference to the fluorescence images by bubbles of saliva or blood from gingival bleeding, drying of the teeth was prolonged (> 5 s) before the images were captured in the darkened dental office. The image capturing was performed on buccal and lingual surfaces of all permanent maxillary and mandibular teeth by a trained examiner (R.H.-W.) 2 wk after the clinical examination. No information about the visual scoring of presence/absence of an initial caries lesion was noted on the fluorescence image. 2.00f software (Inspektor Research Systems) was used to display, store and analyse the images (19). Initially, the images were visually viewed for signs of decalcification, which appear as dark areas surrounded by bright green fluorescent sound tooth tissue (2, 4, 19). If a lesion was detected, the fluorescence loss (DF; %), lesion area (A; mm 2 ) and the fluorescence loss integrated over the lesion area (DQ; % mm 2 ) were analysed using the systems analysis software to determine lesion severity. The image analysis was considered to be optimal when the grey-level image showed a homogeneous status (2, 4, 19). Mean DQ values for the )5% threshold were exported for subsequent statistical analysis. All images were analysed in a blinded manner by two independent investigators (J.K. and S.I.). The reproducibility of the image analysis was assessed by calculation of the intra- and interexaminer reliability. After an intensive calibration training of 1 wk for the analysts (J.K., S.I), 60 smooth surfaces with various appearance of initial caries lesions were randomly selected; a second image analysis was performed 3 d later by both analysts. Statistics For both VI and methods of examination, the smooth surfaces were categorized as sound or with initial caries. Initial lesions were grouped, according to the method of examination, into the following: VI+, VI alone, or alone. The results were calculated separately for buccal and lingual surfaces. For the -related groups, descriptive statistics included mean values and standard deviation (SD) with confidence intervals (95% CI) for the parameters A, DF, and DQ. The Mann Whitney U-test was used to determine differences between the quantitative parameter of smooth surface lesions detected by VI+ and by alone. A P-value of less than 0.001 was considered as statistically significant. The correlation between VI and was analysed by calculating the Spearman rank correlation coefficient (r s ) (20). The quality of intraexaminer reliability for VI was calculated using Cohen s kappa statistic. A kappa (j) value of 1 represents perfect agreement, j-values of < 1, but > 0.75, denote excellent agreement, those in the range of 0.4 0.75 denote good agreement, and values of < 0.4 denote marginal agreement (21). For determining the
496 Heinrich-Weltzien et al. intra- and interexaminer reliability of image analysis, the intraclass correlation coefficient was estimated for both analysts (20). Results The study comprised 34, 15-yr-old students with a caries experience of 7.7 ± 5.8 D 3)4 MFS. The PPI of 71.1% reflected poor oral hygiene, associated with a generalized gingivitis, scored as a PBI of 76.2%. A j-value of 0.84 was calculated for the intraexaminer agreement of VI. A high intra- and interexaminer agreement was also determined for the three parameters of analysis (Table 1). The relationship between VI and revealed a Spearman rank correlation coefficient of r s ¼ 0.63. A total of 1,761 smooth surfaces were examined by VI and (Table 2) The two methods were in agreement with respect to 87.2% (n ¼ 1535) of the surfaces investigated: 72.2% (n ¼ 1,272) were assessed as sound and 15.0% (n ¼ 263) as initial carious. The proportion of initial caries lesions detected by VI alone represented 4.9% (n ¼ 87) of the surfaces and increased to 7.9% (n ¼ 139) with alone. The distribution of the latter was fairly even: 4.3% (n ¼ 76) on lingual surfaces and 3.6% (n ¼ 63) on buccal surfaces. The features of fluorescence images of smooth surfaces with visually scored initial lesions, not detected with (4.9%; n ¼ 87), revealed that a reduced surface size associated with inflamed gingival swelling (50.6%) and a Table 1 Interclass correlation coefficients of intra- and interexaminer agreement for the quantitative laser/light-induced fluorescence () image analysis Investigator (J.K.) Intraexaminer Investigator (S.I.) Interexaminer A (mm 2 ) 0.90 0.96 0.92 DF (%) 0.95 0.99 0.88 DQ (% mm 2 ) 0.81 0.95 0.85 A, lesion area; DF, fluorescence loss; DQ, fluorescence loss integrated over the lesion area (DF A; % mm 2 ). Table 2 Detection of initial caries lesions on smooth surfaces by visual inspection (VI) and quantitative laser/light-induced fluorescence () Surface Sound Initial caries lesion Total VI+ VI+ VI alone alone n % n % n % n % n % Buccal 527 29.9 213 12.1 76 4.3 63 3.6 879 49.9 Lingual 745 42.3 50 2.8 11 0.6 76 4.3 882 50.1 Total 1,272 72.2 263 15.0 87 4.9 139 7.9 1,761 100.0 Table 3 Features of fluorescence images of smooth surfaces with visually scored initial caries lesions, not detected with quantitative laser/ light-induced fluorescence () Feature of images partially plaque covered surface (23.0%) were the main reasons that a image analysis could not be performed properly (Table 3). Table 4 presents the quantitative parameters of the analysis of 263 initial lesions detected by both methods, compared to the 139 lesions that were detected with alone. There were significantly lower values with respect to A, DF and DQ of lesions detected with alone. Discussion Initial caries lesion detected with VI only n % Surface associated with an 44 50.6 inflamed swollen gingival margin, which reduced the surface size Surface is (partially) covered with plaque 20 23.0 Very bright image, no fluorescence 9 10.3 loss detectable Image out of focus, no analysis possible 14 16.1 Total 87 100.0 VI, visual inspection. The clinical management of dental caries has progressed from extraction to restoration and recently from minimally invasive to maximally interceptive approaches. With the shift from operative to non-operative intervention (22), the accurate early detection of caries lesions is a prerequisite for interceptive measures to arrest or reverse caries progression. In this context, seems to be a promising method, which fulfills the essential requirements for detection, quantification and monitoring of caries lesions (1 3, 8, 9, 23, 24). The present study was conducted under standardized clinical conditions and compared with meticulous VI for detection of smooth surface initial caries lesions. The subjects were 34, 15-yr-old students, designated as high-risk patients on the basis of their caries experience of 7.7 D 3)4 MFS as well as high plaque and gingival bleeding scores. The data are based on the examination of 1,761 smooth surfaces by VI and. One finding, in this study, was that was able to detect a greater number of initial caries lesions than meticulous VI. This finding might be an effect of the study design, which was not restricted to visually detectable lesions but performed as a registration of all smooth surfaces with VI and. Thus, it was possible to compare the outcome of lesion detection by the two methods, in contrast to the Indiana study (8). Of all initial caries lesions, 263 (53.8%) were detected by both methods. Of the total number of initial caries
Visual inspection vs. 497 Table 4 Image analysis of initial caries lesions on smooth surfaces detected with visual inspection + quantitative laser/light-induced fluorescence (VI+) and with alone Initial caries lesion Parameter Mean SD )95% CI +95% CI Detected with VI+ (n ¼ 263) A (mm 2 ) 5.8* 5.4 5.1 6.5 DF (%) )9.6* 8.0 )10.6 )8.6 DQ (% mm 2 ) )84.6 148.4 )103.0 )67.0 Detected with alone (n ¼ 139) A (mm 2 ) 3.4* 3.7 2.7 4.7 DF (%) )7.1* 6.6 )8.2 )6.0 DQ (% mm 2 ) )38.8 94.5 )55.0 )22.0 A, lesion area; CI, confidence interval; DF, fluorescence loss; DQ, fluorescence loss integrated over the lesion area (DF A; % mm 2 ); SD, standard deviation. *P ¼ 0.001. lesions registered on the buccal and lingual surfaces, 139 (28.4%) were registered by alone, compared with 87 (17.8%) registered by VI only. As shown in Table 2, 55.5% of 137 initial caries lesions on lingual surfaces were detected by alone (n ¼ 76). In contrast, only 17.9% of lesions on the buccal surfaces were detected by alone (63 of a total of 352 buccal lesions), whereas VI alone detected 76 buccal lesions (21.6%). This discrepancy between detection of lesions on buccal and lingual surfaces might be attributable to difficulty inspecting the lingual surfaces with VI. The results suggest that enables more sensitive judgement than VI for the detection of smooth surface lesions. However, neither method could be validated against any histological reference standard in this study. As a consequence, it could not be excluded that sites which were qualified as carious by or VI alone were really sound (false-positive diagnosis). Considering the results of former histologically validated studies comparing VI and (9, 10, 13), the authors assume that is superior to VI. A dry and clean surface is essential for capturing an optimal image as well as for meticulous VI. However, it was established that it is more difficult to ensure no saliva contamination during image capturing, especially at lingual surfaces, owing to a longer examination time. This is in line with other in vitro studies (25, 26), which determined the drying effect as a sensitive parameter that has to be controlled for lesion detection and quantification. To avoid any detection failure of initial caries lesions, as occurred in this study, effective air-drying of each surface (> 5 s) is of considerable importance for both methods, and should be controlled strictly. With respect to lesions detected by VI alone, the features of the fluorescence appearance of these smooth surfaces reveal several problems (Table 3). In cases of inflamed and swollen gingiva (n ¼ 44), lesion detection was as impossible as in cases with a tooth surface partially covered by plaque (n ¼ 20), which may be invisible in daylight but clearly visible on the computer screen. In our experience, the use of in children or adolescents with a high risk of caries is practical only when the gingiva is not inflamed and initial lesions located at the gingival margin are unobscured and accessible for detection and monitoring. These findings highlight the fact that lesion detection and quantification are limited when confounding factors are not controlled. As tooth surfaces or dental lesions may be covered with dental plaque, calculus and/or extrinsic discolorations, meticulous professional tooth cleaning is essential prior to the use of the method for early caries detection. In 23 cases, inadequate images may have contributed to failure of lesion detection by : very bright images without any chance of determining fluorescence loss (n ¼ 9), and those that were captured out of focus or not perpendicular to the tooth surface (n ¼ 14) (Table 3). Quantitative analysis of the data from images of noncavitated lesions detected by alone disclosed that these lesions had a significantly smaller area and a lower fluorescence loss than lesions detected by VI+ (Table 4). The clinical relevance and consequence of this finding should be considered critically. On the one hand, it seems to be desirable to detect and quantify any lesion as early as possible, in order to optimize the outcome of interceptive measures. On the other hand, the time-consuming investigation of all smooth surfaces for clinically invisible lesions must be questioned. Meticulous VI is the method of choice for routine examination. Visual inspection focusing on the detection of initial lesions enables the dentist to assess the activity of the lesion (18) and provides further information about future caries risk and the appropriate intervention strategy (27 29). Quantitative laser/light-induced fluorescence analysis of inactive lesions that appear visually smooth and glossy without loss of surface integrity (commonly localized at some distance from the gingival margin) does not seem to be practical. In contrast to these easily assessable lesions, active lesions, usually located at the gingival margin, are difficult to capture and analyse with. There is a need for more information about the suitability of for the assessment of active, extensive lesions at the gingival margin commonly found in high-risk patients. For the dentist using this method it would be helpful to have guidelines with respect to suitable cut-off values for the parameters fluorescence loss and/or area, guiding the clinician s decision-making towards preventive or operative intervention. The dark appearance of the lesions, as viewed with, is very obvious and might therefore
498 Heinrich-Weltzien et al. serve to motivate the patient towards more effective oral hygiene (23, 24). In summary, it was concluded that our hypothesis, assuming a similar detection outcome of a meticulous VI and, should be rejected. The data seem to support the thesis that is a sensitive method suitable for the detection of visually undetected initial caries lesions, and that the clinical use of is limited by several confounding factors in caries-risk adolescents. References 1. TranÆus S, Al-Khateeb S, Björkman S, Tweetman S, Angmar-Månsson B. Application of quantitative light-induced fluorescence to monitor incipient lesions in caries-active children. A comparative study of remineralisation by fluoride varnish and professional cleaning. Eur J Oral Sci 2001; 109: 71 75. 2. Angmar-Månsson B, Ten Bosch JJ. Quantitative light-induced fluorescence (): a method for assessment of incipient caries lesions. Dento-Maxillo-Facial Radiol 2001; 30: 298 307. 3. Kühnisch J, Heinrich-Weltzien R. Quantitative light-induced florescence () a literature overview. Int J Comp Dent 2004; 7: 325 338. 4. de Josselin de Jong E, Sundström F, Westerling H, Tran- Æus S, Ten Bosch JJ, Angmar-Månsson B. A new method for in vivo quantification of changes in initial enamel caries with laser fluorescence. Caries Res 1995; 29: 2 7. 5. Al-Khateeb S, Oliveby A, de Josselin de Jong E, Angmar- Månsson B. Laser fluorescence quantification of remineralisation in situ of incipient enamel lesions: influence of fluoride supplements. Caries Res 1997; 31: 132 140. 6. Al-Khateeb S, Forsberg CM, de Josselin de Jong E, Angmar-Månsson B. A longitudinal laser fluorescence study of white spot lesions in orthodontic patients. Am J Orthod Dentofacial Orthoped 1998; 113: 595 602. 7. Tranaeus SXQ, Shi Lindgren LE, Trollsås K, Angmar- Månsson B. In vivo repeatability and reproducibility of the quantitative light-induced flurescence method. Caries Res 2002; 36: 3 9. 8. Ferreira Zandoná AG, Isaacs RL, Van Der Veen MH, Stookey GK. Indiana pilot clinical study of quantitative light fluorescence. In: Stookey GK, ed. Early Detection of Dental Caries II. Proceedings of the 4th Annual Indiana Conference. Indiana, IN: Indiana University School of Dentistry, 2000; 219 230. 9. Ferreira Zandoná AG, Stookey GK, Eggertsson H, Wefel J, Katz B, Ofner S, Eckert G. Clinical validation study of at Indiana. In: Stookey GK, ed. Early Detection of Dental Caries III. Proceedings of the 6th Annual Indiana Conference. Indiana, IN: Indiana University School of Dentistry, 2003; 237 252. 10. Ten Cate BJM, Lagerweij MD, Wefel JS, Angmar-Månsson B, Hall AF, Ferreira Zandoná AG, Stookey GK, Faller RV. In vitro validation studies of quantitative lightinduced fluorescence. In: Stookey GK, ed. Early Detection of Dental Caries II. 2000. Proceedings of the 4th Annual Indiana Conference. Indiana, IN: Indiana University School of Dentistry, 2000; 231 250. 11. Al-Khateeb S, Ten Cate JM, Angmar-Månsson B, de Josselin de Jong E, Sundstroem G, Exterate RA, Oliveby A. Quantification of formation and remineralization of artificial enamel lesions with a new portable fluorescence device. Adv Dent Res 1997; 11: 502 506. 12. Emami Z, de Josselin de Jong E, Sundström S, Trollsås K, Angmar-Månsson B. Mineral loss in incipient caries lesions quantified with laser fluorescence and longitudinal microradiography: a methodological study. Acta Odontol Scand 1996; 54: 8 13. 13. Ando M, Hall AF, Eckert GJ, Schemenhorn BR, Analoui M, Stookey GK. Relative ability of laser fluorescence techniques to quantitate early mineral loss in vitro. Caries Res 1997; 31: 125 131. 14. Stößer L, Tietze W, Heinrich-Weltzien R, Kneist S, Schumann V, Möller M, Busse H. Studiendesign und repräsentativität der erfurter kariesrisiko-studie mit schu lern der ersten und fu nften klasse. In: Stößer L, ed. Kariesdynamik und kariesrisiko. Berlin: Quintessenz, 1998; 168 178. 15. Lange DE, Plagmann HC, Eenboom A, Promesberger A. Bewertungsverfahren zur objektivierung der mundhygiene. Dtsch Zahna rztl Z 1977; 32: 44 47. 16. Saxer UP, Mühlemann HR. Motivation und aufklärung. Schweiz Monatsschr Zahnmed 1975; 85: 905 919. 17. WHO. Basic Methods, 4th edn. Geneva: World Health Organization, 1997. 18. Nyvad B, Machiulskiene V, Baelum V. Reliability of a new caries diagnostic system differentiating between active and inactive caries lesions. Caries Res 1999; 33: 252 260. 19. Van Der Veen MH, de Josselin de Jong E. Application of quantitative light-induced fluorescence for assessing early caries lesions. Monogr Oral Sci 2000; 17: 144 162. 20. Bühl A, Zöfel P. SPSS, Version 10. Einfu hrung in die Moderne Datenanalyse Unter Windows. Mu nchen: Addison-Wesley, 2000; 242. 21. Fleiss IL. Statistical Methods for Rates and Proportions, 2nd edn. New York: Wiley, 1981; 212 225. 22. Pitts NB. Are we ready to move from operative to nonoperative/preventive treatment of dental caries in clinical practice? Caries Res 2004; 38: 294 304. 23. Heinrich-Weltzien R, Kühnisch J, Van Der Veen M, de Josselin de Jong E, Stößer L. Quantitative light-induced lluorescence () a potential method for the dental practitioner. Quintessence Int 2003; 34: 181 188. 24. TranÆus S, Heinrich-Weltzien R, Kühnisch J, Stösser L, Angmar-Månsson B. Potential applications and limitations of quantitative light-induced fluorescence in dentistry. Med Laser Appl 2001; 16: 195 204. 25. Pretty IA, Edgar WM, Higham SM. The effect of dehydration on quantitative light-induced fluorescence analysis of early enamel demineralization. J Oral Rehabil 2004; 31: 179 184. 26. Al-Khateeb S, Exterkate RA, de Josselin de Jong E, Angmar-Mansson B, Ten Cate JM. Light-induced fluorescence studies on dehydration of incipient enamel lesions. Caries Res 2002; 36: 25 30. 27. Alanen P, Hurskainen K, Isokangas P, Pietilä I, Levänen J, Saarni U-M, Tiekso J. Clinican s ability to identify caries risk subjects. Community Dent Oral Epidemiol 1994; 22: 86. 28. Graves RC, Abernathy JR, Disney JA, Stamm JW, Bohannan HM. University of North Carolina caries risk assessment study. III. Multiple factors in caries prevalence. J Public Health Dent 1991; 51: 134 143. 29. Helfenstein U, Steiner M, Marthaler TM. Caries prediction on the basis of past caries including precavity lesions. Caries Res 1991; 25: 372 376.