Eur J Orthod 2003 Benson 149 58

7/27/2019 Eur J Orthod 2003 Benson 149 58

1/10

European Journal of Orthodontics 25 (2003) 149-158 2003 European O rthodontic Society

Quanti fy ing enamel demineral izat ion from teeth withorthodontic bracketsa comparison of two methods.Part 1: repeatability and agreementPhilip E. Benson *, Nei l Pender** and Susan M. H igh am ***Department of Child Dental Health, University of Sheffield and **Cariology Research Group, Department ofClinical Dental Sciences, University of Liverpool, UKSUMMARY The aim of this investigation was to compare the repeatability of m easuring enamel dem ineral-ization surrounding an orthodontic bracket using two techniques: computerized image analysis fromdigitally converted photographic slides and quantitative light-induced fluorescence (QLF). Fifteenhuman molars were halved and shaped to look like incisors. The teeth were individually num bered andorthodontic brackets bonded to the buccal surface. The crowns were covered w ith acid resistant varn ish,except for windows approximately 1.5 x 3 mm adjacent to the gingival, occlusal, mesial, and distaledges of the bracket. The windo ws were va riously exposed to a dem ineralizing gel for 0, 3, 7, or 14 days,and the acid resistant varnish was removed. Standardized photographic slides and QLF images of theteeth were taken. These were repeated after 1 week. The slides were converted to grey scale digitalforma t and analysed using Image-Pro Plus 3.0. The QLF images were stored , processed, and analysedusing customized software. All images were recoded for b lind analysis. The four surfaces of the bracketwere inspected and only areas of suspected demineralization were analysed. This was repeated after1 week.The limits of agreement and mean difference between repeat readings of the area of dem ineralizationwere similar for both techniques (-0.04 0.43 for photographs and -0.10 + 0.63 for QLF). Mean greylevel (photographs) and mean loss of fluorescence from that area (AF) (QLF) showed acceptable limitsof agreement. The Intra Class Correlation (ICC) was below 0.81 for the measurement of area from QLF,suggesting that random error needs to be reduced. There was evidence of systematic bias for the repeatreadings of the grey levels from the photographs (P< 0.001). Enamel demineralization surrounding anorthodontic bracket can be measured reproducibly using these two techniques.

IntroductionDemineralization of enamel surrounding orthodonticbrackets is a significant clinical problem. Iatrogenicwhite spot lesions lead to poor aesthetics and, in severecases, the need for restorative treatment. Strategies toprevent demineralization subsequent to orthodontictreatment must be developed. To test the effectivenessof these preventive regimes, a technique for the record-ing and measurement of enamel white spots should beused that is both reproducible and valid (Houston,1983).The severity of enamel demineralization can bequantified in terms of the area of the tooth surface thatit covers and the degree of mineral loss. A recording andmeasuring technique for demineralization should showgood repeatability for both these parameters and itshould be clear that the technique is measuring enameldefects that are associated with the orthodonticappliance and not one of the many other causes of whitespot lesions (Small and Murray, 1978).

Photographs are commonly used in the clinical envir-onment. They are a convenient and effective means of

permanently recording the optical properties of enamel.It has been shown that an area of demineralization onthe buccal surface of the tooth can be reproduciblymeasured using photographic slides that have beenconverted to digital images and measured with com-puterized image analysis (Benson et al, 2000; Willmotet al, 2000).Various optical metho ds have also been developed toquantify enamel demineralization. These are wellreviewed by Angmar-Mansson et al. (1996). They can becategorized into non-fluorescent methods, such as theoptical caries monitor (ten Bosch et al, 1980), andfluorescent methods. The fluorescent methods havepreviously involved the use of ultraviolet or laser light,which are potentially dangerous forms of radiation,particularly to th e eyes.Recently, a small portable system with a new lightsource and filter system has been described for intra-oral use (Al-Khateeb et al, 1997). It is calledquantitative light-induced fluorescence or QLF. Light

from an arc lamp passes through a blue filter, with apeak intensity of 370 nm, along a liquid light guide to a

byguestonAugust16,2013

http://ejo.oxfordjournals.o

rg/

Downloadedfrom
http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/http://ejo.oxfordjournals.org/


2/10

15 0 P. E. BENSON ET AL.handpiece that can be directed at the tooth surface.Enamel fluorescence is detected using an intra-oralcamera within the handpiece. The reflected light passesthrough a yellow high-pass filter of 520 nm in front ofthe camera to exclude light below that frequency. Thecombination is optimized to minimize reflections. Theimages are stored, processed, and analysed withcustomized software (Inspektor Research Systems BV,Amsterdam, The N etherlands). Q LF provides quantita-tive data for area of demineralization and mean loss offluorescen ce. The latter has been correlated withmineral loss measured using the destructive techniquesof transverse microradiography and chemical analysis(Al-Khateeb et al, 1997).The aim of this study was to compare the quantifi-cation of demineralization surrounding an orthodonticbracket using two methods.1. Computerized image analysis from a digitallyconverted photograph.2. Quantitative light-induced fluorescence.

The experiment is described in two parts. Therepeatability and agreement of the two techniques arecompared in this article, and in Part 2 the validity isassessed.Materials and methodsFifteen extracted human molars were used in thisin vitro study. It would have been preferable to useincisors, but it was not possible to obtain a large enoughsample. The teeth were carefully inspected to ensure theclinical absence of white spot lesions. They were dividedin half by cutting mesio-distally down the long axis ofthe tooth with a diamond wheel (Isomet; Buehler Ltd,Evanston, IL, USA). This produced a buccal and alingual half for each tooth, which were shaped to looklike incisors by using the diamond wheel to contour themesial, distal, and occlusal surfaces. A unique identifyingnumber was engraved on the cut surface of each halftooth. The cemento-enamel junction of each tooth wasgrooved with a small round burr and filled with light-cured composite resin dyed with a red vegetable dye tohighlight the junction and simulate the gingival margin.Identical standard edgewise twin brackets, with a slotsize of 0.018 x 0.025 inch [Ortho-Care (UK) Ltd,Bradford, UK], were bonded to the surface, in the usualposition for an orthodontic attachment. The crowns ofthe teeth were then covered with three coats of acidresistant varnish (Max Factor; Procter and Gamble,Weybridge, UK), except for windows approximately1.5 x 3 mm of enamel surface on the gingival, occlusal,left, and right aspects of the bracket (Figure 1). Theteeth were attached to glass rods and placed in ademineralizing gel (lactic acid, buffered with sodiumhydroxide to a pH of 4.5, in hydroxyethylcellulose).

Distance across thetie-wing used tocalibrate the image

Figure 1 Diagram showing a tooth with an orthodontic bracket andthe enamel area on each of the four sides of the bracket that wasexposed to the demineralizing gel.During the time in the gel, the windows on the enamelsurface of the teeth were incrementally covered ina systematic arrangement to subject the exposedenamel surface to demineralization for 0,3,7 , or 14 days(Figure 2). The patterns were chosen to ensure thatsome teeth had no demineralization challenge, some hadsevere demineralization, and there was a spread ofpatterns in between. The occlusal edge of the brackethas been identified as a site without a high prevalencefor demineralization (Mizrahi, 1982,1983), therefore, it

Tooth 14

Tooth 15

Tooth 16

Tooth 18

Tooth 19

Tooth 20

Tooth 21

1 - 5

Tooth 6

Tooth 7

Tooth 8

Tooth 9

Tooth 10

Tooth 11

Tooth 12 \ 7 'o

Tooth 13

Figure 2 Diagram showing the periods of incremental demineral-ization in days for the gingival, occlusal, left, and right surfaces ofthe bracket for the 30 teeth.

Tooth 23

Tooth 24

Tooth 25

Tooth 26

Tooth 27

Tooth 28

Tooth 29

Tooth 30 \ 3


3/10

MEASURING DEMINERALIZATION AROUND BRACKETS 1 5 1was not given priority when arranging the putativepatterns of demineralization.A tooth that was designated to have areas withdifferent periods of demineralization was removed fromthe gel after the shortest exposure time, washed indistilled water, and three coats of acid resistant varnishapplied to cover the relevant window. Once the varnishhad dried, the tooth was replaced in the gel. This wasrepeated until the maximum exposure for that toothhad been completed (between 3 and 14 days). Followingexposure to the gel, the teeth were washed in distilledwater and the varnish removed with acetone.Photographic technique an d image analysisStandardized photographs were taken of the teeth,using a technique that has been described previously(Benson et al., 1998). The ap erture was opened to F16,as a slightly lighter image was found to be beneficialwhen converting the image to a digital format andperforming the computerized image analysis.Photographs were taken with masking on the ringflash below the lens to reduce reflections (Benson et al.,2000; Willmot et al., 2000). To improve the replication ofthe camera positioning, a sighting jig was placed in thebracket slot (Figure 3). The jig consisted of a full sized(0.018 x 0.025 inch) rectangular stainless steel archwire,with one long and one short arm . The jig was held in thebracket slot with an elastomeric ligature. The camerawas lined up at right angles to the bracket using therectilinear attitude of the jig. The end of the long arm ofthe jig was identified in the viewfinder of the camera.The camera was then moved toward the tooth until theend of the short arm was in view. When the ends of thelong and short arms were adjacent in the horizontalplane, the photograph was taken. After each photograph,the jig was removed. The jig was also constructed with agrey scale consisting of three shades: white, grey, andblack, to allow for grey scale calibration of the digitalimage (Figure 4).

The photographs were repeated after 1 week, toprovide two sets of photographs. When all the photo-graphs had been taken and developed using the samemachine, the slides were recoded by a second investi-gator (NP) to allow a blind assessment by the principalinvestigator (PEB).The photographs were converted to digital images aspreviously described (Benson et al., 2000). The imageswere opened using the image analysis software (Image-Pro Plus, version 3.0 for Windows 95; MediaCybernetics, Silver Spring, MD, US A). E ach image wasindividually calibrated in millimetres, using the bracketas the calibrating m easure (Figure 1). To determine thecalibration measurement, the distance across theoutside of the tie-wings of five brackets (Figure 1) wasmeasured on two occasions, 1 week apart, with digitalcallipers (Mitutoyo Corp., Minato-ku, Tokyo, Japan).The readings were averaged (3.33 0.06 mm) and thisfigure was used to calibrate each image.The grey scale images were opened in Image-Pro Plusand calibrated using the bracket tie-wing measurement

Calibrating grey scale withwhite, grey and black areas.

Figure 4 Grey scale image of the bracketed tooth with thepositioning jig in place and showing the c alibrating grey scale.

Figure 3 Image of the positioning jig as used in a clinical study. Thejig is placed in the orthodontic bracket slot to allow reproduciblepositioning of the camera. Figure 5 QLF image of same tooth as Figure 4 showing demineral-ized dark area to the right of the bracket.



rg/

Downloadedfrom


4/10

15 2 P. E. BEN SO N ET A L.(Figure 1). The four edges of the bracket on the greyscale image were individually inspected and, if an area ofdemineralization was observed, then an area of interest(AOI) was delineated around it. The area and the meangrey scale levels of the AOI were recorded. Only whenthe observer considered an area of demineralization tobe presen t was a reading taken. Therefore, there were twoprocesses occurring. First, subjective visual assessmentsto produce a dichotomous estimate of the enamelsurface (yes or no to demineralization); and secondly, ameasurement of the enamel on those parts of the toothsurface judged to be demineralized.Quantitative light-induced fluorescenceImages of the 30 teeth were captured using the arc lampwith a liquid light guide system described by Al-Khateebet al. (1997). Two images of each tooth were captured1 week apart (Figure 5). The images were stored,processed, and analysed with customized softwaredeveloped by de Josselin de Jong (v 2.00c; InspektorResearch Systems BV). A second investigator (NP)recoded the images, which were stored on the hard driveof a computer. The principal investigator (PEB) thenanalysed each image, blindly, on two occasions at least1 week apart.The images were inspected and only dark areas ofsuspected demineralization were assessed. The custom-ized software allows the construction of a box around thedark area, including some normal enamel. In order toexclude the bracket from the image, the lesion thresholddiscriminators of the software were set at an upper limitof 95 per cent and a lower limit of 55 per cent. This wasfound to exclude the bracket from the calculation. Inaddition, the edge of the analysis box that included thebracket was 'switched off, so that the computer programexcluded this region from the calculation of soundenamel. The outcome measures from QLF are: area ofdemineralization (mm2), mean loss of fluorescence of thatarea, or AF (%), and the parameter AQ, which is themean loss of fluorescence integrated over the lesionarea (mm

2% ).

StatisticsThe repeatability was assessed using descriptivestatistics for the difference between the first and secondreadings of the same image. Only data that includedtwo readings from the same site were analysed. When arecording was taken on one occasion but not the other,this was excluded. This was done because it barred fromthe error calculation the subjective assessment of theimage, and only the error of the method was assessed.The Intra Class Correlation (ICC) (Fleiss, 1986) wasused to assess random error and the one sample t-testfor systematic error (Houston, 1983). The differences

between the readings of the first and second photo-graphic and QLF images were assessed using the limits ofagreement (Bland and Altman, 1986). The agreementbetween the area measurements of the two techniqueswas also assessed using the limits of agreement.ResultsThe descriptive statistics for repeatability are shown inTable 1. There were 60 images produced for each tech-nique (30 teeth, imaged twice). From these 60 imagesthere were 240 possible sites where demineralizationhad occurred (four sites or edges to the bracket perimage). Of the 240 sites, 92 had actually been exposed tothe demineralizing gel and the remaining 148 had notbeen exposed (Figure 2).

The number of measurements carried out from thephotographs was 108 from the first assessment and 108from the second assessment. Seven readings from thefirst assessment were not repeated in the second assess-ment and seven from the second assessment were notcarried out in the first assessment. These isolated meas-urements were excluded from the error assessmentsfor reasons explained in Materials and methods. Therewere therefore a total of 101 recordings when twomeasurements were carried out from the repeatedreadings of the same photograph.The number of measurements carried out using theQLF technique was 87 from the first reading and 85from the second reading, of which 83 were performedon the same site from both assessments.The average area of demineralization measured was2.1 mm2 with the photographic technique and 2.5 mm2with QLF. Table 1 shows that the mean difference forthe area of demineralization measured from the photo-graphs was small (-0.04 0.43 mm2). The mean differencein grey levels between the first and second recordingswas -1.71 4.57 greys. There was no evidence of system-atic bias between the recordings for the area (P = 0.120).The one sample Mest for the mean grey level gave asignificant result (P < 0.001), suggesting that there wassome systematic bias between the repeat readings, withthe second reading being higher than the first. However,a mean difference of-1.71 greys on a scale from 0 to 255can be considered small and would not be clinicallysignificant. The ICC of reliability was slightly betterfor the measurement of the grey scale level (0.89),compared with the area (0.82), but this was still withinacceptable limits.The differences between the repeat readings from theQLF images are also shown in Table 1. The mean differ-ences for the area of demineralization, mean percentageloss of fluorescence (AF), and the integration of fluor-escence loss over area (AQ) were small. The variances ofthe area and AF were also small; however, the variancefor AQ was large, leading to a large confidence interval



rg/

Downloadedfrom


5/10

MEASURING DEMINERALIZATION AROUND BRACKETS 15 3Table 1 Mean difference, standard deviation (SD) of the differences, and 95 per cent confidence intervals (CI) for the twotechniques.

Photographs (n = 101) Q L F (n = 83)

Area (mm 2) Me an grey level Ar ea (mm 2) AF (% ) AQ (m m 2 % )

Mean differenceSDCIPR

-0 .040.43-0.12-0.050.1200.82

-1.714.57- 2 . 6 1 - -0 .81 "

0.5 1.0 3.5 4 .0 . 4.5 5.0

Mean v Difference Upper Limit

Mean DifferenceLower Limit

1.5

3. i.oa2a."O 0.0> '* -0.5 8!-


6/10

15 4 P. E. BEN SO N ET A L.

J2Q>

g 2


7/10

MEASURING DEMINERALIZATION AROUND BRACKETS 15 512.0

1 0 . 0 -

8.0

7n2.0-

> -3 -2.0 oi '40

jfc -6.0 5- 8 . 0 -

-10.0

- 1 2 . 0 -

-U.0 ' ' -10* Mean v Difference

- Upper LimitMean Difference

- Lower Limit

Mean Delta F (%)Figure 9 Limits of agreem ent for AF recorded from the first and second QLF images.

60.0 i

oO) 20.0 *

/0.0 -110 .0 -90-8* * *50.0 30 0 -10.fi *

Mean v Difference- - Upper Limit

Mean DifferenceLower Limit

8 -200IIQ -40.0

Mean Delta Q (mm.'/,)Figure 10 Limits of agreemen t for AQ for the first and second QLF readings.

from a photographic image is very similar tothe technique of analysing a fluorescent image of thetooth using the customized software with QLF. Theadvantage of the QLF technique over the photographictechnique is that it estimates mineral loss by extrapo-lating the change in fluorescence of the lesion com paredwith the fluorescence of the surrounding sound enamel.The photographic technique produces a figure for thegrey level, which in itself is relatively meaningless, as

the absolute figure will be dependent upon lightingconditions, changes in processing, and even film type.This will be discussed more in Part 2.The mean differences between re peat readings of thesame sites on the photographic images were small andof the same magnitude as in a previous study (Bensonet al., 2000). There was some evidence of systematic biasin the grey scale readings, but the error was lowcompared with the size of the scale.



rg/

Downloadedfrom


8/10

156 P. E. BENSON ET AL.

o> 2n5 1oa2 2

Mean v Difference Upper Limits

Mean D ifferenceLower Limits

5 -3-1

Mean Area (mm2)Figure 11 Limits of agreement of area measured from the photograph and QLF.

Random error, as measured using the ICC, washigher than in the previous study (Benson et al., 2000).This might have been due to the different techniquesused in the two investigations. In the earlier study(Benson et al., 2000) the whole of the buccal surface ofthe tooth was automatically placed in nine differentranges of grey scale by the computer. This reduced therandom error because the computer was making theassessment, but it made the results difficult to interpret,as it was not clear which was sound and which wasdemineralized enamel.In the present study, the technique employed byWillmot et al. (2000) was used. A visual assessment ofthe enamel was carried out. If an area of demineral-ization was located then an AOI was manually drawnaround it. Whilst random error could be introducedwhen outlining the AOI, other investigators (Mitchell,1992; Linton, 1996) have found that this error is a smallproportion of the total error. Another potential sourceof random error occurs when calibrating the image, butthe technique for calibrating the images was common toboth investigations.The ICC was of the same order as the photographstaken above the occlusal plane (Willmot et al., 2000)and worse for those taken below the occlusal plane.Those authors speculated that the difference betweenthese two views was due to the position of the maskingused on the flash to reduce the amount of reflection. Inthe present study the masking was placed on the lowerpart of the ringflash as indicated by the improvedresults. However, the camera was orientated withthe jig that was placed in the bracket slot. This wasgenerally perpendicular to the buccal surface of the

tooth and not below the occlusal plane. This couldexplain the poorer random error from this technique.The ideal position of the masking for photographs ofteeth with orthodontic brackets may require furtherinvestigation.The ICC for the area measurement using QLF wasbelow 0.81, which is considered to be acceptable forrandom error (Fleiss, 1986). The random error for QLFneeds to be reduced, possibly with improvements in thecamera resolution and capture facilities.The limits of agreement for the area measurementswere small (Figure 6). This indicates good agreement inmeasuring the area of demineralization between thetwo slides taken 1 week apart. The limits are smallerthan that of the previous study (Benson et al., 2000),because in that investigation the limits were measuredon the to tal area of the buccal surface of the tooth (mean42.0 mm2). In the present study the area measured wasonly that believed to be demineralized, which was muchsmaller (mean 2.1 mm 2). Any discrepancy in the out-lining of the AOI in the earlier study would magnify theerror, because the area involved was greater.In the previous study (Benson et al., 2000), thecamera was placed in a holder that allowed rotation ofthe camera body. Photographs of the teeth were takenat a reproducible angle. In this study, a jig was usedto allow reproducible positioning of the camera. Thegood agreement between the two slides taken 1 weekapart indicates that the jig is a satisfactory method ofproducing reproducible photographs.The limits of agreement for the m easurement of meangrey levels from the two slides taken 1 week apart showsimilar agreement to the results of Willmot et al. (2000).



rg/

Downloadedfrom


9/10

MEASURING DEMINERALIZATION AROUND BRACKETS 1 57Variations in the lighting and processing of the imagemight lead to differences in the images that affect thegrey scale measurement. The photographs were takenand developed using a standardized technique understandardized conditions, which exceeded that possiblein the clinic. It is likely that t her e will be m ore inconsist-encies in photographs taken in the clinical situation. Toovercome these variations, a calibrating grey scale wasincorporated into the jig (Figure 4). The use of thecalibrating grey scale to manipulate the image digitallyso the grey scales of the two images are closely matchedis an area that requires further investigation.

The agreement of the QLF technique as measuredwith the limits of agreement (Figures 8 and 9) was verysimilar to the photographic technique for both the areaan d AF measurements. The limits of agreement for thearea measurement were slightly wider than for thephotographic technique, but were still acceptable. Asimilar precaution applies to the QLF technique as tothe photographic technique. These images were taken inthe laboratory under ideal conditions. In the clinicalsituation, in the presence of saliva the images will notbe so good. In addition, although QLF is optimized toreduce reflections, precautions need to be taken toexclude all fluorescent lighting, which can also adverselyaffect the images.

Delta Q has been advocated as a summary measure-ment of the area and the mean change in fluorescence(van der Veen and de Josselin d e Jong, 1999). How ever,th e AQ measurement in this study showed a largevariation and wide confidence limits. The limits ofagreement were also wider for AQ (Figure 10) indica-ting poorer agreement between the two recordings. Itis suggested that when measuring demineralizationaround o rthodo ntic brackets the area and AF are statedseparately.

Delta F is a more valuable figure than the grey levelfrom a photograph . It represents a quantitative m easure-ment of demineralization for that particular tooth surface.On the other hand, the grey level is a generic numberthat represents relative change only if subtracted froman area that is considered sound. Differences in lightingand developing may denote that a certain grey level onone image that represents demineralized enamel mayrepresent sound enamel on another image. Thereforefurther subjective assessments and calculations arerequired to make the grey level a practical figure foranalysis.

There was reasonable agreement between the photo-graphic and Q LF techniques for the measurement of area(Figure 11). The limits of agreement were -1.8- 1.8 m m 2.The interesting cluster of results running in a straightline with a mean area of 0.5 mm 2 were data that wererecorded from the photograph, but not using QLF, manyof which were false positives. Elimination of these datadid not affect the limits of agreement.

Part 2 of this investigation describes the validity ofeach technique.

ConclusionsThis in vitro study carried out under ideal conditions hasshown that:1. Demineralization surrounding orthodontic bracketscan be quantified reproducibly using the two tech-niques of computerized image analysis from photo-graphic slides converted to digital images and QLF.2. The two techniques show good agreement withrespect to quantifying the area of demineralization.

Address for CorrespondenceDr P. E. B ensonDepartment of Child Dental HealthSchool of Clinical DentistryClaremont CrescentSheffield S10 2TA, UK

ReferencesAl-Khateeb S et al. 1997 Quantification of formation andremineralization of artificial enamel lesions with a new portablefluorescence device. Advances in Dental Research 11: 502-506Angmar-Mansson B, Al-Khateeb S, Tranaeus S 1996 Monitoring thecaries processoptical methods for clinical diagnosis andquantification of enamel caries. European Journal of Oral Science104: 480-485Benson P E, Pender N, Higham S M, Edgar W M 1998Morphometric assessment of enamel demineralisation fromphotographs. Journal of Dentistry 26: 669-677Benson P E, Pender N, Higham S M 2000 Enamel demineralisationassessed by computerised image analysis of clinical photographs.Journal of Dentistry 28: 319-326Bland J M, Altman D G 1986 Statistical methods for assessingagreement between two methods of clinical assessment. Lancet i:307-310Dooland M B, Wylie A 1989 A photographic study of enameldefects among South Australian school children. Australian

Dental Journal 34: 470-473Ellwood R P 1996 A photographic study of developmental defectsof enamel in Brazilian school children. International DentalJournal 46: 69-75Fleiss J L 1986 Design and analysis of clinical experiments. Wiley,New YorkGoreleick L, Geiger A M, Gwinnett A J 1982 Incidence of whitespot formation after bonding and banding. American Journal ofOrthodontics 81: 93-98Houston W J B 1983 The analysis of errors in orthodonticmeasurements. American Journal of O rthodontics 83: 382-390Houwink B, Wagg B J 1979 Effect of fluoride dentifrice usage duringinfancy upon enamel mottling of the permanent teeth. Caries

Research 13: 231-237Ishii T, Suckling G 1991 The severity of dental fluorosis in childrenexposed to water with a high fluoride content for various periodsof time. Journal of Dental Research 70: 952-956



rg/

Downloadedfrom


10/10

15 8 P. E. BENSON ET AL.Linton J L 1996 Quantitative measurements of remineralization ofincipient caries. American Journal of Orthodontics and Dento-facial Orthopedics 110: 590-597Mitchell L 1992 An investigation into the effect of a fluoride

releasing adhesive on the prevalence of enamel surface changesassociated with directly bonded orthodontic attachments. BritishJournal of Orthodontics 19: 207-214Mizrahi E 1982 Enamel demineralisation following orthodontictreatment. Am erican Journal of Orthodontics 82: 62-66Mizrahi E 1983 Surface distribution of enamel opacities followingorthodontic treatment. American Journal of Orthodontics 83:323-331

Small B W, Murray J J 1978 Enamel opacities: prevalence, classifica-tions and aetiological considerations. Journal of Dentistry 6:3 3 ^ 2ten Bosch J J, Borsboom P C F, ten Cate J M 1980 A nondestructive

method for monitoring de- and rem ineralization of enamel. CariesResearch 14: 90-95Van der Veen M H, de Josselin de Jong E 1999 Introduction of a newparameter DQ for incipient caries measurement with QLF. CariesResearch 33: 318 (abstract)Willmot D R, Benson P E, Pender N, Brook A H 2000Reproducibility of quantitative measurement of white enameldemineralisation by image analysis. Caries Research 34: 175-181



rg/

Downloadedfrom

Documents

Eur J Orthod 2003 Benson 149 58