DIAGNOSTIC QUALITY OF CONVENTIONAL AND DIGITAL RADIOGRAPHIC IMAGES OF DENTAL CARIES

DIAGNOSTIC QUALITY OF CONVENTIONAL AND DIGITAL RADIOGRAPHIC IMAGES OF DENTAL CARIES Gui lherme Mon te iro TOSONI* Leo nor de Cas tro Mon te iro LOFFREDO** Ori valdo TAVANO*** Gulnara SCAF* Ana Lúcia Alva res CAPELOZZA*** ABSTRACT: The aim of this study was to inves ti gate intra and interobserver reproducibility and the valid ity in occlusal and approximal car ies diag no ses using con ven tional and dig i tal images. Radio graphs of six teen extracted human pre mol ars and molars teeth were taken with con ven tional films (Kodak Ektaspeed Plus) and stor age phos phor plate sys tem (Digora). Two observ ers read the con ven tional and dig i tal images by using dig i tal pro cess ing modal i - ties such as enlarge ment, inver sion and third dimen sion effect. The car i ous lesions was scored by a scale of five points. Histologic sec tions were the val i - dat ing cri te ria. The intra and interobserver reproducibility of radio graphic images were deter mined by Kappa sta tis tics. The valid ity was ver i fied using sen si bil ity and spec i fic ity. The intraobserver reproducibility for the con ven - tional radi og ra phy was good for approximal and occlusal sur faces, and the interobserver reproducibility was fair for the approximal sur faces, and good for the occlusal sur faces. The intra and interobserver reproducibility for dig i tal radi og ra phy was dif fer ent for both observ ers, accord ing to den tal sur face and dig i tal pro cess ing modal i ties, show ing a range between fair and opti mum * Depar ta mento de Diag nós tico e Cirur gia Facul dade de Odon to lo gia de Ara ra quara UNESP 14801-903 Ara ra quara SP. ** Depar ta mento de Odon to lo gia Social Facul dade de Odon to lo gia de Ara ra quara UNESP 14801-903 Ara ra quara SP. *** Depar ta mento de Esto ma to lo gia Facul dade de Odon to lo gia de Bauru USP 17012-901 Bauru SP. Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 279

results. Con sidering the valid ity, there were no sig nif i cant dif fer ences bet - ween con ven tional and dig i tal images for the occlusal and approximal sur fa - ces. Given the dif fer ent lev els of reproducibility offered by the dig i tal images mo dal i ties, an indi vid ual modal ity is not rec om mended. KEYWORDS: Den tal radi og ra phy; den tal car ies; dig i tal image. Intro duc tion The validity of the radiographic examination increases substantially using digital processing of the radiographic image. 13, 24 Sub jec ti vely, most of the obser vers pre fer the digi tal pro ces sing of the radi o grap hic image, and this pro ces sing seems to be depen dent on the diag nos tic pro ce dure. Image tre at ment must be used in digi tal radi o grap hies. 23 Radi o grap hic inter pre ta tion has many sub jec tive cha rac te ris tics and it depends very much on the obser ver. Accor ding to Grön dahl, 6 the decre ase of caries pre va lence, diag nos tic met hods with gre a ter spe ci fi - city are nee ded, and when con si de ring the radi o graphy, it is ne ces sary to remem ber that the obser ver is an inte gral part of the met hod. Many aut hors have shown the vari a bi lity among the obser vers caries diag no - ses. 1, 15, 17, 20 Even so, these aut hors do not eva lu ate the intra and inte - robserver reproducibility in the caries diagnosis. To eva lu ate the qua - lity of the radi o grap hic inter pre ta tion using dif fe rent obser vers, the in - tra and inte rob ser ver repro du ci bi lity is a deter mi ning fac tor for the qua lity of the diag nos tic met hod. 2 The disa gre e ment bet ween diag no ses leads to the need of kno - wing the iden tity of the inter pre ta ti ons done. The re fore, it is neces sary to verify the repro du ci bi lity of the appro xi mal and occlu sal caries diag - no sis accor ding to the con ven ti o nal and direct digi tal image, tes ting some of the digi tal image pro ces sing moda li ties by the Digora System. When verifying the diagnostic quality of a method, reproducibility and vali dity are deter mi nants. Repro du ci bi lity deter mi nes the intra and in te rob ser ver repro du ci bi lity, that is, if the same obser ver is able to ma ke identical diagnoses on different occa si ons or if dif fe rent obser vers can make the same diag no sis. The vali dity of the met hod is expres sed by the sensitivity, which is the abi lity to deter mine the pre sence of dise ase, and by the spe ci fi city, which is the abi lity to deter mine the absence of dise ase, both pro ved by a met hod that deter mi nes the true sta tus of the dise ase. The posi tive and nega tive pre dic tive values must also be 280 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001

con si de red, since they are affec ted by dise ase pre va lence in the popu - la tion being tes ted and are par ti cu larly impor tant if the dise ase is rare. Al though the aspects of repro du ci bi lity and vali dity are inde pen dent, a method that presents low reproducibility presents low diagnostic quali - ty, which may com pro mise its vali dity. The aim of this study was to inves ti gate intra and inte rob ser ver re - pro du ci bi lity and the vali dity in occlu sal and appro xi mal caries diag no - ses using conventional and digital radiography. Mate rial and met hod For this in vitro study, six teen extrac ted human pre mo lars and molars were used. The teeth were set in appro xi mal con tact, simu la ting a mouth with the pos te rior arches. The appro xi mal and occlu sal sur fa - ces of the teeth had a spec trum of caries ran ging from sound to var ying degrees of dis co lo ra tion to cavi ta tion. The teeth were radiographed by conventional radiography using Ektas peed Plus (East man Kodak, Roches ter, N.Y.) and Digora sto rage phosphor plates (Sore dex/orion Corp., Hel sinki, Fin land). A 15 mm acr y - lic plate was pla ced adja cent to the teeth to pro duce a soft-tissue equi - va lente scat te ring. The X-ray unit was the Gene ral Elec tric 100, ope - ra ting at 70 kvp. The cone was cir cu lar with a dia me ter of 7.5 cm. The tar get-film dis tance was 40 cm, 4mAs for the con ven ti o nal film and 3mAs for the digi tal phosp hor plate. The films were pro ces sed by Peri-Pro II 91000 (Air Tech ni ques Inc., U.S.A.). The image was scan ned imme di a tely after the radi o grap hic expo - sure, resul ting in an image with a 416 x 560 pixels size matrix. The digi - tal ima ges were sto red in a TIF for mat, occup ying a memory of 234 kb each. The digi tal and con ven ti o nal radi o graphs were read twice, in a 15 day inter val, by two obser vers, A and B. The obser vers, both radi o lo - gists, were cali bra ted for caries clas si fi ca tion using sco res of 5 points (0-sound; 1-enamel caries less than ½ way through ena mel; 2-enamel caries pene tra ting at least ½ way through ena mel; 3-caries of ena mel and den tine pene tra ting less than ½ way to pulp cavity; 4-caries of ena mel and den tine pene tra ting more than ½ way den tine toward pulp ca vi ty). Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 281

The radi o grap hies were read in a redu ced illu mi na tion room, under the light of a view box. The sco res were writ ten in files by two obser - vers, inde pen dently, accor ding to the sur face of the teeth exa mi ned. The digi tal ima ges were read on a Pen tium 166 mhz com pu ter with a 14 SVGA Phi lips moni tor, with small fonts and 1024 x 768 pixels reso lu tion. The com pu ter was in a redu ced illu mi na tion room, wit hout any kind of light or shine on the moni tor. The obser vers were wit hin a dis tance of 50 and 70 cm from the moni tor. The Digora System for Win - dows 3.11 was used for the pro ces sing and inter pre ta tion of the ima ges. The ima ges were analy zed the nor mal way with enlar ge ments of 1X, 2X and 3X, in the 3D mode (third dimen sion effect) with enlar ge ments of 1X (3D) and 2X (3DX), and the inver ted mode with enlar ge ments of 1X-, 2X- and 3X-. The ima ges were shown on the moni tor at ran dom, and the obser - vers had the fre e dom to adjust brigh tness, con trast and edge enha ce - ment fil ter. After the adjust ment, accor ding to obser ver s pre fe rence, they gave the score for each den tal sur face exa mi ned. Both obser vers made a total of 1,728 deci si ons. After assess ment, the teeth were seri ally sec ti o na ted in pla nes ori en - ted mesi o dis tally and paral lel to the long axis of the crown, resul ting in 6-10 sec ti ons from each tooth, with a thick ness of 800 µm. Each slice was man nu ally worn down and smo ot hed rea ching 120-150 µm, using a sand pa per under water. The sec ti ons were was hed in water and dried in air and then, were moun ted and fixed onto an object slide and exa mi - ned using a stereomicroscope (x10 magnification) by one independent examiner. One approximal surface was damage during sectioning and, the re fore, exclu ded from the mate rial which the re af ter con sis ted of a total of 47 sur fa ces. The micros co pic analy sis ser ved as the vali da tion for the pre sence of caries lesion. In order to esti mate the vali dity, the diag nose of caries dise ase was clas si fied in a dicho to mous scale: nega tive (score 0) or posi tive (score 1 to 5). Kappa (k) sta tis tics was used, accor ding to Light, 11 to ob ta in the intra and interobserver reproducibility values for the caries diag no sis by the conventional and digital radiographic methods. The value found for this reproducibility was classified, according to the reproducibility level, by the Landis & Koch 9 scale. In addi tion, a test of hypot he sis for repro - du ci bi lity was per for med in order to eva lu ate the sig ni fi cance of k value, supposing that the k variable has an approximately normal distribution. The sig ni fi cant level adop ted was 5% for the deci sion taken. 282 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001

Considering microscopic examination as the gold standard, sen - si ti vity and spe ci fi city were cal cu la ted. Result The intra and inte rob ser ver repro du ci bi lity results are shown in Figu res 1-6. Both obser vers obta i ned simi lar results in intra ob ser ver reproducibility for the approximal surfaces in the conventional radiography (Figu res 1 and 2). On the occlu sal sur fa ces (Figu res 3 and 4), the values of reproducibility in the conventional radiography were distinct, ran ging from good (obser ver B) to opti mum (obser ver A). 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 Fair Good Optimum FIGURE 1 Kappa values for intra ob ser ver repro du ci bi lity (obser ver A) in the con ven ti o nal radi o graphy (C) of appro xi mal sur fa ces and for the dif fe rent digi tal radi o graphy moda - lities. In the digital radiography, the values of intraobserver reproducibility were dis tinct bet ween the two obser vers, sho wing levels of repro - du ci bi lity that varied bet ween fair and opti mum, depen ding on the den - tal sur face and the digi tal image pro ces sing (Figu res 1-4). Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 283

1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 Good Optimum FIGURE 2 Kappa values for intra ob ser ver repro du ci bi lity (obser ver B) in the con ven ti o nal radi o graphy (C) of appro xi mal sur fa ces and for the dif fe rent digi tal radi o graphy moda - lities. 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 Good Optimum FIGURE 3 Kappa values for intra ob ser ver repro du ci bi lity (obser ver A) in the con ven ti o nal radiography (C) of occlusal surfaces and for the different digital radiography modalities. 284 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001

1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 Fair Good Optimum FIGURE 4 Kappa values for intra ob ser ver repro du ci bi lity (obser ver B) in the con ven ti o nal radiography (C) of occlusal surfaces and for the different digital radiography modalities. The conventional radiography presented interobserver reproducibility values on the appro xi mal sur fa ces of 0.60 and on the occlu sal sur fa - ces of 0.63 (Figu res 5 and 6). Although, accor ding to the Landis & Koch 9 scale, the appro xi mal sur fa ces have had fair repro du ci bi lity, the value of 0.60 is very close to 0.61, which is cha rac te ris tic of good repro du ci bi lity. The interobserver reproducibility values varied according to the dental sur face and the digi tal radi o graphy pro cess. The inte rob ser ver repro - du ci bi lity in the digi tal radi o graphy varied from fair (0.51) to good (0.77) on the appro xi mal sur fa ces (Figure 5), and from good (0.76) to opti mum (0.89) on the occlu sal sur fa ces (Figure 6). The interobserver reproducibility results show that some digi tal ima ges may pre sent a good qua lity diag no - sis for approximal and occlusal caries regarding the reproducibility. The true sta tus of the appro xi mal sur fa ces accor ding to micros co - pic exa mi na tion was: 1 sound sur face, 20 sur fa ces with ena mel caries and 10 sur fa ces with den tin caries. The true sta tus of occlu sal sur fa ces was: 7 sound sur fa ces and 9 den tin caries. As there was no sta tis ti cally sig ni fi cant dif fe rence bet ween the obser vers in their assign ments, the first clas si fi ca tion from obser ver A was used to prove the radi o grap hic diag no sis before the micros co pic Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 285

exa mi na tion. The sen si ti vity and spe ci fi city values for the appro xi mal and occlu sal sur fa ces are shown in Tables 1 and 2, res pec ti vely. 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 Fair Good FIGURE 5 Kappa values for inte rob ser ver repro du ci bi lity in the con ven ti o nal radi o graphy (C) of appro xi mal sur fa ces and for the dif fe rent digi tal radi o graphy moda li ties. 1 0,9 0,8 0,7 0,6 0,5 0,4 0,3 0,2 0,1 0 Good Optimum FIGURE 6 Kappa values for inte rob ser ver repro du ci bi lity in the con ven ti o nal radi o graphy (C) of occlu sal sur fa ces and for the dif fe rent digi tal radi o graphy moda li ties. 286 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001

On the appro xi mal sur fa ces (Table 1), the sen si ti vity values were mode rate to low. For the con ven ti o nal image and for the digi tal ima - ges in the 2X and 3X mode, the value was 0.50. The others digi tal ima ges (1X, 3D, 3DX, 1X-, 2X- and 3X-) values ran ged bet ween 0.35 and 0.58. Table 1 Sen si ti vity (S) and spe ci fi city (Sp) values for the appro xi mal sur fa ces accor ding to image moda lity: con ven ti o nal (C), nor mal digi tal (1X, 2X and 3X), digi tal in third dimen sion (3D and 3DX) and inver ted digi tal (1X-, 2X- and 3X-) S 0.50 0.38 0.50 0.50 0.42 0.38 0.46 0.54 0.58 Sp 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 On the occlu sal sur fa ces (Table 2), the vali dity results were more homo ge ne ous: the sen si ti vity was 0.50 for the con ven ti o nal and digi tal ima ges in the 1X, 2X and 3X mode, and 0.42 in the 3D, 3DX, 1X-, 2Xand 3X- ima ges. Table 2 Sen si ti vity (S) and spe ci fi city (Sp) values for the occlu sal sur - fa ces accor ding to image moda lity: con ven ti o nal (C), nor mal digi tal (1X, 2X and 3X), digi tal in third dimen sion (3D and 3DX) and inver ted digi tal (1X-, 2X- and 3X-) S 0.50 0.50 0.50 0.50 0.42 0.42 0.42 0.42 0.42 Sp 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 All the spe ci fi city values, for appro xi mal and occlu sal sur fa ces, were 1.00 (Tables 1 and 2). Dis cus sion The distinct intraobserver reproducibility results prove the subjec - ti vity of the exa mi na tion 7 and show that the digi tal pro ces sing ima ges Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 287

must be used by the obser vers on each den tal sur face, in the radi o gra - p hic exa mi na tion of the caries. When the obser vers used these digi tal processing modalities, the reproducibility values were similar, higher or lower than conventional radiography. The inte rob ser ver repro du ci bi lity values varied accor ding to the dental surface and the digital radiography process. The interobserver repro du ci bi lity in the digi tal radi o graphy varied from fair (0.51) to good (0.77) on the appro xi mal sur fa ces, and from good (0.76) to opti mum (0.89) on the occlu sal sur fa ces. The inte rob ser ver repro du ci bi lity results show that some digi tal ima ges may pre sent a good qua lity diag no sis for approximal and occlusal caries regarding the reproducibility. The lite ra ture men ti ons the vari a bi lity of the obser vers in the radi o - graphic interpretation of the dental caries. 7, 12, 20 Howe ver, the degree of intra and inte rob ser ver repro du ci bi lity, deter mi ned by Kappa (k) sta tis - tics, 11 is not shown in the radi o grap hic diag no sis of the den tal caries. Accor ding to Fires tone et al., 4 the reproducibility between the den - tists in the radi o grap hic exa mi na tion of caries is extre mely vari a ble. The pre sent study sho wed vari a ti ons in the intra and inte rob ser ver repro du ci bi lity values for the diag no sis of caries simi lar to those found in the lite ra ture. 3, 4, 5, 8, 10, 13 It should, howe ver, be noted that these stu - dies were not using the whole sam ple. The vari a bi lity in the radi o grap hic inter pre ta tion may be the result of: diagnostic task, 7 clinical trends, perception, education, training and expe ri ence of the obser vers. 15 Accor ding to Fires tone et al. 4 and Lazar - chik et al., 10 the inte rob ser ver repro du ci bi lity level incre a ses depen ding on the level of expe ri ence and edu ca tion of the obser ver, which sup - ports the con clu si ons made by Dunn & Kantor, 3 who des cri bed radi o - grap hic inter pre ta tion as an acti vity of high cog ni tive level based on the know ledge and expe ri ence of the obser ver. In the pre sent study, the two obser vers inter pre ted the same ima ges under the same con di ti ons, both have the same spe ci alty and expe ri ence and, besi des this, were tra i ned and cali bra ted to use the same sco res, only psychoph ysics, 3 which stu dies the func ti o nal rela ti ons bet ween the mind and physi cal phe no mena, could explain this vari a tion. Accor ding to Straub et al., 18 the radi o grap hic inter pre ta tion is com - po sed by a num ber of physi cal and psycho lo gi cal phe no mena. The per - cep tion of image qua lity by obser vers and their skills in cor rectly diag - no sing do not depend exclu si vely on the image qua lity. Dunn & Kantor 3 repor ted, as exam ples of psychoph ysi cal tests in oral radi o logy, the eva - lu a tion of view box and ambi ent light influ ence in radi o grap hic inter - 288 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001

pre ta tion, mea su ring eye move ment of obser vers when loo king for es - sential diagnostic information in the radiographic image, and mapping of neu ral path ways in the visual cor tex during the obser va tion of the image. While con ven ti o nal radi o graphy pre sents a sta tic image, once deve lo ped, the digi tal radi o grap hic image is dyna mic and may be pro - ces sed by the obser ver during the inter pre ta tion. 5, 23 Some stu dies sho wed impro ve ment in the diag no sis of caries with the digi tal pro ces sing of the image. 13, 19, 22, 24 Howe ver, Wenzel 21 repor - ted that few stu dies have been publis hed about the cli ni cal appli ca tion of the digi tal systems and until now, in vitro stu dies have not shown sig ni fi cant dif fe ren ces in the per for mance bet ween the digi tal systems, including Digora, and conventional films, which is in reproducibility with the vali dity results of this study. In the pre sent study, there were no sta tis ti cally sig ni fi cant dif fe ren - ces bet ween the con ven ti o nal and digi tal ima ges regar ding the vali dity aspect. The sam ple size and the high pre va lence of caries in the teeth, mainly inci pi ent caries in ena mel, were fac tors that con tri bu ted to the vali dity results of this study. This is in repro du ci bi lity with Mile man & Van der Weele, 12 who con si de red the selec tion and pre va lence of dise a - ses to be impor tant in a bat tery of diag nos tic tests. Moys tad et al., 13 Svanaes et al. 19 and Scaf et al., 16 obser ved gre a ter diag nos tic vali dity in the enlar ged ima ges, alt hough there is a limit beyond the diag nos tic vali dity may be redu ced. Enlar ge ment incre a ses the appa rent size of the image, but the infor ma tion con tent of this image is limi ted to the ori gi nal reso lu tion when cap tu red. 3 Regar ding the vali dity aspect, our results didn t show dif fe ren ces bet ween image enlar ge ments on the appro xi mal and occlu sal sur fa ces, with excep tion of inver ted ima ges taken on the appro xi mal sur fa ces, where image en - lar ge ment resul ted in incre a sed sen si ti vity. Rela ted to the inte rob ser ver repro du ci bi lity, the image enlar ge - ment, both on the appro xi mal and occlu sal sur fa ces, resul ted in incre a - sed reproducibility up to 2X enlargement, decreasing reproducibility value at 3X enlar ge ment. In the inver ted ima ges, there was no dif fe - rence in repro du ci bi lity in rela tion to their enlar ge ments. An inte res ting result was found in the 3D and 3DX ima ges of the occlusal surfaces. These images presented a high level of reproducibility, yet they did not show the best sen si ti vity results. This may be ex pla i ned by the high spe ci fi city of these ima ges, 14 bea ring in mind that the vali dity and repro du ci bi lity aspects are inde pen dent, alt hough a high level of dia - gnostic reproducibility does not necessarily imply a correct diagnosis. Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 289

The radi o grap hic image of the caries con sists of a pro ce dure that may lead to dif fi culty in inter pre ta tion, depen ding on inhe rent fac tors of the image itself, or the sub jec ti vity of the clas si fi ca tion. Digi tal radi o gra - phy may offer an impro ve ment in the image qua lity. The incre ase in the types of ima ges obta i ned by the digi tal pro ces sing moda li ties intro duce one more fac tor to be con si de red. The best result would be obta i ned if it were pos si ble to stan dar dize caries clas si fi ca tion in any of the digi tal ima ges moda li ties that, toget her with the other cli ni cal data, would offer ele ments for a con duct to be adop ted. Con clu sion Given the dif fe rent levels of repro du ci bi lity offe red by the digi tal ima ges moda li ties, an indi vi dual moda lity is not recom men ded. Rela ted to vali dity, there were no dif fe ren ces bet ween con ven ti o nal and digi tal ima ges. Acknow ledg ment We thank Prof. Dr. Alberto Con so laro for valu a ble sup port in the miscroscopic analysis. TOSONI, G. M. et al. Qua li dade diag nós tica das ima gens radi o grá fi cas con ven - ci o nal e digi tais da cárie den tá ria. Rev. Odon tol. UNESP (São Paulo), v.30, n.2, p.277-290, 2001. RESUMO: O objetivo deste trabalho foi verificar a reprodutibilidade intra e inte re xa mi na dor e a vali dade das radi o gra fias con ven ci o nal e digi tal no diag - nós tico de cáries pro xi mais e oclu sais. Dezes seis den tes foram mon ta dos em um modelo e radi o gra fa dos uti li zan do-se fil mes Kodak Ekta plus e a placa de ima gem do Sis tema Digora. Dois obser va do res radi o lo gis tas inter pre ta ram as ima gens con ven ci o nais e as digi tais com os recur sos de ampli a ção, inver são e ter ce ira dimen são. As super fí cies den tá rias foram clas si fi ca das com esco - res de 0 a 5, deter mi nan do-se a pre sença e a exten são de lesão cari osa. Os den tes foram sec ci o na dos para vali da ção em exame micros có pico. Uti li - zou-se a es ta tís tica Kappa para veri fi car a repro du ti bi li dade intra e inte re xa - mi na dor. A sen si bi li dade, a espe ci fi ci dade e os valo res pre di ti vos foram obti - 290 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001

dos. Pela ra di o gra fia con ven ci o nal, a repro du ti bi li dade intra-examinador foi boa nas su per fí cies pro xi mais e oclu sais, e a inte re xa mi na dor foi regu lar nas superfícies proximais e boa nas oclusais. Pela digital, a reprodutibilidade intra -exa mi na dor foi dis tinta para os dois exa mi na do res, apre sen tando valo - res que vari a ram entre regu lar e ótimo. Nas pro xi mais, a repro du ti bi li dade inte re xa mi na dor da ra di o gra fia digi tal variou de regu lar a boa e, nas oclu sais, de boa a ótima. Con si de rando o aspecto de vali dade, não houve dife ren ças sig ni fi ca ti vas en tre as ima gens con ven ci o nal e digi tais. Em razão dos dife - ren tes níveis de re pro du ti bi li dade apre sen ta dos pelos recur sos das ima gens digi tais, não é reco men dada a uti li za ção de um recurso indi vi du al mente. PALAVRAS-CHAVE: Radi o gra fia den tá ria; cárie den tá ria; ima gem digi tal. References 1 CAYLEY, A. S., HOLT, R. D. The influ ence of audit on the diag no sis of occlu - sal caries. Caries Res. (Basel), v.31, n.2, p.97-102, Mar. 1997. 2 DOUGLASS, C. W. Eva lu a ting diag nos tic tests. Adv. Dent. Res. (Was hing - ton), v.7, n.2, p.66-9, Aug. 1993. 3 DUNN, S. M., KANTOR, M. L. Digi tal Radi o logy. Facts and fic ti ons. J. Am. Dent. Assoc. (Chicago), v.124, n.12, p.39-47, Dec. 1993. 4 FIRESTONE, A. R. et al. The effect of expe ri ence and tra i ning on the diag no - sis of appro xi mal coro nal caries from bite wing radi o graphs. Schweiz. Mo - na tsschr. Zahn med. (Zurich), v.104, n.6, p.719-23, 1994. 5 GOTFREDSEN, E., WENZEL, A, GRÖNDAHL, H-G. Obser vers use of image enha ce ment in asses sing caries in radi o graphs taken by four intra-oral digi - tal systems. Dentomaxillofac. Radiol. (Erlangen), v.25, n.1, p.34-8, Jan. 1996. 6 GRÖNDAHL, H-G. The influ ence of obser ver per for mance in radi o grap hic caries diagnosis. Swed. Dent. J. (Stockholm), v.3, n.4, p.101-7, June 1979. 7 HOLLENDER, L. Deci sion making in radi o grap hic ima ging. J. Dent. Educ. (Was hing ton), v.56, n.12, p.834-43, Dec. 1992. 8 HUYSMANS, M. C. D. N. J. M., HINTZE, H., WENZEL, A. Effect of expo sure time on in vitro caries diag no sis using the Digora system. Eur. J. Oral Sci. (Cam bridge), v.105, n.1, p.15-20, Feb. 1997. 9 LANDIS, J. R., KOCH, G. G. The mea su re ment of obser ver agre e ment for cate go ri cal data. Biometrics (Ale xan dria), v.33, n.1, p.159-714, Mar. 1977. 10 LAZARCHIK, D. A. et al. Radi o grap hic eva lu a tion of occlu sal caries: Effect of tra i ning and expe ri ence. Caries Res. (Basel), v.29, n.5, p.355-8, Sept. 1995. 11 LIGHT, R. J. Mea su res of res ponse agre e ment for qua li ta tive data: some gene ra li za ti ons and alter na ti ves. Psychop har ma col. Bull. (Was hing ton), v.76, p.365-77, 1971. Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001 291

12 MILEMAN, P. A., VAN DER WEELE, L. T. The role of caries recog na tion: treatment decisions from bitewing radiographs. Den to ma xil lo fac. Radiol. (Erlan gen), v.25, n.5, p.228-33, Nov. 1996. 13 MOYSTAD, A. et al. Effect of image magnification of digitized bitewing radiographs on appro xi mal caries detec tion: an in vitro study. Dentomaxillofac. Radiol. (Erlan gen), v.24, n.4, p.255-9, Nov. 1995. 14 PEREIRA, M. G. Afe ri ção dos even tos. In:. Epi de mi o lo gia. Teo ria e prática. Rio de Jane iro: Gua na bara Koo gan, 1995. cap. 17, p.358-75. 15 PETRIKOWSKI, C. G. et al. Inte rob ser ver vari a bi lity in radi o grap hic inter pre - ta tion of pedi a tric den tal dise a ses: A pilot study. J. Can. Dent. Assoc. (Toronto), v.62, n.9, p.723-30, Sept. 1996. 16 SCAF, G., KAN TOR, M. L., WALSH, S. J. Effect of mag ni fi ca tion on caries detec tion with Radi o Vi si og raphy (RVG). J. Dent. Res. (Chigaco), v.72, sp. iss., p.255, Mar. 1993. (Abstract n.1217). 17 SINTON, J. M. et al. Influ ence of the addi tion of res to ra ti ons on the diag no sis of caries from digi ti zed bite wing radi o graphs. Oral Surg. Oral Med. Oral Pat hol. Oral Radiol. Endod. (St. Louis), v.84, n.4, p.443-8, Oct. 1997. 18 STRAUB, W. H., GUR, D., GOOD, B. C. Visual acu ity tes ting of radi o lo gists Is it time?. Am. J. Roent ge nol. (Spring fi eld), v.156, n.5, p.1107-8, May 1991. 19 SVANAES, D. B. et al. Intra oral sto rage phosp hor radi o graphy for appro xi - mal caries detec tion and effect of image mag ni fi ca tion. Oral Surg. Oral Med. Oral Pat hol. Oral Radiol. Endod. (St. Louis), v.82, n.1, p.94-100, July 1996. 20 WEISS, E. I. et al. Inter pre ta tion of bite wing radi o graphs. Part 2. Eva lu a tion of the size of appro xi mal lesi ons and need for tre at ment. J. Dent. (Guild - ford), v.24, n.6, p.385-8, Nov. 1996. 21 WENZEL, A. Digi tal radi o graphy and caries diag no sis. Dentomaxillofac. Radiol. (Erlan gen), v.27, n.1, p.3-11, Jan. 1998. 22 WENZEL, A., FEJERSKOV, O. Vali dity of diag no sis of ques ti o na ble caries lesi ons in occlu sal sur fa ces of extrac ted third molars. Caries Res. (Basel), v.26, n.3, p.188-94, May 1992. 23 WENZEL, A., HINTZE, H. Per cep tion of image qua lity in direct digi tal radi o - graphy after appli ca tion of vari ous image tre at ment fil ters for detec ta bi lity of den tal dise ase. Den to ma xil lo fac. Radiol. (Erlan gen), v.22, n.3, p.131-4, Aug. 1993. 24 WENZEL, A. et al. Depth of occlu sal caries asses sed cli ni cally, by con ven ti o - nal film radi o graphs, and by digi ti zed, pro ces sed radi o graphs. Caries Res. (Basel), v.24, n.5, p.327-33, Sept. 1990. 292 Rev. Odontol. UNESP, São Paulo, 30(2): 277-290, 2001