Radiographic Diagnosis of Dental Caries

S. Brent Dove, D.D.S., M.S.

Abstract: The purpose of this report was to respond to aspects of the RTI/UNC systematic review relating to the radiographic diagnosis of dental caries. The systematic review was commissioned as part of the NIH Consensus Development Conference on Diagnosis and Management of Dental Caries Throughout Life. The systematic review evaluated the dental literature from 1966 to 1999. Well-defined search criteria along with clear inclusion and exclusion criteria were used to perform the review. Some of the inclusion and exclusion criteria used in the systematic review may have limited the evidence supporting the use of radiography, especially for the diagnosis of proximal surface caries. The RTI/UNC review only included studies in which sensitivity and specificity were reported or could be derived from the data presented. Studies that used the receiver operating characteristic as a measure of diagnostic accuracy were not included. Although the strength of evidence is considered poor, this does not mean that the use of radiographic methods is of no diagnostic value. It simply means that, using the criteria established by the systematic review, the evidence is inadequate to validate the method. Guidelines should be developed for assessing diagnostic methods that assist researchers in developing study designs that will hold up to critical review. Dr. Dove is Division Head, Department of Dental Diagnostic Sciences, University of Texas Health Science Center. Direct correspondence to him at Department of Dental Diagnostic Sciences, University of Texas Health Science Center, 7703 Floyd Curl Drive, San Antonio, TX 78229 ; 210-567-3332 phone; 210-567-3334 fax; dove@uthscsa.edu. Key words: dental caries, caries diagnosis, dental radiology, dental radiography, oral diagnosis

lmost since the discovery of x-rays by Wilhelm Conrad Roentgen in 1895, radiography has been used to detect the effects of dental caries on dental hard tissues. Radiography has been primarily used for the detection of lesions on the proximal surfaces of teeth, which are not clinically visible for inspection. Radiographs are also recommended as a supplement to the clinical examination of occlusal surfaces for the detection of pit and fissure caries. Over the years, it has been well established that more dental caries are detected by radiography than by clinical examination alone.1-6 Radiographic diagnosis of dental caries is fundamentally based on the fact that as the caries process proceeds, the mineral content of enamel and dentin decreases with a resultant decrease in the attenuation of the x-ray beam as it passes through the teeth. This is recorded on the image receptor as an increase in radiographic density. This increase in radiographic density must be detected by the clinician as a sign of a carious lesion. Many different factors can affect the ability to accurately detect these lesions: exposure parameters, type of image receptor, image processing, display system, viewing conditions, and ultimately, the training and experience of the human observer. A systematic review of the existing literature was performed by the RTI/UNC Evidence-Based Practice Center to address the question of the validity of six different diagnostic methods for the detection of dental caries in primary and permanent teeth. The diagnostic

methods assessed were: visual and visual/tactile inspection, radiography, fiber-optic transillumination (FOTI), electrical conductance (EC), laser fluorescence (LF), and combinations of these methods. Three primary computer indexes used in searching the literature were MEDLINE, EMBASE, and the Cochrane controlled trials register. The period searched was from January 1966 to December 1999. Inclusion and exclusion criteria were clearly defined prior to performing the search. Studies were limited to those with human subjects and natural carious lesions, publication language in English, and histological validation of caries status for each surface studied or visual/tactile validation of intact surface for cavitation only; outcomes must be expressed as sensitivity and specificity or provided data from which these outcomes could be derived. While both in vitro and in vivo studies were included in the review, only those methods that are commercially available to the general practitioner were assessed. Thirty-nine studies were selected from among 1,407 diagnostic reports that satisfied all criteria. These studies reported 126 different assessments of different diagnostic methods. Of these studies, 65 assessements evaluated the diagnostic performance of radiographic methods. The studies were critically reviewed and a quality rating scale assessed that appraised several elements of internal validity, including study design, duration, sample size, blinding of examiners, baseline assessments, and examiner reli-

October 2001

Journal of Dental Education

985

ability. The overall strength of evidence supporting the validity of a method was judged in terms of the extent to which it offered clear, unambiguous assessment of a particular method for identifying a specific type of lesion on a specific type of surface. Some of the inclusion and exclusion criteria used in the systematic review may have limited the evidence supporting the use of radiography, especially for the diagnosis of proximal surface caries. The RTI/UNC review only included studies in which sensitivity and specificity were reported or could be derived from the data presented. Studies that used the receiver operating characteristic as a measure of diagnostic accuracy were not included. Receiver Operating Characteristic (ROC) analysis is a method to determine the diagnostic accuracy of a particular method of assessment. ROC is based upon signal detection theory and provides for an unbiased measure of discrimination in the paired-comparison or forced-choice situation.7 This is exactly the type of choice the dentist faces when determining the presence or absence of a carious lesion on a radiograph. Using sensitivity and specificity values for a diagnostic test or imaging modality can be ambiguous as dentists exhibit a wide variation in their decision criteria.8 The ROC analysis gives a measure of discrimination that is independent of the cut-off points of the decision criterion and, therefore, unbiased by them.7 Recently, many researchers evaluating the diagnostic performance of radiographic methods have advocated the use of ROC analysis to evaluate imaging systems for the diagnosis of dental caries.9 A search of the Medline database from January 1966 to December 2000 using [exp dental caries/ or dental caries.mp. (21,172)] and [exp radiography/ or dental radiology.mp. or exp radiography, dental, digital/ or exp radiographic image enhancement (101,704)] and [exp ROC curve/ or ROC curve.mp. (2,167)] as search criteria resulted in sixty-two reports. Including these studies may have improved the strength of evidence for radiographic methods for the detection of dental caries. Another criteria used for inclusion also had a significant effect on the overall outcome of the assessment. Studies included were required to have histological validation of caries status for each surface studied. Some exceptions were made with regard to those studies where cavitation was the extent of lesions to be detected. Due to the practical and ethical limitations of obtaining histological confirmation, the majority of assessments were in vitro (six in vivo and fifty-nine in vitro). In the determination of a quality rating, the maximum score for experimental setting was twenty points. Considering that an in vitro study was given a score of zero and an in vivo study was given a score of two for

experimental setting, 10 percent of the overall quality score was affected by this criterion. While some of the methods such as visual inspection, fiber-optic transillumination, and electric conductance can be greatly affected by the setting in which they were performed, the basic physics of image formation should not be greatly affected by difference between a laboratory and clinical setting, provided extracted human teeth and natural caries are being studied. A meta-analysis of factors involved in the validity of radiographic diagnosis for proximal surface caries indicated that experimental setting did in fact have an impact on diagnostic performance.10 Contrary to the results of this meta-analysis, a more recent direct experimental examination of the same question supports the idea that no such relationship exists. Hintze and Wenzel11 directly compared the diagnostic accuracy of radiographs obtained both in vivo and in vitro of the same teeth for the detection of occlusal and proximal surface caries. The results of their study suggest that no difference could be found between in vivo and in vitro results using ROC analysis. The RTI/UNC systematic review of the dental literature indicates that the strength of evidence for radiographic methods for the detection of dental caries is poor for all types of lesions on posterior proximal and occlusal surfaces. This was primarily due to the large amount of variation in the reported sensitivity and specificity of this method. Little if any evidence exists to support the use of radiographic methods for primary teeth, anterior teeth, and root surfaces. The literature is severely limited by problems associated with both internal and external validity. These include: incomplete descriptions of sample selection, diagnostic criteria and examiner reliability, the use of small numbers of examiners, nonrepresentative teeth, samples with high lesion prevalence, and the use of reference standards of questionable reliability. Although the strength of evidence is considered poor, this does not mean that the use of radiographic methods is of no diagnostic value. It simply means that, using the criteria established to evaluate the existing evidence, we find the evidence is inadequate to validate the method. Better studies designed to address the limitations of the current literature could in fact indicate that the method is valid. It does call into question the relative importance of this method in making treatment decisions. A review of the RTI/UNC report indicates that most of the variability in diagnostic performance of posterior proximal and occlusal surfaces was in fact associated with the sensitivity of the method and not the specificity. Table 1 shows the radiographic assessments of the diagnosis of cavitated lesions, lesions involving dentin, and any lesions on proximal surfaces

986

Journal of Dental Education Volume 65, No. 10

of posterior teeth. For those assessments involving cavitated lesions, the standard deviation of the mean sensitivity was 0.21, whereas the standard deviation of the mean specificity was only 0.04 (Table 1). This same trend is consistent for all evaluations of proximal surfaces regardless of lesion progression. This trend is not apparent when considering the diagnosis of occlusal caries involving dentin (Table 2). However, the variability in sensitivity is high when compared to that of specificity when we evaluate the occlusal surfaces of permanent posterior that had lesions of different depths (Table 2). On further review, considerable outliers ap-

pear in each dataset. These have been indicated in bold. Some of the variability may be explained by the detection task itself and the decision criteria used by different evaluators. When radiographic interpretation is performed, the decisions are presented as either the presence or absence of a lesion. But the decisions are not always so black-and-white, but in fact lie in a gray continuum of negative to positive. Therefore, several different values along this continuum could be selected as a cutoff to determine if dental caries is present or absent. Depending upon whether more stringent or more lenient criteria are used for detection, the sensitivity and specificity can vary dramatically. If more stringent

Table 1. Permanent posterior teethproximal surfaces

Citation, Method Rugg-Gunn, 197212 D speed film Downer, 197513 D-speed film Mejare, et al., 198514 D-speed film Pitts & Rimmer, 199215 D-speed film Hintze, et al., 199816 E-speed film Espelid & Tveit, 198617 D-speed film Mean Performance Mileman & van der Weele, 199018 D-speed film Verdonschot, et al., 199119 D-speed film Russell & Pitts, 199320 D-speed film E-speed film RVG Ricketts, et al., 199721 D-speed film Mean Performance Heaven, Firestone, & Feagin, 199222 D-speed film with image analysis Russell & Pitts, 199320 D-speed film E-speed film RVG Ricketts, et al., 199721 D-speed film Firestone, et al., 199823 D-speed film Film image analysis Sensaray image analysis Mean Performance 16 1 240 3 75% NR 105 276 21 3 240 3 43% NR NR Lesion into outer of dentin Lesion reaching DEJ Lesion penetrating DEJ Sites Raters 370 NR 185 NR 598 3 1468 1 338 4 151 7 Prevalence 9% 36% 5% 1% 6% 19% Criteria Lesion in enamel & outer dentin Lesion at DEJ or beyond Lesion 2/3 enamel thickness Lesion into dentin Lesion into dentin Lesion involving DEJ Sensitivity 0.35 0.73 0.36 0.87 0.63 0.69 0.61 0.21 0.54 0.50 0.29 0.30 0.16 0.16 0.33 0.16 1.0 Specificity 1.00 0.97 0.98 0.99 0.93 0.89 0.96 0.04 0.97 0.94 0.92 0.96 0.96 0.99 0.96 0.02 1.0 All Lesions All Lesions Dentin Lesions Dentin Lesions Dentin Lesions Type Cavitated Lesions Cavitated Lesions Cavitated Lesions Cavitated Lesions Cavitated Lesions Cavitated Lesions

96teeth 5

13%

Lesion into dentin

Dentin Lesions

NR

Lesion penetrating DEJ Lesion into dentin NR

0.26 0.25 0.15 0.27 0.61 0.78 0.73 0.51 0.31

0.90 0.90 0.92 0.97 0.86 0.74 0.82 0.89 0.08

96teeth 5 102 1

37% 66%

All Lesions All Lesions

October 2001

Journal of Dental Education

987

Table 2. Permanent posterior teethocclusal surfaces

Citation, Method Wenzel, et al., 199024 D-speed film Digitized film enhanced Wenzel, et al., 199125 E-speed film Digitized film Digitized film enhanced RVG contrast enhanced RVG density saturation Wenzel & Fejerskov, 199226 E-speed film Digitized film Digitized film enhanced Nytun, Raadal, & Espelid, 199227 Film Keltey & Holt, 199328 D-speed film Russell & Pitts, 199320 D-speed film E-speed film RVG Lussi, 199329 D-speed film Verdonschot, et al., 199330 E-speed film Lussi, et al., 199531 D-speed film Ricketts, et al., 199432 D-speed film Ekstrand, Ricketts, & Kidd, 199733 D-speed film Huysmans, Hitze, & Wenzel, 199734 Digora Ricketts, et al., 199721 Ashley, Blinkhorn, & Davies, 199835 E-speed film Digora Huysmans, Longbottom, & Pitts, 199836 E-speed film Mean Performance Wenzel, et al., 199024 D-speed film Digitized film enhanced Russell & Pitts, 199320 D-speed film E-speed film RVG Lazarchik, et al., 199537 D-speed film Ricketts, et al., 199721 D-speed film Mean Performance Sites Raters 46 6 81 4 Prevalence 72% Criteria caries into dentin caries into dentin Sensitivity 0.63 0.68 0.63 0.72 0.62 0.69 0.64 0.48 0.71 0.54 0.66 Specificity 0.94 0.98 0.85 0.83 0.83 0.84 0.82 0.81 0.85 0.77 0.50 Type Dentin Lesions Dentin Lesions

67%

78 1

67%

caries reaching dentin

Dentin Lesions

30 10 100 1 120 3

77%

51% 28%

radiolucency involving dentin radiolucency into dentin radiolucency penetrating beyond DEJ caries into dentin dentinal caries caries beyond the DEJ dentine caries radiolucency to middle 1/3 of dentin caries into dentin radiolucency into dentin radiolucency into dentin

Dentin Lesions Dentin Lesions Dentin Lesions

0.67 0.18 0.21 0.21 0.45 0.61 0.62 0.62 0.54

0.92 0.98 0.99 0.97 0.83 0.79 0.77 0.76 1.00

63 24 81 4 26 6 48 12 100 3 189 3 96 5 103 1

44% 67% 42% 67% 39%

Dentin Lesions Dentin Lesions Dentin Lesions Dentin Lesions Dentin Lesions Dentin Lesions Dentin Lesions Dentin Lesions

55%

0.60

0.94

39% 36%

0.14 0.24 0.19

0.95 0.89 0.89

107 2

41%

radiolucency into dentin

0.58 0.51 0.19

0.87 0.86 0.11 0.80 0.90 0.95 0.96 0.97 0.79 0.97 0.91 0.08

Dentin Lesions

46 2 120 3

89%

caries into enamel radiolucency penetration beyond DEJ caries present radiolucency into dentin

0.73 0.79 0.12 0.12 0.15 0.58 0.27 0.39 0.30

All Lesions All Lesions

NR

100 15 96 5

79% 70%

All Lesions All Lesions

988

Journal of Dental Education Volume 65, No. 10

criteria are applied, fewer false-positives will occur, resulting in higher specificity at the expense of sensitivity and the associated increase in false-negative decisions. If, however, more lenient criteria are used, the inverse is typically true. All of the evidence suggests that radiographic methods have a higher degree of specificity than sensitivity, which means that false-negative diagnoses are proportionally more apt to occur in the presence of disease than are false-positive diagnoses in the absence of disease. This outcome may be beneficial if the negative consequences of a false-positive diagnosis outweigh that of a false-negative diagnosis. Currently the most common type of intervention is surgical removal of the lesion, so a false-positive diagnosis results in a perfectly normal tooth being irreversibly damaged. A false-negative results in further progression of the lesion and potentially further loss of tooth tissue. This outcome is somewhat mitigated by the fact that the lesion may be detected at a later time. Nonsurgical interventions are gaining in popularity as alternatives to mechanical replacement of damaged tooth tissue with artificial materials. These nonsurgical methods are only effective if the lesion is detected prior to cavitation. This means that the lesion must be detected early. To detect the lesion earlier, a diagnostic method must provide for higher sensitivity, which may result in more false-positive diagnoses. If early interventions consist of nonsurgical management, which does not result in any permanent damage to the tooth, then the negative consequences of a false-negative diagnosis outweigh those of a falsepositive diagnosis. Every attempt should be made to increase the sensitivity of currently available methods of radiographic diagnosis to coincide with this change in intervention strategy. New digital radiographic techniques that eliminate the use of silver halide emulsion x-ray film by capturing radiographic images on photostimulable phosphor imaging plates or charge-coupled devices may improve detection of dental caries. The images acquired with these technologies are digital and can be processed or analyzed to enhance diagnostic performance. The weight of available evidence suggests that the use of some digital methods offers small gains in sensitivity without reduction in specificity and that image analysis techniques may offer more substantial gains. Renewed effort should be made to ensure that future studies address the question of diagnostic validity adequately. Guidelines should be developed for assessing diagnostic methods that assist researchers in developing study designs that will hold up to critical review.

REFERENCES
1. Mann J, Pettigrew JC, Revach A, Arwas JR, Kochavi D. Assessment of the DMF-S index with the use of bitewing radiographs. Oral Surg Oral Med Oral Pathol 1989;68:661-5. 2. Creanor SL, Russell JI, Strang DM, Stephen KW, Burchell CK. The prevalence of clinically undetected occlusal dentine caries in Scottish adolescents. Br Dent J 1990;169:126-9. 3. de Vries HCB, Juiken HMHM, Konig KG, vant Hof MA. Radiographic versus clinical diagnosis of approximal carious lesions. Caries Res 1990;24:364-70. 4. Kidd EAM, Naylor MN, Wilson RF. Prevalence of clinically undetected and untreated molar occlusal dentine caries in adolescents on the Isle of Wight. Caries Res 1992;26:397-401. 5. Weerheijm KL, Groen HJ, Bast AJJ, Kieft JA, Eijkman MAJ, van Amerongen WE. Clinically undetected occlusal dentine caries: a radiographic comparison. Caries Res 1992;26:305-9. 6. Hintze H, Wenzel A. Clinically undetected dental caries assessed by bitewing screening in children with little caries experience. Dentomaxillofac Radiol 1994;23:19-23. 7. Swets JA. The relative operating characteristic in psychology. Science 1973;182:990-1000. 8. Vining DJ, Gladish GW. Receiver operating characteristic curves: a basic understanding. RadioGraphics 1992;12:1147-54. 9. ten Bosch JJ, Angmar-Mansson B. Characterization and validation of diagnostic methods. Monogr Oral Sci 2000;17:174-89. 10. van Rijkom HM, Verdonschot EH. Factors involved in validity measurements of diagnostic tests for approximal cariesa meta-analysis. Caries Res 1995;29:364-70. 11. Hintze H, Wenzel A. Clinical and laboratory radiographic caries diagnosis: a study of the same teeth. Dentomaxillofac Radiol 1996;25:115-8. 12. Rugg-Gunn AJ. Approximal carious lesions: a comparison of the radiological and clinical appearances. Br Dent J 1972;133:481-4. 13. Downer MC. Concurrent validity of an epidemiological diagnostic system for caries with the histological appearance of extracted teeth as validation criterion. Caries Res 1975;9:231-46. 14. Mejare I, Grondahl HG, Carlstedt K, Grever AC, Ottosson E. Accuracy at radiography and probing for the diagnosis of proximal caries. Scand J Dent Res 1985;93:178-84. 15. Pitts NB, Rimmer PA. An in vivo comparison of radiographic and directly assessed clinical caries status of posterior approximal surfaces in primary and permanent teeth. Caries Res 1992;26:146-52. 16. Hintze H, Wenzel A, Danielsen B, Nyvad B. Reliability of visual examination, fibre-optic transillumination, and bite-wing radiography, and reproducibility of direct visual examination following tooth separation for the identification of cavitated carious lesions in contacting approximal surfaces. Caries Res 1998;32:204-9.

October 2001

Journal of Dental Education

989

17. Espelid I, Tveit AB. Clinical and radiographic assessment of approximal carious lesions. Acta Odontol Scand 1986;44:31-7. 18. Mileman PA, van der Weele LT. Accuracy in radiographic diagnosis: Dutch practitioners and dental caries. J Dent 1990;18:130-6. 19. Verdonschot EH, van de Rijke JW, Brouwer W, ten Bosch JJ, Truin GJ. Optical quantitation and radiographic diagnosis of incipient approximal caries lesions. Caries Res 1991;25:359-64. 20. Russell M, Pitts NB. Radiovisiographic diagnosis of dental caries: initial comparison of basic mode videoprints with bitewing radiography. Caries Res 1993;27:65-70. 21. Ricketts DN, Whaites EJ, Kidd EA, Brown JE, Wilson RF. An evaluation of the diagnostic yield from bitewing radiographs of small approximal and occlusal carious lesions in a low prevalence sample in vitro using different film types and speeds. Br Dent J 1997;182:51-8. 22. Heaven TJ, Firestone AR, Feagin FF. Computer-based image analysis of natural approximal caries on radiographic films. J Dent Res 1992;71(Spec Iss):846-9. 23. Firestone AR, Sema D, Heaven TJ, Weems RA. The effect of a knowledge-based, image analysis and clinical decision support system on observer performance in the diagnosis of approximal caries from radiographic images. Caries Res 1998;32:127-34. 24. Wenzel A, Fejerskov O, Kidd E, Joyston-Bechal S, Groeneveld A. Depth of occlusal caries assessed clinically, by conventional film radiographs, and by digitized, processed radiographs. Caries Res 1990;24:327-33. 25. Wenzel A, Hintze H, Mikkelsen L, Mouyen F. Radiographic detection of occlusal caries in noncavitated teeth: a comparison of conventional film radiographs, digitized film radiographs, and RadioVisioGraphy. Oral Surg Oral Med Oral Pathol 1991;72:621-6. 26. Wenzel A, Fejerskov O. Validity of diagnosis of questionable caries lesions in occlusal surfaces of extracted third molars. Caries Res 1992;26:188-94.

27. Nytun RB, Raadal M, Espelid I. Diagnosis of dentin involvement in occlusal caries based on visual and radiographic examination of the teeth. Scand J Dent Res 1992;100:144-8. 28. Ketley CE, Holt RD. Visual and radiographic diagnosis of occlusal caries in first permanent molars and in second primary molars. Br Dent J 1993;174:364-70. 29. Lussi A. Comparison of different methods for the diagnosis of fissure caries without cavitation. Caries Res 1993;27:409-16. 30. Verdonschot EH, Wenzel A, Truin GJ, Konig KG. Performance of electrical resistance measurements adjunct to visual inspection in the early diagnosis of occlusal caries. J Dent 1993;21:332-7. 31. Lussi A, Firestone A, Schoenberg V, Hotz P, Stich H. In vivo diagnosis of fissure caries using a new electrical resistance monitor. Caries Res 1995;29:81-7. 32. Ricketts D, Kidd E, Smith B, Wilson R. Radiographic detection of occlusal caries: effect of X-ray beam factors on diagnosis. Eur J Prosthodont Restor Dent 1994;2:14954. 33. Ekstrand KR, Ricketts DN, Kidd EA. Reproducibility and accuracy of three methods for assessment of demineralization depth of the occlusal surface: an in vitro examination. Caries Res 1997;31:224-31. 34. Huysmans MC, Hintze H, Wenzel A. Effect of exposure time on in vitro caries diagnosis using the Digora system. Eur J Oral Sci 1997;105:15-20. 35. Ashley PF, Blinkhorn AS, Davies RM. Occlusal caries diagnosis: an in vitro histological validation of the Electronic Caries Monitor (ECM) and other methods. J Dent 1998;26:83-8. 36. Huysmans MC, Longbottom C, Pitts N. Electrical methods in occlusal caries diagnosis: an in vitro comparison with visual inspection and bite-wing radiography. Caries Res 1998;32:324-9. 37. Lazarchik DA, Firestone AR, Heaven TJ, Filler SJ, Lussi A. Radiographic ealuation of occlusal caries: effect of training and experience. Caries Res 1995;29:355-8.

990

Journal of Dental Education Volume 65, No. 10