Abstract
When radiography is applied in the clinic for caries detection, the recommended technique is bitewing projection (Gröndahl, 1994). This technique was introduced by Raper (1925) and has continued with only minor alterations. The aims of this report are to provide evidence for (1) optimal bitewing recording for individual examinations and clinical trials, (2) advantages and disadvantages of digital receptors for bitewing examinations, (3) the diagnostic outcomes and limitations of bitewing radiography, and (4) computerautomated detection of caries.
Bitewing Recording with Film or Digital Receptors
The recommendation for a posterior bitewing examination is that it should capture an image of the crowns of the teeth from the distal surface of the canine to the distal surface of the most posterior erupted molar (see textbooks, Mauriello et al., 1995; White and Pharoah, 2000). To cover this area in an adult patient with erupted third molars, the clinician must use 2 size #2 dental films or 1 long bitewing film (size #3) in each side of the mouth. The advantage of 1 long film over 2 conventional films is that the patient receives less radiation. However, significantly more overlapping surfaces, cone cuts, and missing surfaces have been found in bitewing examinations undertaken with 2 long films compared with examinations with 4 conventional films (Kaffe et al., 1984; Jensen et al., 1987). The recommendation dates from a period when caries was common in canines and first premolars. It may be argued that the four-film examination provides an additional view of most surfaces, which potentially increases the likelihood of a lesion being detected. In a recent study, however, only small differences were found between the prevalence of caries (enamel and dentin) in unrestored occlusal and approximal surfaces when the examination was performed with only 1 size #2 bitewing compared with the recommended 2 bitewings in each side of the mouth. For dentinal caries, the prevalence was 3.2% and 3.6%, respectively, and very little disease was seen in canines and first premolars (Hintze and Wenzel, 1999).
For optimal image quality to be obtained, a bitewing examination should involve an aiming device, and for exposure to be kept “as low as reasonably achievable”, caries risk assessment should precede radiographic examination. In individuals—and in clinical trials in populations with low caries prevalence—one exposure in each side of the mouth, covering the most posterior teeth, may suffice.
Projection errors and retakes should be thoroughly considered when recent years’ digital radiographic receptors are brought into use in the dental clinic (Wenzel, 1999). There are two basically different concepts for digital bitewing radiography: solid-state sensors (Charge-Coupled Device [CCD] and Complementary Metal Oxide Semiconductor [CMOS] technology), where a cord connects the receptor and the computer, and storage phosphors, which appear like film and must be processed (scanned) after exposure (for reviews, see Wenzel and Gröndahl, 1995; Wenzel, 1998, 1999). The solid-state sensors are sub-optimal for bitewing examination in the clinic, since not only is the effective radiation field smaller than a size #2 film, but also the problem lies in their thickness and the cable that must come out of the subject's mouth. General practitioners in Denmark who work with these sensors stated that they often replaced a traditional bitewing examination with from 4 to 8 periapical exposures (Wenzel and Gotfredsen, 2000). Film was still in use significantly more for bitewing examinations by Norwegian general practitioners who worked with sensors (27%) than by those who worked with storage phosphor systems (4%) (Wenzel and Møystad, 2001a, b). The phosphor plates are available in sizes similar to film. The holders available for bitewing examinations with phosphor plates appear like those for film, while the sensor holders still leave much to be desired.
A recent study using two sensor and two phosphor plate systems for bitewing examinations showed more positioning errors with the sensors (Bahrami et al., 2003). A previous study of technical errors in periapical radiography reported more retakes with a CCD sensor (28%) than with film (6%) (Versteeg et al., 1998). Thinner, cordless sensors with larger effective image areas may be available in the future, and for field studies of population groups at various sites, a sensor system connected to a portable computer might be practicable, since the image is displayed immediately after exposure and no processing has to be performed, either in wet chemicals as for film, or through scanning as for storage phosphors. Other benefits of the digital systems include reductions in radiation dose and time savings (Wenzel and Møystad, 2001b). Since the digital systems are based on re-usable receptors, the possibility for cross-contamination should be considered.
A microbiological study on contamination of the CCD sensor and the storage phosphor plate in connection with bitewing examinations demonstrated that total counts of cultivable bacteria were low (< 20), the majority being catalase-positive, Gram-positive cocci, and Gram-positive rods (Wenzel et al., 1999). It was concluded that cross-contamination poses a minor problem with both digital techniques when simple hygiene procedures are followed.
Diagnostic Outcomes and Limitations of Bitewing Radiography
In laboratory experiments, sensitivities for caries detection have been fairly high (50-70%) for detection of lesions reaching into the dentin (approximal and occlusal), while the fraction of false-positive recordings ranges from 3 to 30% (for review, see Wenzel, 1995). Detection rates for approximal enamel lesions are considerably lower (Hintze et al., 1994; White and Yoon, 1997). Current digital intra-oral radiography systems and recently developed dental film have been reported to be as accurate as traditional films for the detection of caries (for review, see Wenzel, 1999, 2000), although there are exceptions (Hintze and Wenzel, 2002; Hintze et al., 2002) (Fig. 1), which stresses the need for evidence when a new radiographic system is implemented. Data from laboratory experiments also seem to be reliable in the clinic (Hintze and Wenzel, 1996).
Differences in diagnostic accuracy (Receiver Operating Characteristics areas) for caries detection in approximal surfaces between recent digital sensor, storage phosphor, and film modalities.
In the clinical situation, caries lesions have traditionally been diagnosed by visual inspection in combination with radiography. It is agreed in the literature that radiography is a more sensitive diagnostic method than clinical inspection for detecting both approximal and occlusal lesions in dentin, for estimating depth of the lesion, and for monitoring lesion behavior. How much information radiography adds to a conventional clinical examination has been reviewed in several papers (for reviews, see Kidd and Pitts, 1990; Pitts and Kidd, 1992; Wenzel et al., 1993; Wenzel, 1995; Pitts, 1996), and extensive reviews have keyed the information from bitewing radiographs to treatment strategies and given surveys on selection criteria for radiographic examinations (Pitts, 1996; Verdonschot et al., 1999).
However, there are limits to the radiographic examination which should be considered, particularly since lesion behavior has changed, with cavitation occurring much later than previously (Pitts and Rimmer, 1992), opening the possibility for the demineralized but non-cavitated lesion to be arrested. With only one measurement point, radiography cannot distinguish between an active and an arrested lesion. A demineralized “scar” may often be left in the hard tissues after lesion progression has been arrested, since fluoride, as a marker of remineralization, only rarely diffuses into the body of the lesion, even if it maintains an intact outer surface (Pearce et al., 1995). Moreover, radiography is not convincingly able to distinguish between cavitated and noncavitated surfaces (Nielsen et al., 1996), and the relationship between the depth of the radiographic lesion and clinical cavitation is not unequivocal. In low-prevalence populations, only a few percent of radiographic, approximal enamel lesions are cavitated. For radiographic dentinal lesions, the fraction for surfaces with cavitation has been reported to range between 50 and 90% (Ratledge et al., 2001). Radiographically deep dentinal lesions, however, are very likely to be cavitated. Lesions in occlusal surfaces, with a clinically visible breakdown of even a minor part of the surface, most often penetrate deep into dentin (Wenzel and Fejerskov, 1992), and such surfaces hardly need radiographic examination. Clinically “sound” and apparently intact occlusal surfaces, however, may develop lesions which penetrate into the dentin, sometimes named “hidden” caries (Ricketts et al., 1997). Radiography may reveal a high fraction of dentinal occlusal lesions in intact occlusal surfaces with clinical signs of caries activity (Hintze, 1993; Hintze and Wenzel, 1994). There is no evidence, however, that “automatic” radiographic screening will benefit populations with low caries experience (for review, see Pitts, 1996). In fact, such screening procedures may do more harm than good, particularly in low-prevalence populations, since the predictive value of the positive test decreases significantly with decreasing disease prevalence, leading to overdiagnosis (White and Yoon, 1997).
Since cavitation should be the ultimate endpoint before operative treatment is initiated, optimally the state of the surface should be determined before treatment planning occurs. Temporary tooth separation (Pitts and Longbottom, 1987) offers the clinician the possibility of determining whether the lesion is active or inactive, cavitated or non-cavitated, with some degree of accuracy (Hintze et al., 1998) and may be suitable in low-caries, highly motivated individuals. In a recent longitudinal study on dental students using temporary tooth separation, approximately 20% of approximal lesions with an intact surface, which were in the outer dentin radiographically, developed cavitation during the first 1½-year period, indicating that not all dentinal lesions may be arrested, even among motivated individuals (Hintze et al., 1999). Temporary tooth separation is not an applicable tool in all patients or clinical trials, and monitoring lesion behavior may still rely on radiography, with its inaccuracies and observer variations taken into account.
New Diagnostic Methods Facilitated by Digital Radiography
When 2 radiographs are recorded with at least partly controlled projection angles, the information from the most recent one may be digitally subtracted from that from the former. Optimally, all unchanged anatomical background structures will cancel, and unchanged areas will be displayed in a neutral grey shade in the subtraction image (for review, see Wenzel, 1991). Areas with mineral loss are conventionally displayed in darker shades of grey, while areas of gain appear lighter than the background. For the detection of caries, subtraction has been used in laboratory studies (Wenzel and Halse, 1992; Halse et al., 1994), but the technique may also be applicable for monitoring caries in the clinic. A recent prospective study, evaluating lesion behavior over time (Fig. 2), found higher inter-observer agreement with assessment of subtraction images than in comparison of 2 bitewing films (Wenzel et al., 2000). A subtraction image is still a two-dimensional display of three-dimensional structures, and recent years’ developments of the tomosynthesis principle, now known as tuned-aperture computed tomography, or TACT™, have demonstrated that the technology's ability to display structures in three dimensions increases the accuracy of detecting dental pathology, including caries lesions (Webber et al., 1997).
Subtraction image between two digital bitewing radiographs taken in general dental practice two years apart in the same individual. Due to a non-optimal standardization of the recordings, the teeth in each jaw are subtracted separately, here the upper jaw. The box illustrates a new filling, the white circle a new deep approximal caries lesion, and the black circle progression of a former enamel lesion into the dentin.
As emphasized by Rossmann (1969), “the recording of the image and the visual detection and interpretation of it form essential links in the overall chain of operations constituting the radiologic process. Each of these operations introduces parameters that can have a profound effect on the transmission of diagnostic information.” The interpreter more than the image receptor may be the limiting factor in this diagnostic imaging chain. Therefore, automated caries detection systems with high reproducibility and accuracy would be welcomed in clinicians’ search for more objective caries diagnostic methods. Computer-aided image analysis for detecting caries in digital radiographs has been explored since the 1980s. Early systems were promising (Pitts, 1984b), and more recent systems have been demonstrated to perform, on average, as well as human observers (Heaven et al., 1992; Firestone et al., 1998). None of the systems developed in universities, however, has resulted in a product available for general dentists. A program for automated caries detection has recently been marketed together with a digital sensor system (Trophy Radiologie Inc., Croissy-Beaubourg, France). It has been concluded, from studies on this program, that it lacked accuracy and consistency, since it provided different opinions on the caries status in a surface (Wenzel, 2001; Wenzel et al., 2002). Thus, an accurate, objective method for radiographic caries diagnosis is still not at hand for the clinician.
Conclusions
Bitewing radiography may still be used in individual caries diagnosis and clinical trials for detecting demineralization caused by microbial activity on the tooth surface: the caries lesion. Radiography plays a part in monitoring lesion development over time, since remineralization procedures can arrest or reverse progression of the lesion and lead to changes in mineral quality and quantity. The evidence for dental caries is not restricted to the levels of surface cavitation, and radiography can add information about many of the clinical stages of the caries process at approximal surfaces and the more advanced stages on occlusal surfaces. The most appropriate technique, low-dose receptors, and selection criteria must be considered. Subtraction radiography and the TACT technique may be promising developments in radiographic caries diagnostics and are obvious objects for further research studies.