Over the past few years, systems that can generate radiographic digital images without the need for radiograph film have become available for use in clinical practice and are gaining in popularity among practitioners.^5–7 All digital radiography systems contain a solid-state detector sensitive to X-rays. The system we used in our study contained a charged-coupled device, or CCD, sensor. The CCD is a silicon chip made of a group of photodiodes on top of a semiconductor. The size of the sensor is slightly smaller than a conventional PA radiograph. It is placed in the mouth and connected to a computer via a universal serial bus, or USB, cord. When the sensor is exposed to X-rays, the photodiodes generate electric currents corresponding to the strength of the X-rays hitting the surface in different zones. The semiconductor sends the electric signals to the computer to be processed and analyzed. The computer assigns a digital number (density value/gray level) to each signal. The digital numbers then are grouped together to generate an image on a computer monitor.^5–7

This new technology offers many advantages over conventional radiography. It eliminates the need for film and film developing, and it allows for lower radiation exposure. The generated image is available immediately for evaluation on a computer screen and can be manipulated digitally to enhance viewing—for example, the image could be enlarged to focus on specific regions. In addition, digital tools are available to record electronic measurements and to cut, paste and colorize the image. The images can be easily filed on and retrieved from a hard disk or removable storage medium, or the images can be transferred electronically to third-party carriers.^5–7

Digital radiography eliminates the need for film and film developing, and it allows for lower radiation exposure.

Many studies have evaluated the use of filmless digital radiography in caries detection^8–14 and endodontics.^15–17 A few studies examined the use of digital radiography in evaluating crestal alveolar bone loss,^18–23 and most were performed in vitro.^18–21 Some of the studies used direct digital imaging devices,^18–20 and others used a storage phosphor system^20,21 or digitized conventional radiographs.^22,23 Because of the in vitro nature of most of these studies, the findings may not be clinically applicable.

We conducted a study to compare direct digital and conventional radiographic estimates of alveolar bone levels under normal clinical use.

	MATERIALS, METHODS AND STATISTICS

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The New Jersey Dental School internal review board approved the study. We recruited 25 subjects from the periodontal clinic. To meet the study’s inclusion criteria, all subjects had to be adults between the ages of 18 and 65 years and had to require a full set of conventional radiographs (PA and BW) for diagnostic purposes and treatment planning. All of the subjects had to have a minimum of 15 teeth and be systemically healthy. We did not include any pregnant or nursing women in this study. All of the subjects read and signed an informed consent form. They all received clinical oral and periodontal examinations to rule out any intraoral pathological conditions and determine clinical periodontal status.

Radiographic images. Experienced dental assistants licensed to take radiographs in the state of New Jersey took all of the conventional and digital radiographs. They took all of the conventional radiographs under normal clinical practice conditions without using any special devices. They used a long cone paralleling technique for all PA radiographs, along with a plastic film holder attached to a metal arm with a cone-guiding ring to position the film intraorally. A paper sleeve with biting tab was used for all BW radiographs. They used Kodak Ekta Speed Plus (Kodak, Rochester, N.Y.) film (no longer available) for all conventional images and set radiation exposure time to 28 pulses for maxillary posterior sextants and 25 pulses for all other areas.

Within four weeks of having the conventional radiographs taken, each subject returned to have the digital radiographs taken. Under the same conditions as the conventional radiographs were taken, the dental assistants took a set of digital PA and BW radiographs that matched the conventional radiographs. Instead of radiographic film, however, they used a size 2 digital sensor (Schick Technologies, Long Island City, N.Y.). They used a plastic holder with a metal guiding arm supplied by the manufacturer to position the sensor for all PA images. The dental assistants used a paper sleeve with biting tab used for all BW images. They set radiation exposure time to 14 pulses for maxillary posterior regions and 10 pulses for all other areas.

Figure 1 shows an example of a PA conventional radiograph and a PA digital image from the same subject. Both PA images look similar. Figure 2 shows an example of a BW conventional radiograph and a BW digital image from the same subject. The digital BW image shows a wider opening between maxillary and mandibular teeth than does the conventional BW image. We observed this wider opening on most digital BW images and speculate that it may be related to the USB cord’s being present between the upper and lower teeth.

View larger version (62K): [in this window] [in a new window]   Figure 1. Periapical radiograph examples. A. Conventional. B. Digital.

View larger version (65K): [in this window] [in a new window]   Figure 2. Bitewing radiograph examples. A. Conventional. B. Digital.

One examiner (L.H.) determined all of the conventional radiographic measurements. Under standardized viewing conditions, she used a transparent plastic ruler to measure to the nearest millimeter the distance from the CEJ to the interproximal alveolar crest on both the mesial and distal aspects of each tooth present, excluding third molars. She excluded surfaces with nonidentifiable CEJ due to overlapping or restorations. She determined the crest of the alveolar bone by the point at which the periodontal ligament space ends on the root surface.

The other examiner (K.C.) determined all of the digital radiographic measurements. He viewed the images on the monitor of a personal computer. He made the same measurements on the digital radiographs as were taken on the conventional radiographs. He digitally determined the distance from the CEJ to the interproximal alveolar crest on both the mesial and distal aspects of each viewable tooth using a software program (Schick CDR, Version 2.1, Schick Technologies). The examiner measured all of the images at x100 magnification.

The examiners performed their measurements independently from each other. Each examiner measured the conventional or digital radiographs twice and took two sets of measurements for each subject. The examiners took the second set of measurements without having access to the initial set. The overall percentage agreement of the measurements for first and second readings for the conventional images was 99 percent, statistic of 0.97. The overall percentage agreement of measurements for the first and second readings for the digital images was 92 percent, statistic of 0.78.

Statistical analysis. All data were entered into a personal computer. We combined the initial and second sets of measurements for both the conventional radiographs and digital images to generate an average score for each site measured. For each subject, we calculated the average bone level measurements for each sextant and for all sites.

We assessed the agreement between digital and conventional radiographs two ways. First, we assessed absolute agreement by comparing the mean difference between methods with a paired samples t test. Then we computed Pearson’s correlations to examine the relative agreement between the paired observations. We first computed these statistics separately for BW and PA observations averaged over the whole mouth and then by sextant, except for anterior BW measures, which were not assessed.

To examine how each radiographic method revealed bone level and loss, we assigned bone level measurements for each site to one of the following categories: normal bone level ( 3 mm), early-to-moderate bone loss (4–6 mm) and advanced bone loss ( 7 mm). We computed the statistic to determine the categorical agreement between the two methods. We performed ² analysis to determine the differences in the number of sites assigned to each bone level category, by each radiographic system.

	RESULTS

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Twenty-five subjects were entered in the study. The total number of matched PA sites we measured was 857, and the average number of sites we measured per subject was 34.3 ± 8.0 standard deviation, or SD. The total number of matched BW sites we measured was 315, and the average number of sites we measured per subject was 12.6 ± 6.6 SD.

Table 1 shows the absolute agreement between average measurements of alveolar bone levels for both digital and conventional images by sextant and for the whole mouth, using the paired samples t test. Table 2 shows the relative agreement or correlation between the average measurements of alveolar bone levels for both conventional and digital radiographs by sextant and for the whole mouth.

Given the overall difference between conventional and digital BW radiographs, we wanted to know if this difference was consistent across all sextants of the mouth. Therefore, we computed paired t test for each of the four sextants available. Results showed differences in mandibular, but not maxillary, regions of the mouth. In the mandibular left sextant, conventional images showed mean bone loss of 1.8 mm, while digital images indicated a mean bone loss of 2.1 mm (P = .002). In the mandibular right sextant, conventional images showed a mean bone loss of 1.7 mm, while the digital images showed an average loss of 2.1 mm (P = .002). By contrast, in the maxillary left sextant, bone loss averaged 2.2 mm for conventional images, which was similar to the 2.4 mm mean bone loss for digital images (P = .5). In the maxillary right sextant, bone loss averaged 2.2 mm for conventional images and 2.1 for digital images (P = .5). Thus, the difference between the conventional and digital radiographs was not found in all mouth sextants. Rather, more bone loss was indicated by digital radiographs only in the posterior mandibular region; measures of bone loss in the posterior maxillary region were similar between the two radiographic methods. It seems that digital radiographs impart a constant addition of millimeters to measures taken in the posterior mandibular region.

The correlation between radiographic methods in the maxillary right, maxillary left, mandibular right and mandibular left sextants ranged from r = .67 in the maxillary left sextant to r = .85 in the maxillary right sextant, all P < .001. These data suggest that the two radiographic methods maintain consistent orderings of the observations, regardless of mean regional differences we noted previously.

For the PA radiographs, we noted no consistent difference between radiographic methods. Averaged over the whole mouth, conventional radiographs showed 2.2 mm of bone loss as did digital radiographs (P = .41). A correlation of r = .92 between these measures indicated a strong relative agreement between these measures.

Together, the absence of a mean difference and presence of a strong correlation indicate both relative and absolute agreement. Both methods seem to provide similar measurements.

A different story emerged when we compared radiographic methods by sextant. As shown in Table 1, measurements of the maxilla in conventional radiographs showed greater bone loss than those in digital radiographs, while measurements of the mandible in digital radiographs showed greater bone loss than those in conventional radiographs for the anterior region and no difference in either posterior regions. Thus, the presence of an interaction of radiographic method by jaw effectively reduced to zero the difference by radiographic method averaged over the region, suggesting that there was no difference between radiographic methods. In fact, the maxillary measure of bone loss is larger with conventional radiographs than with digital PA radiographs, while there is either no difference or a difference favoring digital radiographs for assessments of bone level around teeth on the mandible. In contrast to the BW radiographs for which digital radiographs indicated more bone loss in the posterior mandibular sextants, measures from the PA radiographs suggested less bone loss in the digital radiographs of the maxillary teeth. Like the BW data, correlations between radiographic methods ranged from r = .57 for the maxillary left sextant, to r = .91 for the maxillary anterior sextant, again indicating a close correspondence in the relative ordering of cases by the two radiographic methods.

To evaluate how each radiographic method revealed bone level and loss, we assigned measurements for each site to one of the following three bone level categories: normal bone level ( 3 mm), early-to-moderate bone loss (4–6 mm) and advanced bone loss ( 7 mm). We created contingency tables by cross-tabulating categorical measures from digital PA and BW radiographs with categorical measures from conventional PA and BW radiographs by number of sites we assigned to each category on the basis of the radiographic method (Table 3 and Table 4). The percentage agreement between categorical measures from conventional and digital radiographs is shown in parentheses.

For PA radiographs, out of the 168 sites we categorized as having early-to-moderate bone loss on the basis of the digital images, we categorized 78 (46.4 percent) as having normal bone levels on the basis of the conventional images. For BW radiographs, out of the 52 sites we categorized as having early-to-moderate bone loss on the digital images, we categorized 36 (69.2 percent) as having normal bone levels on the basis of the conventional images.

On both PA and BW images, we tended to categorize a higher number of sites as having early-to-moderate bone loss on the basis of the digital radiographs than we did on the basis of the conventional radiographs. These data suggest that the small average absolute increase in bone loss measures seen on digital BW radiographs and less consistently on PA radiographs hides a larger clinical diagnostic difference. That is, the small increase in measures of bone loss on digital radiographs results in a larger increase in the number of sites we categorized as diseased—early-to-moderate bone loss or advanced bone loss—since many of these observations were located near the cut point. Therefore, clinical classification between these methods indicate less agreement than would be expected from the comparison of actual measures of bone loss.

	DISCUSSION

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
The results of our study showed that under normal clinical use significant differences existed between bone level measurements on digital and conventional radiographs in several regions in the mouth in both PA and BW images. In addition, both radiographic systems significantly differed on revealing alveolar bone levels in the normal, early-to-moderate bone loss or advanced bone loss categories. Both PA and BW digital images tended to reveal a higher number of sites as having early-to-moderate bone loss than did conventional images. For PA radiographs, we categorized 78 sites as having early-to-moderate bone loss on the basis of the digital images and as having normal bone levels on the basis of the conventional images. For BW radiographs, we categorized 36 sites as having early-to-moderate bone loss on the basis of the digital images and as having normal bone levels on the basis of the basis of the conventional images. The statistics of 0.48 for PA images and 0.37 for BW images indicated poor agreement between digital and conventional radiographs on revealing sites with early-to-moderate bone loss.

Despite the significantly higher measurements on all PA maxillary conventional images, the categorical assignment of bone level as either normal, having early-to-moderate loss or having advanced loss was comparable on the basis of both the digital and the conventional images. On the other hand, the categorical assignments showed that there were a higher number of sites with bone loss on the digital radiographs than on the conventional radiographs in the mandibular anterior sextant in PA images and in both the mandibular right and mandibular left sextants in BW images. In those sextants, the average absolute increase in bone loss measurements resulted in a larger increase in the number of sites we categorized as diseased on the basis of the digital images. These data suggest that the disagreement between the two systems is influenced by the type of image—PA or BW—and the region in the mouth.

The differences we noted between the two imaging systems may be attributed to variations in the size and flexibility of the conventional radiograph film and the sensor used in digital radiography. Conventional radiographic film is larger than the sensor; however, it can flex and bend, making it easier to position in the mouth. Even though the sensor is smaller than the conventional radiographic film, it is difficult to position comfortably in the mouth because of its rigidity. These differences in size and flexibility between the sensor and the conventional radiographic film may have influenced the positions and angles they were in in the different regions of the mouth. In addition, having the USB cord attached to the sensor may have interfered with the subjects’ biting on the paper tabs used for BW images. This may have contributed to jaw shifting on mouth closure and resulted in the differences noted in measurements between digital and conventional BW images in the mandibular posterior sextants.

A different examiner measured each type of radiographic system. To control for interexaminer variability, the examiners were calibrated in both radiographic systems to recognize and agree on what they would consider the position of the alveolar bone crest to be in relation to the CEJ on the root surface. In addition, each examiner was compared with himself or herself. The examiners’ agreement between the first and second measurements was better for the conventional images than for the digital images. The computerized measurements for the digital images were given to the nearest 0.1 mm. The fractional millimeter values computed by the digital software may have contributed to the lower agreement between the first and second sets of measurements. To control for intraexaminer variability, we averaged both sets of measurements on both radiographic systems.

	CONCLUSIONS

TOP ABSTRACT MATERIALS, METHODS AND... RESULTS DISCUSSION CONCLUSIONS REFERENCES

 
Under normal clinical use and without a standardized method of film positioning, the average alveolar bone level measurements varied significantly between conventional radiographs and digital images in multiple regions of the mouth. In addition, the digital images tended to reveal a higher number of sites with early-to-moderate bone loss than did the conventional images. These findings suggest that evaluation of alveolar bone loss using intraoral digital radiographs is not comparable with that of conventional radiographic film under normal clinical use.