|Year : 2010 | Volume
| Issue : 3 | Page : 143-152
The effect of exposure reduction on the diagnosis of caries: An ex vivo comparison of film and a CMOS digital imaging system
James R Geist1, Ashok Balasundaram1, Shin-Mey Rose Y Geist1, Vijay Parashar2
1 Department of Biomedical and Diagnostic Sciences, University of Detroit Mercy School of Dentistry, Detroit, Michigan, USA
2 Department of Clinical Dentistry, Midwestern University College of Dental Medicine, Glendale, Arizona, USA
|Date of Web Publication||21-Apr-2012|
James R Geist
University of Detroit Mercy School of Dentistry, 2700 Martin Luther King, Jr, Blvd, Detroit, MI 48208-2576
Source of Support: This project was funded by Delta Dental Fund of Michigan grant number 260694, Conflict of Interest: None
| Abstract|| |
Objectives: To determine the effect on caries diagnosis of exposure reduction on intraoral digital radiographs compared with optimally exposed film images. Materials and Methods: F-speed film radiographs of 61 extracted molars and premolars were made with optimal exposure parameters. The teeth were radiographed using a complementary metal oxide semiconductor (CMOS) digital system with exposures equal to 50%, 30%, 20%, 10%, and 5% of the film exposure. Five observers, who were permitted to adjust brightness and contrast on the digital images, scored the proximal and occlusal surfaces for the presence of caries using a 5-point confidence scale. Areas under receiver operating characteristic (ROC) curves represented accuracy of caries detection. Sensitivity and specificity were also calculated. The significance level was P = 0.05. Results: All digital images resulted in lower diagnostic accuracy than film images for all lesions (P ≤ 0.036) and for caries in enamel only (P ≤ 0.030). With dentin caries, there were no significant differences between film and any digital radiographs (P ≥ 0.065) except the 5% exposures (P ≤ 0.034). Digital radiographs of 5% of the exposure of film were significantly poorer than all other exposure categories (P ≤ 0.014) except for the 10% exposures for accuracy for all lesion sizes and for dentin lesions only. Exposures at 10% and 20% resulted in lower sensitivity scores for enamel caries, while 50% exposures were associated with the poorest specificity. Conclusions: Exposures of 30% of optimal F-speed film exposure settings appear to balance acceptable levels of accuracy, sensitivity, and specificity for caries detection. Observer-controlled enhancements were ineffective at extremely high and low exposures.
Keywords: Dental caries, dental digital radiography, diagnosis, oral, radiography, dental, receiver operating characteristic curve, sensitivity and specificity
|How to cite this article:|
Geist JR, Balasundaram A, Geist SMY, Parashar V. The effect of exposure reduction on the diagnosis of caries: An ex vivo comparison of film and a CMOS digital imaging system. J Int Clin Dent Res Organ 2010;2:143-52
|How to cite this URL:|
Geist JR, Balasundaram A, Geist SMY, Parashar V. The effect of exposure reduction on the diagnosis of caries: An ex vivo comparison of film and a CMOS digital imaging system. J Int Clin Dent Res Organ [serial online] 2010 [cited 2018 Nov 15];2:143-52. Available from: http://www.jicdro.org/text.asp?2010/2/3/143/95288
| Introduction|| |
Digital radiography (DR) has gained acceptance as a replacement for film radiography in the detection of dental caries. It provides diagnostic information comparable to film radiographs, generally with lower radiation exposure to the patient. ,,,,,,,,,,,,,,,,,, Digital systems using solid-state charge coupled device (CCD) or complementary metal oxide semiconductor-activated pixel sensors (CMOS-APS) transfer the exposure data directly to the computer, producing an image on the screen in a few seconds. Photostimulable phosphor (PSP) sensors must be scanned with a laser beam to send the data into the computer, which results in a slight delay in visualizing the radiograph. Most investigations of solid-state systems have subjectively chosen radiation doses ranging from 10 to 50% of the exposures needed for film radiographs, ,,,,,,,,,,,, although some have used exposures equivalent to those of film radiography. ,,,,
However, there has been little research providing objective evidence regarding the lowest limit of radiation exposure that is compatible with an accurate diagnosis of caries. In DR systems, operators have the option of altering brightness and contrast subjectively to enhance under- or overexposed radiographs to detect caries, which is not possible with film radiographs. In addition, many DR systems include software that can be activated automatically to correct the density and contrast of the acquired image if it is suboptimally exposed, although the benefits of this enhancement have not been extensively tested. Compared to PSP sensors, CCD and CMOS sensors have narrower latitude or range of exposures that can produce an acceptable radiograph. , This suggests that it may not be possible to decrease exposures with some CCD or CMOS systems without degrading the image to the point where it is no longer diagnostic even with the assistance of brightness and contrast enhancement.
This ex vivo investigation was conducted to measure the diagnostic accuracy of caries detection at decreasing levels of exposure with a CMOS DR system and compare it with optimally exposed radiographic film images. The goal was to identify the exposure level at which the diagnostic ability of observers is significantly diminished with respect to film exposure and other digital exposure times.
| Materials and Methods|| |
A total of 61 extracted maxillary and mandibular molar and premolar teeth were used in this investigation. They had no cavitated carious lesions or restorations and were free of obvious developmental and acquired defects. The university's Institutional Review Board (IRB) granted exempt status to this project.
The teeth were mounted in plaster blocks in groups of three or four per block and arranged to produce proximal surface contacts simulating the usual proximal contact position in the jaws. The plaster blocks were positioned in a device that permitted reproducible alignment of the X-ray beam, block of teeth, and film or sensor, with no overlapping of the proximal contact points. The device also included a 2.5-cm Plexiglass plate between the position-indicating device of the X-ray beam and the blocks of teeth to simulate soft tissue scatter and absorption of radiation.
A GX-770 X-ray unit (Gendex Corp., Milwaukee, WI, USA) operating at 70 kVp and 7 mA was used to expose the radiographs. Exposures were made with a rectangular position-indicating device producing a 45-cm source-to-image distance.
For each block, radiographs were made with Insight film (Eastman Kodak Co., Rochester, NY, USA) using a range of exposure times. The radiation dose for each exposure was measured with a Mult-O-Meter Model 512 dosimeter (Unfors Instruments, Billdal, Sweden) that had been factory calibrated within the previous 6 months. The films were processed in an AT 2000 automatic processor (Air Techniques, Inc., Hicksville, NY, USA) operating at 82°F in a 5.5-minute cycle and using fresh Ready Matic developer and fixer (Eastman Kodak).
The optical density in the coronal dentin of the teeth in each processed film was measured with a Digital Densitometer II Model 07-424 (Nuclear Associates, Carle Place, NY, USA) using a 1-mm circular aperture. The densitometer had been factory calibrated 6 months earlier. The radiograph of each block that had an optical density between 0.95 and 1.05 in the coronal dentin was selected for inclusion in the project, as described previously. ,,, Research indicates that maximum detectability of caries occurs at an optical density of approximately 1.0.  The radiation dose required for the optimal film exposure in all blocks was 2287 μGy.
Following selection of the optimal film radiographs, each block was radiographed five times with a CMOS-APS sensor (CDR, Schick Technologies, Long Island City, NY, USA). Radiographs were made with exposure times to produce exposures of 50%, 30%, 20%, 10%, and 5% of the dose needed to produce the optimal image on Insight film. The PerfectShot® automatic exposure control algorithm, which is optional on the Schick CDR system, was not activated.
The digital radiographs were assigned random numbers and stored on the hard drive of the CDR system. The proprietary CDR Schick software accompanying the CMOS sensor was used for image capture. All images were saved in a tagged image file format and stored in a compact disc storage medium along with the viewer interface software (Image J, National Institutes of Health, Bethesda, MD, USA) to enable the observers to view the digital images. The film images were randomly numbered and placed in an opaque mount for viewing.
Five dentists, with experience ranging from 5 to 35 years in clinical practice, served as observers. Three observers were oral and maxillofacial radiologists, one had graduate education and extensive experience in oral medicine and oral diagnosis, and one was a general dentist. All images were presented in a random order for viewing. The observers examined the radiographs individually and were not placed under any time limitation. Before beginning their evaluation, all observers participated in a calibration session to demonstrate the use of the viewer interface software for examining the digital radiographs.
The DR images were shown on a 17″ flat panel display monitor with a resolution of 1024 × 768 and an 8-bit gray scale. The observers were allowed to adjust only the brightness and contrast of the images as displayed through the Image J software. The film radiographs were evaluated by observers on a light box. A hand-held 2× magnifying lens was provided to view the film images. All digital and film radiographs were viewed under dimmed lighting conditions.
The observers were asked to examine the mesial, occlusal, and distal surfaces of the molars and premolars. For each radiograph, the observers indicated their confidence in the presence of a carious lesion in each tooth surface with a 5-point scale: 1 = caries definitely not present; 2 = caries probably not present; 3 = uncertain if caries is present; 4 = caries probably present; and 5 = caries definitely present. Instructions were given to use the full range of the scale during the observations, as described previously.  The observers evaluated the digital images in four sessions to avoid fatigue, with each session separated by at least 1 week. The film radiographs were evaluated in one session.
The teeth were cut with a diamond saw (Mager Scientific, Inc., Dexter, MI, USA) in sections of approximately 500 μm in the mesio-distal dimension. Two investigators used a dissecting microscope to evaluate each surface of the sections for the presence and depth of caries. In cases of disagreement on the status of a surface, a forced consensus was obtained between the investigators. Caries depth was classified as: E1 = caries in outer half of enamel, E2 = caries in inner half of enamel, D1 = caries in outer half of dentin, and D2 = caries in inner half of dentin. This examination provided the ground truth for analysis of the observers' responses.
While sectioning the teeth, three surfaces were destroyed, resulting in a total of 180 surfaces (121 proximal, 59 occlusal) that were evaluated histologically. [Table 1] summarizes the status of caries in the teeth.
Receiver operating characteristic (ROC) analysis with the maximum likelihood technique, corrected by bootstrap sampling of data for the use of the same cases in each imaging modality and multiple observers, was used to determine the diagnostic accuracy of each observer. The areas under the ROC curves (A z ) were used as measures of diagnostic accuracy. , This area can be thought of as the probability that an observer looking at a pair of radiographs, one with a carious lesion and one without, will correctly identify the caries. Sensitivity and specificity were also calculated using bootstrap sampling. Due to the relatively small numbers of carious surfaces, E1 and E2 lesions were combined as enamel caries and D1 and D2 lesions were combined as dentin caries for statistical evaluation. The null hypothesis stated that there was no difference in A z or the other diagnostic measures between any of the six exposure conditions. The significance level was established a priori at P = 0.05.
| Results|| |
The mean values of the ROC A z and standard deviations for all lesions are listed as multiple pair-wise comparisons in [Table 2], along with 95% confidence intervals (CI) and P values for differences between each pair of exposure conditions. The values for all digital images were significantly lower than for film exposures (P ≤ 0.36). Digital radiographs made at 5% exposure had significantly lower A z values than all other digital exposure times (P ≤ 0.014) except at 10% (P = 0.078).
|Table 2: Average ROC Az values, standard deviations, and 95% confidence intervals for the film exposure and each of the five sets of digital images for all lesions|
Click here to view
When considering enamel lesions only [Table 3], A z values of all digital images were significantly poorer than film radiographs (P ≤ 0.030), while none of the digital exposures differed significantly from the others (P ≥ 0.078). ROC calculations for dentin caries [Table 4] revealed that the 5% exposure was significantly worse than all other exposure categories (P ≤ 0.034) except the 10% exposure (P = 0.294). There were no significant differences between any of the other exposure conditions (P ≥ 0.065). The A z values for dentin caries were significantly better than the corresponding values for enamel lesions for each digital and film exposure condition (P ≤ 0.050).
|Table 3: Average ROC Az values, standard deviations, and 95% confidence intervals for the film exposure and each of the five sets of digital images for enamel lesions only|
Click here to view
|Table 4: Average ROC Az values, standard deviations, and 95% confidence intervals for the film exposure and each of the five sets of digital images for dentin lesions only|
Click here to view
The mean values for sensitivity, along with the statistical analysis, are listed in [Table 5]. Sensitivity is the percentage of carious surfaces correctly identified as carious. Images exposed at 5% of the film dose performed significantly worse than film and all other digital images (P ≤ 0.039). The 10%, 20%, and 30% digital radiographs were significantly lower in sensitivity (P ≤ 0.050) than the 50% digital images and film, which were not significantly different amongst each other (P = 0.876).
|Table 5: Overall sensitivity values, standard deviations, and 95% confidence intervals for the film exposure and each of the five sets of digital images for all lesions|
Click here to view
[Table 6] lists the mean values for sensitivity limited to enamel lesions, with statistical measurements. The 5% exposures were significantly lower in mean values than film, 50%, and 30% images (P ≤ 0.013). The 10% and 20% radiographs were worse than film (P ≤ 0.013), and the 20% exposures were lower than 50% images (P = 0.020), while no significant differences were detected in the mean values between 30%, 50% and film exposures (P ≥ 0.056).
|Table 6: Enamel sensitivity values, standard deviations, and 95% confidence intervals for the film exposure and each of the five sets of digital images for all lesions|
Click here to view
Overall specificity is represented in [Table 7]. Specificity is the percentage of sound surfaces correctly identified as non-carious. Digital images exposed at 50% of the film exposure dose had significantly poorer specificity (0.79) than all other imaging systems (P ≤ 0.014). The specificity of the 30% exposure (0.85) was slightly poorer than film (P = 0.050).
|Table 7: Overall specificity values, standard deviations, and 95% confidence intervals for the film exposure and each of the five sets of digital images for all lesions|
Click here to view
The intra-class correlation coefficient ranged from 0.17 (poor agreement) for the 5% exposures to 0.41 (moderate agreement) for film.
| Discussion|| |
Radiographic diagnosis of dental caries involves a risk/benefit calculation: the risk of radiation exposure to patients must be balanced against the benefits of diagnosis of caries. While it is desirable to keep radiation exposure to the patient as low as reasonably achievable, it must not be set below a threshold that compromises the ability to detect carious lesions. Digital images exposed with 5% of the exposure dose for Insight film radiographs performed poorly with respect to measures of accuracy and sensitivity. The A z values for all lesions and for caries in dentin were significantly lower for the 5% exposure than all other exposure conditions except for the 10% exposure time, and sensitivity for all carious lesions was significantly poorer at 5% exposure than all other digital and film exposures. The null hypothesis for these parameters can be rejected for the 5% exposure radiographs. The low scores are related to the fact that as exposure dose decreases, the degree of noise caused by quantum mottle, or variation in quantity of X-ray exposure, increases. ,,, This might reduce the signal/noise ratio to the point where the signal (depiction of an abnormality) cannot be detected even with enhancement of the digital image, leading to misdiagnoses. It was not surprising that sensitivity, which represents the percentage of actually carious surfaces correctly identified as caries, was therefore significantly lower in the 5% exposures than all other imaging conditions.
Sensitivity of diagnosis of lesions limited to enamel was generally poor for the 5%, 10%, and 20% radiographs, while the 30%, 50%, and film images were significantly better. Exposures as low as 6% of the dose needed for radiographs on E-speed film have been found to permit accurate diagnosis of caries when using PSP plates.  However, the exposure latitude, or range of exposure doses over which acceptable images can be produced, is narrower for CCD and CMOS sensors than for PSP sensors , or film.  This suggests that it may not be possible to decrease exposures as significantly with some CCD or CMOS systems without reducing the signal so severely that enhancement cannot strengthen it.
While sensitivity and accuracy are important in detecting carious lesions that are actually present, an imaging system must also enable observers to correctly identify sound surfaces as caries free. Sensitivity, therefore, must be balanced against specificity. It is known that the ability to visualize caries is enhanced with greater exposure because contrast improves with increasing darkness and caries detection is dependent on radiographic darkness and contrast. ,,,, However, the number of false-positive decisions also increases with darker images, so specificity decreases as sensitivity improves. , In this research, most of the 50% images were notably darker than the other radiographs, including film. Sensitivity at 50% exposure was comparable to film and significantly better than the 5%, 10%, and 20% exposures for both enamel and dentin caries. But digital radiographs exposed at 50% of the film dose had significantly lower specificity than all other conditions.
It appears that the use of brightness and contrast controls was not effective in improving the diagnostic accuracy at the extremes of exposure. The significantly underexposed digital radiographs, with their low signal/noise ratio, could not be enhanced to the point where diagnosis of enamel caries could be comparable to images with greater exposure. We speculate that if these radiographs were darkened or given greater contrast, the image of small carious lesions near the enamel surface might have been lost, yielding poor accuracy and sensitivity scores for the 5% exposure series. Similarly, the excessive darkness of many of the 50% exposure radiographs could not be brightened sufficiently during the electronic transformation of gray levels that occurs during enhancement. The tendency toward misinterpreting sound surfaces as carious in dark radiographs could not be counteracted. We believe that this may partly explain the low specificity levels in the 50% exposure images.
The ineffectiveness of observer-controlled modification of the electronic image is a common finding. ,,,,,, Some investigators have proposed that ad libitum adjustments of brightness and contrast may actually diminish the accuracy of diagnosis. ,,,, Alteration of contrast may degrade the image by accentuating noise, leading to more false-positive diagnoses. 
However, many digital systems come equipped with an automatic exposure compensation (AEC) algorithm that manipulates the acquired exposure data to optimize the image as it appears on the screen. ,, Compensation is usually performed by remapping the gray range into an 8-bit scale and applying it over the range of recorded exposures. In this way, under- and overexposed areas are represented by more gray shades than in the unaltered image. AEC has been shown to minimize the decrease in diagnostic accuracy that occurs at excessively low and high exposure levels when using a solid-state sensor due to its narrow latitude.  The AEC program on the Schick CDR system was not activated in our project in order to directly evaluate the effect of very low exposures on diagnosis. Considering the disappointing results in caries diagnosis at the extremes of exposure, it would appear that AEC should be used when available.
In addition, caries-specific enhancement algorithms have been researched on radiographs captured with PSP sensors. ,,,, Research indicates that some caries-enhancement programs are effective in improving the accuracy of caries diagnosis or in decreasing the variance in diagnostic scores of multiple observers. ,,, High-pass filters may preserve the image at the edge of a tooth surface, thereby decreasing the chance that incipient enamel caries may become invisible.  They may also compensate for the diminished image quality resulting from poor signal/noise ratios. However, some caries-specific programs have been found to have no effect on diagnosis  or to significantly reduce diagnostic accuracy for small lesions when compared to the original 8-bit image through reductions in sensitivity and specificity.  Continued research is needed on the effects of these algorithms in caries diagnosis.
Our results indicated that film radiographs outperformed all digital exposure conditions for enamel caries. Possible explanations include the following.
Familiarity with film
At the time this research was conducted, three of the observers had had limited experience with intraoral DR. Lack of familiarity with this modality has been attributed to poor observer performance in other research of similar nature in the past. [3, 7, 12]
In many of the investigations that reported equivalent performance of digital and film radiography in caries detection, the exposure time for the digital images was not reduced to as great an extent as in our investigation relative to the time needed for optimal film radiographs. ,,, This would have the effect of not decreasing the signal/noise ratio of the images as significantly and would be expected to elevate the accuracy scores of observers when looking for caries. But this defeats one of the advantages of digital systems: the ability to reduce patients' radiation burden.
Radiography in general is largely unsuccessful in visualizing many carious lesions limited to enamel. ,,,,, The author of a review of dozens of studies concluded that radiography has very little value in diagnosing enamel caries, and digital systems are poor in this task.  Correspondingly, observers' confidence in diagnosis for enamel decay was reported to be lower than their confidence in diagnosis of dentin caries in one investigation.  Although our results indicated that digital systems differed from film images in depicting caries, the accuracy and sensitivity of the film radiographs were also fairly low. In our research, the large majority of carious lesions were limited to enamel, unlike many other studies in which dentinal decay was much more prevalent. ,,,,,,,, It might be expected that visualization of enamel caries would pose a challenge for all imaging systems, film as well as digital. This might explain the poor inter-observer agreement as measured in the intra-class correlation coefficient in our project.
Some published reports indicate that A z values in ROC analysis are significantly higher for dentin caries than for enamel lesions, similar to our findings. ,, When examining the results of the diagnosis of dentinal caries, we found that only the digital images at 5% exposure were significantly poorer than all other systems, including film, for which there were no significant differences. This might suggest that there is no difference between film radiographs and digital images with sufficient exposure in tasks for which radiography can be realistically expected to provide diagnostic information.
As mentioned above, subjective modification of the image by observers has been shown to be ineffective. Some researchers have suggested that such enhancement results in lower accuracy for digital images than for film, especially when AEC has been applied. ,,
Although many reports indicate that digital radiographs are not significantly different from film images in caries detection, some investigators have found certain digital systems, both solid state and storage phosphor, to be outperformed by film-based radiographs, ,,, especially at very low exposures.  This may limit the ability to generalize these findings to all solid-state sensor systems.
| Conclusion|| |
An exposure time of 5% of the dose of an optimally exposed F-speed film radiograph resulted in A z scores overall and for lesions extending into dentin that were significantly worse than digital images with greater exposures and film-based radiographs. Exposures of 5%, 10%, and 20% produced significantly poorer sensitivity values than digital radiographs at higher exposures and film images. Specificity was significantly lower in the 50% digital radiographs, indicating the highest of risk of false positivity. Observer-controlled enhancement of brightness and contrast appear to be ineffective at the extremes of exposure times. These findings suggest that an acceptable compromise between sensitivity and specificity, with accuracy comparable to film radiographs for dentinal caries, would occur at exposure settings of approximately 30% of the dose required for a properly exposed radiograph using an F-speed film.
| Acknowledgments|| |
This project was funded by Delta Dental Fund of Michigan grant number 260694. None of the authors has a conflict of interest. The authors gratefully acknowledge Drs George Eckert and Rebeka Tabbey for statistical evaluation of the data.
| References|| |
|1.||Wenzel A. Digital radiography and caries diagnosis. Dentomaxillofac Radiol 1998;27:3-11. |
|2.||Abreu M Jr, Mol A, Ludlow JB. Performance of RVGui sensor and Kodak Ektaspeed plus film for proximal caries detection. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2001;91:381-5. |
|3.||Nair MK, Nair UP. An in-vitro evaluation of Kodak Insight and Ektaspeed Plus film with a CMOS detector for natural proximal caries: ROC analysis. Caries Res 2001;35:354-9. |
|4.||Tyndall DA, Ludlow JB, Platin E, Nair M. A comparison of Kodak Ektaspeed film and the Siemens Sidexis digital imaging system for caries detection using receiver operating characteristic analysis. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1998;85:113-8. |
|5.||Hintze H, Wenzel A. Influence of the validation method on diagnostic accuracy for caries. A comparison of six digital and two conventional radiographic systems. Dentomaxillofac Radiol 2002;31:44-9. |
|6.||Hintze H, Wenzel A, Frydenberg M. Accuracy of caries detection with four storage phosphor systems and E-speed radiographs. Dentomaxillofac Radiol 2002;31:170-5. |
|7.||Syriopoulos K, Sanderink GC, Velders XL, van der Stelt PF. Radiographic detection of approximal caries: A comparison of dental films and digital imaging systems. Dentomaxillofac Radiol 2000;29:312-8. |
|8.||Khan EA, Tyndall DA, Ludlow JB, Caplan D. Proximal caries detection: Sirona Sidexis versus Kodak Ektaspeed Plus. Gen Dent 2005;53:43-8. |
|9.||Castro VM, Katz JO, Hardman PK, Glaros AG, Spencer P. In vitro comparison of conventional film and direct digital imaging in the detection of approximal caries. Dentomaxillofac Radiol 2007;36:138-42. |
|10.||Wenzel A. A review of dentists' use of digital radiography and caries diagnosis with digital systems. Dentomaxillofac Radiol 2006;35:307-14. |
|11.||Kaeppler G, Dietz K, Reinert S. Influence of tube potential setting and dose on the visibility of lesions in intraoral radiography. Dentomaxillofac Radiol 2007;36:75-9. |
|12.||Uprichard KK, Potter BJ, Russell CM, Schafer TE, Adair S, Weller RN. Comparison of direct digital and conventional radiography for the detection of proximal surface caries in the mixed dentition. Pediatr Dent 2000;22:9-15. |
|13.||Svanaes DB, Moystad A, Larheim TA. Approximal caries depth assessment with storage phosphor versus film radiography. Caries Res 2000;34:448-53. |
|14.||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. |
|15.||Wenzel A, Borg E, Hintze H, Grondahl HG. Accuracy of caries diagnosis in digital images from charge-coupled device and storage phosphor systems: An in vitro study. Dentomaxillofac Radiol 1995;24:250-4. |
|16.||Yoshiura K, Nakayama E, Shimizu M, Goto TK, Chikui T, Kawazu T, et al. Effects of the automatic exposure compensation on the proximal caries diagnosis. Dentomaxillofac Radiol 2005;34:140-4. |
|17.||Berkhout WE, Beuger DA, Sanderink GC, van der Stelt PF. The dynamic range of digital radiographic systems: Dose reduction or risk of overexposure? Dentomaxillofac Radiol 2004;33:1-5. |
|18.||Moystad A, Svanaes DB, van der Stelt PF, Grondahl HG, Wenzel A, van Ginkel, et al. Comparison of standard and task-specific enhancement of Digora storage phosphor images for approximal caries diagnosis. Dentomaxillofac Radiol 2003;32:390-6. |
|19.||Moystad A, Svanaes DB, Risnes S, Larheim TA, Grondahl HG. Detection of approximal caries with a storage phosphor system. A comparison of enhanced digital images with dental x-ray film. Dentomaxillofac Radiol 1996;25:202-6. |
|20.||Ludlow JB, Mol A. Digital Imaging. In: White SC, Pharoah MJ, editors. Oral Radiology: Principles and Interpretation. 5 th ed. St. Louis: Mosby; 2004. p. 232-3. |
|21.||Svenson B, Lindvall AM, Grondahl HG. A comparison of a new dental x-ray film, Agfa Gevaert Dentus M4, with Kodak Ektaspeed and Ultraspeed dental x-ray films. Dentomaxillofac Radiol 1993;22:7-12. |
|22.||Svenson B, Welander U, Anneroth G, Soderfeldt B. Exposure parameters and their effects on diagnostic accuracy. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1994;78:544-50. |
|23.||Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operator characteristic (ROC) curve. Radiology 1982;143:29-36. |
|24.||Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 1983;148:839-43. |
|25.||Skodje F, Espelid I, Kvile K, Tveit AB. The influence of radiographic exposure factors on the diagnosis of occlusal caries. Dentomaxillofac Radiol 1998;27:75-9. |
|26.||Svenson B, Welander U, Anneroth G, Soderfeldt B. Exposure parameters and their effects on diagnostic accuracy. Oral Surg Oral Med Oral Pathol 1994;78:544-50. |
|27.||Kositbowornchai S, Basiw M, Promwang Y, Moragorn H, Sooksuntisakoonchai N. Accuracy in diagnosing occlusal caries using enhanced digital images. Dentomaxillofac Radiol 2004;33:236-40. |
|28.||Haiter-Neto F, Casanova MS, Frydenberg M, Wenzel A. Task-specific enhancement filters in storage phosphor images from the Vistascan system for detection of proximal caries lesions of known size. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2009;107:116-21. |
|29.||Li G, Sanderink GC, Berkhout WE, Syriopoulos K, van der Stelt PF. Detection of proximal caries in vitro using standard and task-specific enhanced images from a storage phosphor plate system. Caries Res 2007;41:231-4. |
[Table 1], [Table 2], [Table 3], [Table 4], [Table 5], [Table 6], [Table 7]