Basic ResearchTechnology

A Comparison between Panoramic Digital and Digitized Images to Detect Simulated Periapical Lesions Using Radiographic Subtraction
Sergio Augusto Quevedo Miguens Jr, DDS, MSc,* Elaine Bauer Veeck, DDS, MSc, PhD, Vania Regina Camargo Fontanella, DDS, MSc, PhD, and Nilza Pereira da Costa, DDS, MSc, PhD
Abstract
This study compared digital and digitized panoramic radiographs to detect incipient periapical lesions using the digital subtraction radiography (DSR) technique. Bone defects were created in 77 apical sites of dried human mandibles through the application of 70% perchloric acid. Conventional and digital radiographs were obtained twice at time zero and 2, 4, and 6 hours after application of the acid solution. Conventional radiographs were digitized by using the same attributes than direct digital radiographs. The baseline images were subtracted of the negative control and experimental groups and assessed by an experienced and blinded observer who assigned score 0 (normal) or 1 (mineral loss) to each image. The sensitivity of both modalities was statistically different in relation to experimental times. The percentage of correct responses significantly increased regarding experimental times. The specificity was high both for digitized and digital images. DSR of panoramic images, both digitized and digital, allows for the diagnosis of incipient-simulated periapical lesions with no differences in their performance. (J Endod 2008;34: 15001503)

ince the studies performed by Bender and Seltzer (1, 2), several other studies have been performed using different methodologies to radiographically detect the early stages of periapical lesions (3, 4). Digital subtraction radiography (DSR) has proven to be a sensitive instrument in the detection of changes in tissue mineral content since its introduction in dental care in the 1980s (5), taking into account that a 30% to 40% mineral content loss is needed to be visually identified in conventional radiography (6 11). In vivo and in vitro studies have been conducted to establish the diagnostic accuracy of DSR in the detection of bone changes (12). Most of these studies were performed using digital radiographs, but only a few of them report on the use of DSR in digitized images (1315) and extraoral radiographs (16). The aim of this study was to compare the sensitivity, specificity, and diagnostic accuracy of DSR in digital and digitized panoramic images in the detection of incipient periapical lesions created by the application of acid solution.

Materials and Methods

This study was approved by the Research Ethics Committee at Pontifcia Universidade Catlica do Rio Grande do Sul, no. 06/03450. Ten dried human mandibles were selected for this in vitro experimental study with the inclusion criteria: to have at least one tooth or a root without previous changes in the periapical region. The sample was composed of 77 apical sites of lower molars, which were carefully extracted from their alveoli. The integrity of tooth structure and bone tissue was analyzed, and the teeth were returned to their respective alveoli. The mandibles were placed on an acrylic base to standardize the position of the occlusal plane horizontal to the ground and were then radiographed by using both the conventional and digital methods to obtain panoramic radiographs. Two-centimeter thick wax plates were interposed between the radiation source and the mandible in order to simulate soft tissue absorption and scatter of x-ray (17). Conventional radiographs were obtained using a T-Mat G/RA film (Kodak, So Jos dos Campos, SP, Brazil) in a Planmeca Proline CC-Planmeca (Helsinki, Finland) panoramic x-ray unit; exposure settings of 60 kVp and 4 mA and constant attenuation through the placement of two screen layers interposed to the x-ray output were used. The films were processed in new RPX-Omat solutions (Kodak) at a 4.5-minute dryto-dry time cycle and temperature at 28 C (A/T 2000 XR, Hicksville, NY). Conventional radiographs were digitized using an Epson (Long Beach, CA) Perfection 2450 table scanner equipped with a transparency reader and a black acrylic mask. Images of the same mandible were sequentially captured. Automatic settings for brightness and contrast were used in the capture program for the first image and were then maintained for all other images. All images were captured in their original size, 8-bit mode, and 150-dpi scanning resolution. A noncompressed TIFF format was used for disk storage. Direct digital images were obtained using the DenOptix system (Gendex, DesPlaines, IL) using the same device and following the same standardization procedures applied to the conventional radiographs. All images were stored as TIFF, 8-bit mode, 150-dpi resolution using the software VixWin (Gendex).

Key Words
Panoramic radiography, periapical lesion, subtraction technique

From the *Graduate Program in Clinical Stomatology and Department of Radiology, School of Dentistry, Pontifcia Universidade Catlica do Rio Grande do Sul, Porto Alegre, RS, Brazil; and Department of Diagnosis, School of Dentistry, Universidade Luterana do Brasil, Canoas, RS, Brazil. Address requests for reprints to Dr Vania Fontanella, Rua Coronel Paulino Teixeira, 169/403, 90420-160 Porto Alegre, RS, Brazil. E-mail address: vaniafontanella@terra.com.br. 0099-2399/$0 - see front matter Copyright 2008 American Association of Endodontists. doi:10.1016/j.joen.2008.09.006

1500

Miguens et al.

JOE Volume 34, Number 12, December 2008

Basic ResearchTechnology
brightness. There was no time restriction for the assessment of each image. For further assessment of observer reproducibility, 59 images of each modality were digitally reproduced and randomly inserted into the presentation. A total of 430 subtraction images were analyzed, 215 in each modality, in daily sessions of 20 images at the most. The observer assigned a score to each image according to the characteristics of bone mineral loss: score 0 for the absence of a dark image representing mineral loss in the dental apex region and 1 for the presence of a dark image representing mineral loss in the dental apex region. The rater agreement in assessment repetition was analyzed by kappa statistics, according to image modality. The sensitivity, specificity, and diagnostic accuracy values were calculated for each image modality considering a 0.5 prevalence (77 apices at T0: negative control; 77 apices at T2, T4, and T6: experimental comparison groups). The percentage of correct diagnoses according to the experimental time and image modality was compared using the analysis of variance, complemented by post hoc test adequate to sample characteristics, at 5% significance level.

Figure 1. The sequence of subtraction images of a sample tooth in different experimental times. T0, time 0; T0, repeat time 0; T2, after 2 hours of acid application; T4, after 4 hours of acid application; T6, after 6 hours of acid application.

To evaluate radiograph reproducibility and to compose a negative control group, two baseline radiographic images were taken at the start of the experiment: one image was taken at time zero period (T0) and one repeat image at the same time period and thus designated T0 (repeat T0). The simulation of apical lesions was performed by the sequential application of 0.5 mL of 70% perchloric acid at the bottom of each alveolus (18, 19). A pellet made of cotton was placed at the bottom of the alveolus before application of the acid, and the alveolus was later sealed with wax. At the end of each acid application, the pellet was removed and the teeth returned to their previously dried alveoli. Radiographic images were repeated after 2 hours (T2), after 4 hours (T4), and after 6 hours (T6) of perchloric acid application in the apical area of the alveolus. The acid was reapplied after each radiograph was taken. The radiographic images were digitally reproduced and cut so that each image showed only one tooth (two apices per image; although in one sample tooth only, one apex was analyzed). This procedure resulted in 39 images for each experimental time and in a total of 195 digitized images and 195 digital images. The software Adobe Photoshop v. 7.0.1 (Adobe Inc, San Jose, CA) was used for subtraction of the images (Fig. 1). A grayscale inversion filter was applied to the initial image, which was copied and pasted as a new layer in the repeat and experimental images (T0, T2, T4, and T6). That layer had transparency attribute equal to 50%. The digital and digitized images resulting from subtraction were recorded, codified, randomly inserted into a multimedia presentation, and then shown to the observer experienced in the DSR technique who was blinded to which group each sample belonged. All assessments were performed under standardized visualization conditions, and the observer was not allowed to manipulate or adjust image contrast or

Results
Observer reproducibility resulted in kappa values expressing moderate (0.574) and good (0.640) agreement for digitized and digital image modalities, respectively. However, in the assessment by experimental time, diagnostic reproducibility was 100% for DSR of the T0 images, 54.5% for the T2 images, 91.7% for the T4 images, and 85% for the T6 images. The sensitivity, specificity, and accuracy values for each experimental time were then calculated for each modality. There were no significant differences concerning the imaging method (Table 1). Nevertheless, sensitivity significantly increased proportionally to the length of exposure to acid. Multiple analysis of variance complemented by the Dunnetts T3 multiple comparison test (significance level at 5%) confirmed these results (Table 2). T6 images had the highest percentage of correct responses, which was significantly higher than that of other times regardless of the imaging method used. T0 and T4 images showed no statistically significant differences when compared with each other, but both groups showed significantly higher percentage of correct responses than those observed for T2 images (F 19.25, p 0.001). There was no difference in the percentage of correct responses as to the method for any of the experimental times (F 0.37, p 0.543).

Discussion
The creation of periapical bone lesions based on a chemical model proved an effective method of simulating alveolar bone lesions that can be subsequently evaluated on periapical radiographs. The created lesions had an indefinite and diffuse contour, which made diagnosis dif-

TABLE 1. Accuracy, Specificity, and Sensitivity Values of the Analysis According to Experimental Time and Imaging Method Imaging Method Parameter (95% CI)
Specificity T2 sensitivity T4 sensitivity T6 sensitivity 0.896 (0.8000.951) 0.688 (0.5710.786) 0.870 (0.7700.933) 0.961 (0.8830.990)

Digitized Accuracy
0.792 0.883 0.929

Digital (95% CI)

0.844 (0.7400.913) 0.714 (0.5980.809) 0.922 (0.8320.968) 1.000 (0.9411.000)

Accuracy
0.779 0.883 0.922

CI, confidence interval; T2, 2 hours of acid application; T4, 4 hours of acid application; T6, 6 hours of acid application.

JOE Volume 34, Number 12, December 2008

Panoramic Digital vs. Digitized Images to Detect Simulated Periapical Lesions

1501

Basic ResearchTechnology
TABLE 2. Comparison of Percentage of Correct Diagnoses According to Experimental Time and Imaging Method Method Time Percentage
T0= T2 T4 T6 Total 0.896 0.688 0.870 0.961 0.854

Total Digital Percentage*

0.870B 0.701C 0.896B 0.981A 0.862

Digitized Standard Error

0.035 0.053 0.039 0.022 0.020

Percentage
0.844 0.714 0.922 1.000 0.870

Standard Error
0.042 0.052 0.031 0.000 0.019

Standard Error
0.027 0.037 0.025 0.011 0.014

T0=, repeat time 0; T2, 2 hours of acid application; T4, 4 hours of acid application; T6, 6 hours of acid application. *Means followed by distinct letters are significantly different, analysis of variance complemented by Dunnetts T3 multiple comparison test, significance level at 5%.

ficult and consequently closer to what actually occurs in periapical inflammatory lesions (15, 18 20) based on the mechanical model (13, 14, 2123). The DSR technique provided identification at the shortest length of exposure to acid, which corroborates results reported in previous studies (18 20, 24). The occurrence of these small changes emphasizes the need of technique accuracy and reproducibility of techniques used in radiographic assessments of bone losses (25). In addition, the use of more than one observer to assess technique reproducibility lowers the chances of successfully obtaining measurements of desirable quality (26). When radiographic images are analyzed by several observers, a great discrepancy between the obtained results is observed; however, when the same observer (intrarater) performs the comparisons individually using different techniques and images, the accuracy of the obtained results increases (27, 28). The present study used a single observer similarly to the characteristics described earlier, taking into account that the analysis of DSR images requires training and experience in order to increase result reliability (29). In the evaluation of intrarater reproducibility agreement, kappa values varied between 0.574 and 0.640, which provided moderate and good agreement for digitized and digital images, respectively. These results were similar to those reported by Koenig et al (19) who investigated diagnostic reliability of periapical lesions and had moderate to good intrarater agreement. However, when reproducibility was evaluated according to experimental time, there was 100% agreement at T0 and 91.7 and 85% at T4 and T6, respectively, which corresponds to a nearly perfect agreement. The earliest lesion stage (represented herein by T2) revealed the worst reproducibility result, and that sample fraction accounted for poor agreement of the total sample. Our results allow us to infer that DSR in both modalities provides similar conditions for the diagnosis of incipient periapical lesions. It has been previously observed that DSR increases diagnostic ability of small bone defects (24, 29, 30). However, the defects resulting from T2 had the lowest percentage of correct diagnosis in both modalities (digital and digitized) when compared with the other experimental times, confirming the findings reported by other researchers (29, 30). The observers diagnostic ability increased proportionally to defect size regardless of imaging modality. This can be explained by the reduced contrast discrimination in DSR images. However, even exhibiting a lower value of correct diagnoses than that obtained at other experimental times, the observer was able to identify incipient periapical lesions in more than 70% of cases. Although sensitivity values were higher in the DSR images in panoramic digital radiographs, there was no statistically significant difference between both modalities. The specificity value in the detection of simulated periapical lesions was high for both modalities. We verified that the use of digitized images only slightly increased the specificity 1502

value when compared with the use of digital images, with no statistical significance. The problems of radiographic diagnosis have always been associated with low specificity (30). The comparison between DSR image accuracy and conventional and direct digital images in the identification of small periapical defects in this study showed accuracy moderate to high for both DSR modalities, in opposition to results previously reported in the literature (21). When the experimental times were analyzed individually, accuracy at T6 was higher than that of other times, both for the digitized and digital modality. The diagnostic accuracy for DSR increased proportionally to the length of exposure of the bone defect to acid. Therefore, the subjective analysis of DSR images in this study showed satisfactory results concerning the verification of true-positive and true-negative in digitized and digital panoramic radiographs with and without periapical lesions generated by application of acid solution, allowing for detection of incipient periapical lesions.

References
1. Bender IB, Seltzer S. Roentgenographic and direct observation of experimental lesion in bone: I. J Am Dent Assoc 1961;62:152 60. 2. Bender IB, Seltzer S. Roentgenographic and direct observation of experimental lesion in bone: II. J Am Dent Assoc 1961;62:708 16. 3. Barbat J, Messer HH. Detectability of artificial periapical lesions using direct digital and conventional radiography. J Endod 1998;24:837 42. 4. Low KM, Dula K, Brgin W, von Arx T. Comparison of periapical radiography and limited cone-beam tomography in posterior maxillary teeth referred for apical surgery. J Endod 2008;34:557 62. 5. Nair MK, Nair UP. Digital and advanced imaging in endodontics: a review. J Endod 2007;33:1 6. 6. Jeffcoat MK, Reddy MS, Webber RL, Williams RC, Ruttimann UE. Extraoral control of geometry for digital subtraction radiography. J Periodontal Res 1987;22:396 402. 7. Reddy MS, Jeffcoat MK. Digital subtraction radiography. Dent Clin North Am 1993;37:553 65. 8. Lehmann TM, Grndahl HG, Benn DK. Computer-based registration for digital subtraction in dental radiology. Dentomaxillofac Radiol 2000;29:323 46. 9. Yoshioka T, Kobayashi C, Suda H, Sasaki T. An observation of the healing process of periapical lesions by digital subtraction radiography. J Endod 2002;28:589 91. 10. Gegler A, Mahl C, Fontanella VC. Reproducibility of and file format effects on digital subtraction radiography of simulated external root resorptions. Dentomaxillofac Radiol 2006;35:10 3. 11. Wakoh M, Nishikawa K, Otonari T, et al. Digital subtraction technique for evaluation of peri-implant bone change in digital imaging. Bull Tokyo Dent Coll 2006;47:57 64. 12. Carvalho FB, Gonalves M, Tanomaru-Filho M. Evaluation of chronic periapical lesions by digital subtraction radiography by using Adobe Photoshop CS: a technical report. J Endod 2007;33:4937. 13. Nicopoulou-Karayianni K, Brgger U, Brgin W, Nielsen PM, Lang NP. Diagnosis of alveolar bone changes with digital subtraction images and conventional radiographs. An in vitro study. Oral Surg Oral Med Oral Pathol 1991;72:251 6. 14. Parsell DE, Gatewood RS, Watts JD, Streckfus CF. Sensitivity of various radiographic methods for detection of oral cancellous bone lesions. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1998;86:498 502. 15. Cunha FS, da Silva AE, Larentis N, Fontanella V. Diagnostic performance of periapical radiographs in digitized simulated bone loss in apical region. Rev Fac Odontol Univ Passo Fundo 2005;10:88 93.

Miguens et al.

JOE Volume 34, Number 12, December 2008

Basic ResearchTechnology
16. Masood F, Katz JO, Hardman PK, Glaros AG, Spencer P. Comparison of panoramic radiography and panoramic digital subtraction radiography in the detection of simulated osteophytic lesions of the mandibular condyle. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2002;93:626 31. 17. Braga CPA, Gegler A, Fontanella V. Evaluation of the influence of the thickness and of the relative position of soft tissues simulating materials in the optic density of periapical radiographies of the posterior region of the mandible. Cienc Odontol Bras 2006;9:52 8. 18. Meier AW, Brown CE, Miles DA, Analoui M. Interpretation of chemically created periapical lesions using direct digital imaging. J Endod 1996;22:516 20. 19. Koenig L, Parks E, Analoui M, Eckert G. The impact of image compression on diagnostic quality of digital images for detection of chemically-induced periapical lesions. Dentomaxillofac Radiol 2004;33:37 43. 20. Tirrell BC, Miles DA, Brown CE Jr, Legan JJ. Interpretation of chemically created lesions using direct digital imaging. J Endod 1996;22:74 8. 21. Dove SB, McDavid, WD, Hamilton KE. Analysis of sensitivity and specificity of a new digital subtraction system: an in vitro study. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2000;89:771 6. 22. Almeida SM, Bscolo FN, Haiter Neto F, Santos JCB. Evaluation of three radiographic methods (conventional periapical, digital periapical and panoramic) in the diagnosis of artificially produced periapical lesions. Pesqui Odontol Bras 2001;15:56 63. 23. Folk RB, Thorpe JR, McClanahan SB, Johnson JD, Strother JM. Comparison of two different direct digital radiography systems for the ability to detect artificially prepared periapical lesions. J Endod 2005;31:304 6. 24. Southard KA, Southard TE. Detection of simulated osteoporosis in human anterior maxillary alveolar bone with digital subtraction. Oral Surg Oral Med Oral Pathol 1994;78:655 61. 25. Bittar-Cortez JA, Passeri LA, de Almeida SM, Haiter-Neto F. Comparison of periimplant bone level assessment in digitized conventional radiographs and digital subtraction images. Dentomaxillofac Radiol 2006;35:258 62. 26. Pecoraro M, Azadivatan-le N, Janal M, Khocht A. Comparison of observer reliability in assessing alveolar bone height on direct digital and conventional radiographs. Dentomaxillofac Radiol 2005;34:279 84. 27. Wu DM, Wu YN, Guo W, Sameer S. Accuracy of direct digital radiography in the study of the root canal type. Dentomaxillofac Radiol 2006;35:2635. 28. Ekberg EC, Petersson A, Nilner M. An evaluation of digital subtraction radiography for assessment of changes in position of the mandibular condyle. Dentomaxillofac Radiol 1998;27:230 5. 29. Tyndall DA, Kapa SF, Bagnell CP. Digital subtraction radiography for detecting cortical and cancellous bone changes in the periapical region. J Endod 1990;16:173 8. 30. Stassinakis A, Brgger U, Stojanovic M, Brgin W, Lussi A, Lang NP. Accuracy in detecting bone lesions in vitro with conventional and subtracted direct digital imaging. Dentomaxillofac Radiol 1995;24:2327.

JOE Volume 34, Number 12, December 2008

Panoramic Digital vs. Digitized Images to Detect Simulated Periapical Lesions

1503