Dentomaxillofacial Radiology (2006) 35, 10-13
© 2006 British Institute of Radiology
doi: 10.1259/dmfr/86879455
Reproducibility of and file format effect on digital subtraction radiography of simulated external root resorptions
A Gegler*,1,
CEW Mahl1 and
V Fontanella2,3
1 Universidade Luterana do Brasil, Torres, RS, Brazil;
2 Universidade Luterana do Brasil, Canoas, RS, Brazil;
3 Universidade Federal do Rio Grande do Sul, Porto Alegre, RS, Brazil
*Correspondence to: Aderson Gegler, Rua Cel. Jõo Correa, 97/307, CEP 91350-190, Porto Alegre, RS - Brazil; E-mail: adersongegler{at}hotmail.com
Received 9 December 2004;
revised 9 May 2005;
accepted 14 June 2005
 |
Abstract
|
|---|
Objective: To evaluate, in simulated external root resorptions (ERR), two factors that may affect results of digital subtraction radiography (DSR): (1) intraobserver and interobserver reproducibility and (2) effects of digital file formats (TIFF, BMP and JPEG) on the estimation of mineral loss.
Methods: Eleven incisors were radiographed three times (NR, no resorption; SR, small – #1/4 round bur; and LR, large – #2 round bur) on standardized projections. The resulting images were reproduced and saved as TIFF, JPEG and BMP file formats. The pairs of TIFF images (NR x SR and NR x LR) were subtracted three times at 1 week intervals by three observers. One observer subtracted pairs of images (NR x SR and NR x LR) for all file formats. For each subtraction the resorption area was selected and mean pixel density values were calculated.
Results: Analysis of variance (ANOVA) (P=0.05) showed no statistical differences for intraobserver and interobserver values. Mean pixel density values were: observer A, 121.60±2.56 (NR x SR) and 111.84±4.04 (NR x LR); observer B, 121.86±2.50 (NR x SR) and 110.92±3.36 (NR x LR); and observer C, 121.70±2.39 (NR x SR) and 111.10±2.67 (NR x LR). Also, no statistical differences were found between file formats for LR (TIFF, 110.88±2.79; JPEG, 111.35±3.35; BMP, 111.00±2.70) and for SR between TIFF (121.30±2.34) and JPEG (120.46±1.51) or BMP (121.67±2.18) file formats. Differences were found between the JPEG and BMP groups.
Conclusions: DSR is reproducible in simulated ERR, and JPEG or BMP file formats do not affect results.
Keywords: diagnostic imaging;; subtraction technique;; image processing, computer-assisted;; image interpretation, computer-assisted
|
Introduction
|
|---|
Digital subtraction radiography (DSR) was developed in the 1980s1 and has since been studied and used for different clinical applications with the purpose of establishing diagnoses as early as possible. This technique is currently used for the evaluation of caries, alveolar bone loss and periapical lesions, and for the control of maxillary bone lesions and external root resorption (ERR),2 a process in which an injury to dental hard tissues leads to substantial tissue loss.3 The DSR technique provides better diagnoses of ERR than conventional radiography methods.4,5
Scanned radiographic images are converted into several types of digital files for use in digital subtraction radiography. However, the file formats more frequently studied are the Tagged Image File Format (TIFF), the Bitmap - Windows® pattern (BMP) and the Joint Photographic Experts Group (JPEG).6–15 The TIFF file format has been widely accepted as the standard for grey scale images,6 but, depending on the image resolution, the size of this file complicates its transmission and even its digital processing for DSR. The JPEG file format compresses the image, making it significantly smaller and facilitating its use. This compression occurs by discarding statistical redundancy – that is, data related to the similarity, correlation and predictability of image values – thus reducing the number of bits necessary to represent an image.11 This loss of information is not detected by the human eye without the help of adequate tools,7,11 but may be a source of error in the analysis by digital means.7 The loss of information in JPEG is irreversible,7,8 and the purpose of such loss is to eliminate unnecessary data.7,10 Several studies have investigated loss of information in digital images,7,9–14 but only one of these studies investigated the significance of this loss for the DSR technique.15
The use of DSR requires knowledge of computer sciences and demands special care in the performance of radiographic procedures. A broad range of software programs, file formats, resolutions, compression rates, and techniques have been developed. Therefore, the use of standardized procedures and the assessment of the different variables, such as file formats and software, are important to avoid biases and to provide reliable data when comparing results obtained in different studies. The objective of this study was to evaluate, in simulated external root resorptions and using the ImageLab® software, two factors that may affect the results of digital subtraction radiography: (1) intraobserver and interobserver reproducibility and (2) the effects of three digital file formats (TIFF, BMP and JPEG) on the estimation of mineral loss.
|
Materials and methods
|
|---|
The images used in this study were scanned from 11 groups of radiographs of teeth with external root resorption simulated by round burs (#1/4–0.25 mm and #2–2 mm) on the buccal surfaces. The teeth were central incisors previously used for students to practice cleaning and shaping techniques in an endodontics course. Teeth with visual irregularities on the root surface were excluded. A maxilla simulator model was made by removing the alveolar portion of a dry skull and cutting out the buccal and palatal portions of the anterior region with a diamond disc. These pieces were externally joined by a 2 cm thick acrylic block16,17 to simulate soft tissues, and wax was interposed between them in the region where the teeth were placed. This study was approved by the Ethics Committee of Universidade Federal do Rio Grande do Sul.
Resorptions were simulated on the buccal surfaces of each tooth, at the cervical, middle and apical thirds of the roots. The teeth were removed from the wax and underwent two drilling sequences with hand-held #1/4 and #2 round steel burs to simulate external resorption on the root surface. The cavities were made by penetration of all the cutting area of the bur in the centre of the root surface.
The teeth were radiographed individually, before and after each simulated resorption was made, using a dental X-ray unit (Spectro 70; Dabi Atlante, Ribeirão Preto, SP, Brazil), 70 kVp and 10 mA, 0.7 s exposure time, 40 cm focus–film distance, and no. 2 size Insight films (Eastman Kodak Co., Rochester, NY). The films were processed automatically with the Dent-X 9000 automatic processor (Dent-X, Elmsford, NY) in 4.5 min full cycle, all at the same time, with fresh processing solutions.
The radiographs were scanned with an Epson Perfection 2450® scanner equipped with a transparency reader (Epson, USA) and an adapted black acrylic mask to standardize the scanning position and limit the incidence of light. The images were acquired all at the same time and without changing brightness or contrast parameters. The Photoshop 7.0 software (Adobe Systems Inc., San Jose, CA) was used to save each radiographic image digitally at 8 bits and 300 dpi in three different file formats: JPEG, with minimal compression and maximum quality (level 12); TIFF without compression; and BMP without compression. Tools of the Windows XP system were used to compile the size of each file, and means and standard deviations where then calculated. The resulting values were compared using analysis of variance (ANOVA) and Tukey tests.
The ImageLab® software (version 2.3; Softium Informatics System, SP, Brazil) was used for DSR. The radiographs of teeth with resorptions were subtracted from those of teeth without resorptions. In each subtraction image, a circle, always of the same size and in the same position, was added to outline the reference area for the calculation of mean pixel density and standard deviations based on the visual detection of a darker area in the image. Three observers, experienced in the technique, performed DSR three times for each tooth at 1 week intervals using the TIFF digital file format for the large as well as for the small resorption images. For comparison of the different digital file formats, one of the observers also performed DSR using the other file formats (BMP and JPEG) under the same conditions but only once for each resorption. Therefore, DSR results of the same teeth were obtained for the comparison of the three digital file formats under study.
To detect intraobserver and interobserver variability and differences between the three digital file formats for DSR, randomized block ANOVA (P=0.05) was used to compare mean pixel density and standard deviations obtained from the repeated observations using the TIFF format and from the observations using the other file formats, and the Tukey test was used for the groups where differences were found.
|
Results
|
|---|
Mean pixel values and respective standard deviations of the subtractions performed by the three observers using images of large and small resorptions in TIFF file format at the three sessions are shown in Table 1
. Mean pixel values obtained by observer A for the subtraction of large and small resorptions at the first session were 110.93±2.55 and 121.69±2.68. The results of randomized block ANOVA (P=0.05) did not show a significant difference from mean pixel values in images of large and small resorptions found by other observers or by the three observers at the other sessions.
View this table:
[in this window]
[in a new window]
|
Table 1 Mean pixel values of digital subtraction radiography of small and large resorptions obtained by three observers at three sessions (TIFF files)
|
|
Mean pixel values and their standard deviations obtained from the subtractions performed by one of the observers using the TIFF, BMP and JPEG images of small and large resorptions are shown in Tables 2 and 3. The results of ANOVA and Tukey test revealed a significant difference (P=0.017) for the small resorptions between the BMP (121.67±2.18) and the JPEG (120.46±1.51) file formats. However, no difference was found between TIFF (121.30±2.34) and JPEG or BMP file formats. Randomized block ANOVA results for large resorptions (P=0.05) did not show significant differences in any of the groups – mean pixel values were 110.88±2.79 for TIFF files, 111.00±2.70 for BMP files, and 111.35±3.35 for JPEG files.
View this table:
[in this window]
[in a new window]
|
Table 2 Mean pixel values of digital subtraction radiography of small resorptions for three different file formats obtained by one observer
|
|
View this table:
[in this window]
[in a new window]
|
Table 3 Mean pixel values of digital subtraction radiography of large resorptions for three different file formats obtained by one observer
|
|
Table 4 shows the mean sizes of the digital files and their standard deviations. The BMP and TIFF files had the greatest mean value (108.91±5.35 kB and 108.90±5.35). The lowest mean and standard deviation values were found for the JPEG files (35.21±3.63 kB). The results of randomized block ANOVA and Tukey test (P<0.001) revealed significant differences between the mean sizes of JPEG and BMP or TIFF files.
|
Discussion
|
|---|
The search for alternative digital file formats not only for use in clinical settings, but also for storage, diagnosis or research – which includes their use in DSR – has been motivated by the need to obtain files that use less space and take less time to be transferred and processed.7 In this study, a difference was found between the sizes of digital files, and JPEG files were significantly smaller.
The search for smaller digital files is an important endeavour because of the uses that are commonly assigned to digital images. Digital images should be quickly and feasibly made available in clinical settings and through digital communications networks, and to be transmitted via the World Wide Web (WWW) for conference calls, teaching or distance consultation. This has motivated us to search for smaller files with quality parameters that do not interfere with the functions for which the files were assigned. However, these parameters must be constantly reviewed because of the fast pace of improvement in information technology.
The file formats tested in this study are those most often used in other investigations.6–15 As was seen above, most studies that have evaluated loss of digital file information have not focused on the use of digital files for DSR.7,9–14 They have, however, provided data about other characteristics of storage formats that can be correlated with clinical and research needs, similarly to the findings provided here for DSR.
A study comparing density between TIFF and JPEG radiographic images14 revealed that the results for JPEG images were not compatible with the criterion standard, but only one level of JPEG compression (1:21) was used in that investigation. Another study confirmed those findings,7 and suggested that JPEG file formats should not be used for quantitative digital functions. However, our results suggest that this format can be used for diagnosis.7,10,13 Some authors used JPEG images with a compression ratio of 1:21 and obtained acceptable results to support decision making about caries treatment, with no significant difference in the ability to detect caries.10 The study that suggested that the JPEG format should not be used for quantitative functions14 might have yielded better results – even for quantitative evaluations – if lower compression ratios had been used.
A study comparing different digital file formats and compressions for the diagnosis of occlusal or interproximal caries using the ROC curve did not find significant differences between different JPEG compression ratios (3 [1:5], 5 [1:12], 7 [1:20], 9 [1:33]), original TIFF image (1:1), and TIFF with LZW reversible compression (1:2).7 However, the authors admitted the possibility of visual differences in the case of interproximal caries when using the two greater compression ratios (1:20 and 1:33), even though no statistically significant differences were found. Other authors also failed to find significant differences in receiver operating characteristic (ROC) curve values when using original BMP images (1:1), JPEG 27 (1:14) and JPEG 53 (1:21).10 However, a more detailed analysis of data (MANOVA) revealed differences in the diagnosis of caries depth in the same images. The results of these two studies suggest that there are no diagnostic differences between JPEG and TIFF or BMP files, which is in agreement with our findings for DSR.
A compression ratio of up to 1:7 did not affect the detection of a 4 mg or greater bone gain in a study that investigated the use of the JPEG file format for DSR.15 Similarly to our results, their findings did not reveal any differences in the DSR mean pixel values of large resorptions for the different file formats. When using 1 mg of bone, however, they observed that bone gain was underestimated in the high compression ratios. This was also seen in our study, as the mean pixel values found for small resorptions using the JPEG file format were the smallest values (120.46±1.51), significantly different from the mean found for BMP files (121.67±2.18), but not different from the value found for TIFF files (121.30±2.34), a format widely used in DSR.6 However, number differences between means for TIFF and BMP files were very small, which may be explained by the fact that observers were allowed to choose the region for analysis within the outlined area.
Software programs use different mathematical functions (algorithms) to process images and to decompress files. Different results are thus obtained when using different software programs.14,18 In this study, intraobserver and interobserver reproducibility was achieved for the ImageLab® software, but this software was not compared with other programs.
When comparing quantitative DSR and conventional radiography in the determination of small changes in bone thickness,19 other authors found DSR detection limits of 200 µm for cortical bone and 500 µm for cancellous bone, whereas the corresponding detection limits of conventional radiography were 600 µm and 2850 µm. Small resorptions were easily identified in our study, and the values of their mean densities were reproducible for different observers and different sessions. Although measured for dental tissue, mean resorption size in our study was 250 µm, only 50 µm larger than the minimal bone thickness detected in the study mentioned above.19
In conclusion, good intraobserver and interobserver reproducibility was obtained for digital subtraction radiography using the ImageLab® software. No difference was found between the most widely used and accepted digital file format (TIFF) and the other file formats (BMP and JPEG) in DSR diagnosis of simulated external root resorptions. Therefore, the use of the JPEG file format is justified because of its significantly smaller size.
|
References
|
|---|
- Lehmann TM, Gröndahl HG, Benn DK. Computer-based registration for digital subtraction in dental radiology. Dentomaxillofac Radiol 2000; 29: 323–346.[Abstract]
- Heo MS, Lee SS, Lee KH, Choi HM, Choi SC, Park TW. Quantitative analysis of apical root resorption by means of digital subtraction radiography. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 2001; 91: 369–373.[Medline]
- Tronstad L, Debelian GJ. Reabsorções radiculares: etiologia, manifestações clínicas e terapia. In: Melo LL (ed). Traumatismo alvéolo-dentário. São Paulo: Artes Médicas, 1998: pp. 127–158.
- Andreasen JO. Experimental dental traumatology: development of a model for external root resorption. Endod Dent Traumatol 1987; 3: 269–287.[Medline]
- Hintze H, Wenzel A, Andreasen FM, Sewerin I. Digital subtraction radiography for assessment of simulated root resorption cavities. Performance of conventional and reverse contrast modes. Endod Dent Traumatol 1992; 8: 149–154.[CrossRef][Medline]
- Wenzel A, Gröndahl HG. Direct digital radiography in the dental office. Int Dent J 1995; 45: 27–34.[Medline]
- Wenzel A, Gotfredsen E, Borg E, Gröndahl HG. Impact of lossy image compression on accuracy of caries detection in digital images taken with a storage phosphor system. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1996; 81: 351–355.[CrossRef][Medline]
- Wenzel A. Digital radiography and caries diagnosis. Dentomaxillofac Radiol 1998; 27: 3–11.[Abstract]
- Yuasa H, Ariji Y, Ohki M, Naitoh M, Shiojima M, Ushida M, et al. Joint Photographic Experts Group compression of intraoral radiographs for image transmission on the World Wide Web. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1999; 88: 93–99.[Medline]
- Janhom A, van der Stelt PF, van Ginkel FC. Interaction between noise and file compression and its effect on the recognition of caries in digital imaging. Dentomaxillofac Radiol 2000; 29: 20–27.[Abstract]
- Roa L, Gómez-Cía T, Acha B, Serrano C. Digital imaging in remote diagnosis of burns. Burns 1999; 25: 617–623.[Medline]
- Slone RM, Foos DH, Whiting BR, Muka E, Rubin DA, Pilgram TK, et al. Assessment of visually lossless irreversible image compression: comparison of three methods by using an image-comparison workstation. Radiology 2000; 215: 543–553.[Abstract/Free Full Text]
- Janhom A, van der Stelt PF, van Ginkel FC, Geraets WGM. Effect of noise on the compressibility and diagnostic accuracy for caries detection of digital bitewing radiographs. Dentomaxillofac Radiol 1999; 28: 6–12.[Abstract]
- Gürdal P, Hildebolt CF, Akdeniz BG. The effects of different image file formats and image-analysis software programs on dental radiometric digital evaluations. Dentomaxillofac Radiol 2001; 30: 50–55.[Abstract]
- Fidler A, Likar B, Pernus F, Skaleric U. Impact of JPEG lossy image compression on quantitative digital subtraction radiography. Dentomaxillofac Radiol 2002; 31: 106–112.[Abstract]
- Braga CPA, Machado LX, Gegler A, Mahl CEW, Fontanella V. Materiais simuladores de tecidos moles na região posterior da mandíbula: avaliação por subtração radiográfica digital. In: Abstracts Book, XIV Salão de Iniciação Científica, UFRGS, Porto Alegre, Brazil. Porto Alegre: UFRGS, 2002, p. 541 (Abstr 110). (CD-ROM.).
- Meurer E. Análise ótica da densidade óssea da região parassinfisária por um sistema de radiografia digital, utilizando simuladores de tecidos moles [dissertation]. Porto Alegre: PUCRS; 2000, p. 123.
- Analoui M. Radiographic digital image enhancement. Part I: spatial domain techniques. Dentomaxillofac Radiol 2001; 30: 1–9.[Abstract]
- Christgau M, Hiller KA, Schmalz G, Kolbeck C, Wenzel A. Quantitative digital subtraction radiography for the determination of small changes in bone thickness. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1998; 85: 462–472.[CrossRef][Medline]
Top
Abstract
Introduction
