Optimization of radiation doses at panoramic radiography for examination of children and adolescents through digital image processing Björn Svenson1, Lars Larsson1, Lars Gunnar Månsson2, Anne Thilander-Klang2 1Department of radiology, Skaraborg Hospital, Skövde, Sweden
2Department of radio physics, University of Göteborg, Sweden
Running title: Address for correspondence and reprints: Dr. Björn Svenson, Department of Radiology, Skaraborg Hospital, SE-541 85 Skövde, Sweden
+46 (0)500 431376
+46 (0)500 431344
Optimization of radiation doses in panoramic radiography may be seen as a necessary step as many children and adolescents will be subject to panoramic examinations in connection to orthodontic treatment planning. The panoramic x-ray examination is the one most chosen examination before the orthodontic treatment1 and it was shown that the major absorbed dose to the thyroid gland was received from that examination.1 The radiation dose from a panoramic examination to the thyroid gland is, however, rather small, but despite that a cautious attitude should be taken to reduce the radiation dose to the thyroid gland of children and adolescents. The anatomic position of the thyroid gland with its close connection to the mandible and that the thyroid gland is one of the most sensitive organs for radiation-induced oncogenesis2 make it an organ of concern in dental radiography.3 The need for limiting exposure to the thyroid of young people has actually been stressed and emphasised by the findings following the Chernobyl accident in which it was shown that the incidence of thyroid cancer in children in the Belarus area was increased from less than 1 case per million to 100 per million per year in certain areas after the accident.2, 4 It was concluded in a newly published study that there is a small increased risk of cancer from low doses.5 These findings altogether are important for justifying the measures for limiting thyroid exposure.4, 6, 7
All x-ray examinations should be optimized in order to reduce the patient radiation dose as low as reasonable achievable without reducing the diagnostic information needed.8, 9 Optimization of the x-ray exposure has been a reasonable easy task when a film/screen was used as detector. The only parameters that could be altered were the tube voltage, tube current and filtration. If the tube voltage was lowered in order to reduce the radiation dose it resulted in an image with increased noise. An increase in radiation energy by increasing the tube voltage and/or filtration results in a lower radiation dose to the patients but also to an increased level of noise.10 The settings of the exposure systems has since long been tried out for the film/screen system. These settings may still be in use despite the introduction of digital technique. This will result in a radiation dose larger than necessary as the potential of digital technique for a dose reduction is not used. Modern digital detectors are considerably more sensitive to radiation than the old film/screen system. This implies that the tube current can be lowered without reaching an unacceptable noise level. This has been used in connection to a change to digital technique.11, 12 The introduction of digital technique in the x-ray departments radically changed the conditions to optimization. In a number of studies in x-ray departments it was proven that an optimization of the radiographic process will lead to a lowered absorbed dose without any hazardous reduction of diagnostic image quality.12-14 Thus, in panoramic radiography attempts have been made to reduce the radiation dose by either reducing the tube voltage,15tube current16 or a combination of tube voltage and tube current17 when a CCD detector used. The image quality was found not to be the optimum as the image contained more noise when tube voltage and/or tube current were lowered. However, the diagnostic image quality was found to be equivalent to that obtained with a conventional film/screen system.15, 16, 18 In another study using a storage phosphor system it was shown that compared to a film/screen system it did not result in a decreased radiation dose.19
By using image processing the digitally obtained radiographs can be processed in order to reduce the noise and enhance image contrast. The technique for that type of task is under rapid development and the latest methods have great opportunities in displaying the diagnostic interesting information. This new opportunity make it more complicated to optimize the images but implies a great possibility to either increase the content of the image and/or in combination with dose reduction reduce the radiation dose to the patient at a retained image quality.20 The objects of the study are to study the impact of dose optimization in panoramic radiography in combination with image processing on diagnostic image quality.
Material and methods
Radiographic equipment and technique
A panoramic X-ray machine, Scanora® (Soredex, Orion Cos., Helsinki, Finland), multifunctional x-ray unit was used for the study. The Scanora unit has a focal spot of 0.3x0.3 mm and a focus-film distance of 575 mm. The size of the collimator slit was 0.6 x 29.5 mm2 and the corresponding secondary collimator placed on the cassette holder 3.5 x 140 mm2. Panoramic examinations were performed using the panoramic dental program (#003). A skull phantom consisting of a natural human cranium embedded in a thermoplastic material to simulate human soft tissue was used for obtaining the panoramic images. The skull phantom was firmly fixed in the head holder of the x-ray machine (Figure 1). The panoramic radiographs of the phantom were obtained by using a 15x30 cm cassette with storage phosphor plates (ADCC HR MD 10 plates, Agfa-Gevaert NV, Mortsel, Belgium). Exposures were made for 57, 70 and 85 kV combined with exposures ranging from 24 mAs to 480 mAs (Table 1). Three series of radiographs were exposed the first having a sensitivity of 200 and the second 400 both with a filtration of 2.75 mm Al, and the third a sensitivity of 200 with added extra filtration, 0.1 mm Cu equivalent. In each series there were 13 images and in all there were 39 exposures made.
The image plates were scanned with the Agfa ADC Compact (Agfa-Gevaert NV, Mortsel, Belgium). Every image used in the study was processed with neutral “processing” parameters in the Agfa Quality Assurance (QA) station. The neutral processed image files were sent to a process station where the images could be processed and optimized by using a specially developed program for image processing (Context Vision, Stockholm, Sweden) in which the so-called GOP technique21 is applied. In GOP technique a method called adaptive filtering22 is used and is the key of the GOP technique. Image filtering techniques for production of an improved image can be divided into restoration and enhancement. Techniques for image restoration are intended to minimize some generally objective error criterion. Techniques for image enhancement are intended to improve the visual quality of an image and make it possible to enhance edges and lines to display sharply defined structures; it reduces the noise, enhances image contrast, and provides latitude compression so that structures in both dense and translucent areas are visible simultaneously. The experimental images were optimized to the structural image quality closest to that of the reference image by using the above mentioned technique. Every radiograph was given a random three digit number. All images were saved in DICOM format on a CD for viewing. In order to minimize the difference of different imaging systems at different x-ray departments the observers viewed the images using the same computer displayed on the same monitor. The computer and the monitor was transported and set up at the different x-ray departments participating in the study.
Observers and evaluation
Fifteen oral radiologists from four different x-ray departments evaluated all the experimental radiographs for structural quality using a five point rating scale ranging from 1 to 5, where 1=image quality much worse than reference image, 2=image quality worse than reference image, 3=image quality equal to reference image, 4= image quality better than reference image, 5= image quality much better than reference image. The overall evaluation was based on the comparison of structures as the mandibular canal, lamina dura, crista alveolaris, periodontal membrane, the enamel dentinal junction, trabecular bone, and compact bone of mandibular basis (Figure 2). A lap top computer (Dell Inspiron, Ireland) with a graphics card (Integrerated Intel Extreme Graphics 64MB) was used for displaying the radiographs on a monitor (Vista Line TFT-LCD Monitor, Olorin AB, Kungsbacka, Sweden). The monitor is at delivery from the manufacturer DICOM part 14 set. For the evaluation of the experimental radiographs a reference radiograph was used as a standard. The reference radiograph was obtained by using the usually employed exposure parameters. The reference image was exposed on an image plate which was scanned with the Agfa ADC Compact (Agfa-Gevaert NV, Mortsel, Belgium) and processed in the QA station. All experimental radiographs were compared with the reference image on the structural quality. The observers were asked to compare the experimental radiographs with the reference image and they were allowed to set the window for every radiograph to best suite their viewing. The reference image appeared also in the series of 39 experimental radiographs, but at this time neutrally processed. All viewing of the radiographs was performed in a dimmed room.
A dose-area-product KAP ionization chamber (RTI Doseguard 100, Mölndal, Sweden) was placed in front of the first collimator of the panoramic x-ray machine as described by Helmrot and Alm Carlsson.23 Exposures were performed as for the exposures of the radiographs as seen in Table 1. In all there were 39 measurements made.
Statistical analysis was performed by using descriptive statistics.
Due to technical errors two radiographs of 39 were deleted because they were not processed accurately. Both radiographs were exposed with low kV, low mAs and a sensitivity of 400 and may not possibly have influenced the result, because no radiographs exposed with low kV, low mAs and a sensitivity of 400 scored higher than 3. The only radiograph that scored higher than 3 with exposed with low kV and a sensitivity of 400 was exposed with a high mAs. Thus, there were just 37 radiographs in all that were evaluated by the 15 observers. The mean score for the observers and images were 3.1 with a range of 2.7-4. Two of observers had an average score of 2.7 and 2 around 4. The remaining 11 observers had an average of around 3.
Exposure settings together with the measured radiation doses in mGy for the different exposures are shown in Table 1. The reference image was obtained at 70 kV and 150 mAs with a sensitivity of 200 resulting in a dose of 70 mGy. In the experimental series of 37 images the original reference image # 488 was included but now neutral processed and dose optimized. It could be noted that when the observers compared the reference image with the experimental image obtained with the same exposure parameters as the reference image the experiment image had an average score of 3.3. Thus, the dose optimized experimental image resulted in a better diagnostic result than the reference image. It could be observed that dose optimization of the images may lead to a better image quality. There were 22 images (59%) scoring higher than 3, out of the 37 images in the study. Nine of these with a score higher than 3 (24%) resulted also in a lower radiation dose than the reference image. The dose reduction for those images was between 0-60 percent. For the remaining13 images with a score higher than 3 the dose was raised by 21-151%. For the remaining 15 radiographs with a score lower than 3 all were obtained with a lower dose compared with the reference image.
In Figure 2 the relationship between the evaluation score and change of radiation dose is displayed. It could be seen in the shaded area that there were 9 images which were obtained with a lower radiation dose and scoring equal or higher than 3. The other images are placed in the first and third quadrant. There are no images with a score lower than obtained with a high radiation dose. Of the 9 images with both a score higher than 3 and with a radiation dose lower than the reference, 4 were exposed with the sensitivity of 400, 3 with 200 plus 1 mm added Cu filtration, and 2 with the 200 sensitivity setting. The dose reduction for the 9 images was between 0-60 percent. Of the 22 images having an average score larger than 3, 8 were obtained with the 400 sensitivity setting, 7 with the sensitivity of 200 plus 1 mm added Cu filtration and 7 with sensitivity of 200.
In Fig 3 image 165 is displayed. This image was exposed with 70 kV and 240mAs and a sensitivity of 200 and extra filtration resulting in a radiation dose of 90.2 mGy and scoring 3.27. It could be noted the dramatic change in quality after image processing using the processing algorithm (Context Vision, Stockholm, Sweden) in which the so-called GOP technique21 is applied using adaptive filtering.22 The change in quality is accomplished by heavily increased contrast enhancement for large objects, heavily increased contrast enhancement for small objects and increased edge enhancement.
The present study showed that it is possible to reduce the radiation dose in panoramic radiography without jeopardizing the image quality by 60%. This finding was corroborated in a couple of other studies.15, 16, 18 Thus, Dula et al showed that a dose reduction of 43% was possible by reducing the kV, while Dannewitz et al concluded that a reduction in mA by 50% would be appropriate without loss of diagnostic quality.15, 16 In this study we both altered the kV and the mAs and added extra filtration Cu 0.1 mm. All images were dose optimized
A storage phosphor system was used as detector in the present study and it was shown that it is possible reduce the radiation dose by 5-60 % without losing structural image quality. This however is not the finding in a study by Farman et al where it was concluded that using a storage phosphor system did not result in a decreased radiation dose compared to a film/screen system.19 The reason for the diverging observation may be that all images in this study were dose optimized using a specially designed soft ware (Context Vision, Stockholm, Sweden).
Using the diagnostic score of 3 as cut off point, implying that the images should have at least a diagnostic quality of the level of the reference image, there were 22 images having a diagnostic quality equal or better than 3. If both the diagnostic criteria with a cut off point of 3 were used together with the criteria that the radiation should be lower than for the reference then there 7 images of the 37 that resulted in both diagnostic score of 3 and lower radiation lower dose compared to the reference.
The standard exposure usually used at our x-ray department for the panoramic examination has since long been 70 kV and 150 mAs using an image plate system with a sensitivity of 200. This system will give a radiation dose of 70 mGy. By using the 400 image plate system with a sensitivity double that of a 200 system it would be possible to reduce the mAs by half resulting in a dose reduction of 50%. Thus in the present study it was shown that using the 400 system the dose could be reduced by 0-60% and still giving a structural image quality with a score of 3-3.47. By using 96 mAs and 70 kV, it resulted in a reduction of exposure with 36%, and subsequent a dose reduction of 36% and a structural image quality score of above 3, image #288. Using 85 kV and the 400 system also resulted in a score of 3 or above and a radiation dose 41 % lower than with the reference, image #143. The selection of 70 kV instead of 85 kV may be seen as a compromise between dose and structural image quality. This study shows that it is possible reduce the dose by 36% without losing image quality. Other studies have shown when lowering the kV and the mAs that it is possible to reduce the radiation dose without jeopardizing the image quality. 15, 16, 18 It was shown in these studies that the radiation dose could be lowered to around 43% to 50%. However, the results in the present study show that it is possible to reduce the radiation dose by up to 60% without jeopardizing the image quality, but we would like to take a more moderate standpoint and propose that a dose reduction of around 40 % should be adequate, and still not jeopardizing the diagnostic quality. This moderate standpoint is taken due to a number of reasons. First all images were trimmed by using the Context Vision soft ware, second the variation of the observers’ subjective assessment of the radiographs and third the observer’s windowing in connection with the assessment. The windowing option together with the computer graphics card may possibly have affected the outcome of the evaluation of the radiographs. However, if we assume that 80% (12 observers) of the observers should coincide in a diagnostic score equal or higher than 3 then there were 20 radiographs of 37 that were in agreement with that statement. Only nine of these were obtained with a radiation dose equal or lower to the reference. This indicates that the observers might be more comfortable in their diagnostics with images obtained with a higher radiation dose than necessary.
In a couple of studies it was found that dose reduction was dependent on type of panoramic equipment used.19, 24 Hayakawa et al. found that a specific programme setting designed for children resulted in lower absorbed doses during panoramic radiography, irrespective of machine and receptor 25. In Scanora two panoramic programmes are available for use, the dental programme and the jaw programme. The dental programme is recommended for children as it limits the area exposed to radiation 26 and is also recommended for patients in which the region of the alveolar process is of interest and there is no need to image the temporomandibular joints. In the present study the dental program was used. It was shown that the absorbed dose to the thyroid gland could be reduced by 6% for the dental programme compared to the jaw programme.1 Thus, using the dental programme instead of the jaw programme in children an additional reduction of radiation dose could be possible just by adjusting the equipment to the task. Further dose reduction is possible by combining selection criteria1 and the use of dose optimization with GOP technique.
It could be concluded from the present study that it is possible to reduce the radiation dose by up to 60% without jeopardizing the image quality.
I wish to thank all the participants for their effort in reading all the radiographs. The study was supported by research grants from the research committee of Skaraborg Hospital VGSKAS-3160. I also wish to thank Context Vision for providing us with the GOP soft ware for optimizing the radiographs.
1. Svenson B, Sjöholm B, Jonsson B. Reduction of absorbed doses to the thyroid gland in orthodontic treatment planning by reducing the area of irradiation. Swed Dent J 2004; 28: 137-147.
2. Robbins J, Schneider AB. Thyroid cancer following exposure to radioactive iodine. Rev Endocr Metab Disord 2000; 1: 197-203.
3. Bristow RG, Wood RE, Clark GM. Thyroid dose distribution in dental radiography. Oral Surg Oral Med Oral Pathol 1989; 68: 482-487.
4. Nagataki S, Nystrom E. Epidemiology and primary prevention of thyroid cancer. Thyroid 2002; 12: 889-896.
5. Cardis E., Vrijheid M., Blettner M., Gilbert E., Hakama M., Hill C., Howe G., Kaldor J., Muirhead C. R., Schubauer-Berigan M., et al. Risk of cancer after low doses of ionising radiation: retrospective cohort study in 15 countries. Bmj 2005; 331: 77.
6. Jacob P, Kenigsberg Y, Goulko G, Buglova E, Gering F, Golovneva A, Kruk J, Demidchik EP. Thyroid cancer risk in Belarus after the Chernobyl accident: comparison with external exposures. Radiat Environ Biophys 2000; 39: 25-31.
7. Hindie E, Leenhardt L, Vitaux F, Colas-Linhart N, Grosclaude P, Galle P, Aurengo A, Bok B. Non-medical exposure to radioiodines and thyroid cancer. Eur J Nucl Med Mol Imaging 2002; 29 Suppl 2: 497-512.
8. SSI. FS 2000:1. The Swedish Radiation Protection Institute’s Regulations on General Obligations in Medical and Dental Practices using Ionising Radiation. Swedish Radiation Protection Authority. 2000.
9. European Union. Council Directive 97/43 Euratom, on health protection of individuals against the dangers of ionizing radiation in relation to medical exposure, and repealing Directive 84/466 Euratom. Official Journal of the European Communities No L 180, 9th July 1997:22-7. 1997.
10. Johns HE, Cunningham RJ. The physics of radiology. Chapter XVI. Springfield: Charles C Thomas, 1977:
11. McVey G, Sandborg M, Dance DR, Alm Carlsson G. A study and optimization of lumbar spine X-ray imaging systems. Br J Radiol 2003; 76: 177-188.
12. Persliden J, Helmrot E, Hjort P, Resjo M. Dose and image quality in the comparison of analogue and digital techniques in paediatric urology examinations. Eur Radiol 2004; 14: 638-644.
13. Geijer H. Radiation dose and image quality in diagnostic radiology. Optimization of the dose-image quality relationship with clinical experience from scoliosis radiography, coronary intervention and a flat-panel digital detector. Acta Radiol Suppl 2002; 43: 1-43.
14. Persliden J, Beckman KW, Geijer H, Andersson T. Dose-image optimisation in digital radiology with a direct digital detector: an example applied to pelvic examinations. Eur Radiol 2002; 12: 1584-1588.
15. Dula K, Sanderink G, van der Stelt PF, Mini R, Buser D. Effects of dose reduction on the detectability of standardized radiolucent lesions in digital panoramic radiography. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1998; 86: 227-233.
16. Dannewitz B, Hassfeld S, Eickholz P, Muhling J. Effect of dose reduction in digital dental panoramic radiography on image quality. Dentomaxillofac Radiol 2002; 31: 50-55.
17. Farman TT, Farman AG, Kelly MS, Firriolo FJ, Yancey JM, Stewart AV. Charge-coupled device panoramic radiography: effect of beam energy on radiation exposure. Dentomaxillofac Radiol 1998; 27: 36-40.
18. Farman TT, Farman AG. Clinical trial of panoramic dental radiography using a CCD receptor. J Digit Imaging 1998; 11: 169-171.
19. Farman AG, Farman TT. A comparison of image characteristics and convenience in panoramic radiography using charge-coupled device, storage phosphor, and film receptors. J Digit Imaging 2001; 14: 48-51.
20. Larsson L. Optimization of patient dose vs. image quality in a digitized department (conference poster). Proceedings of second Malmö conference on medical X-ray imaging. Malmö. 2004:
21. Granlund GH, Arvidsson J, Knutsson H. GOP a paradigm in hierarchical image processing. Proceedings of the first IEEE computer society international symposium on medical imaging and image interpretation, ISMI II'82. Berlin, Federal Republic of Germany. 1982:
22. Granlund GH, Knutsson H. Adaptive filtering. In: Signal processing for computer vision, 309-333. 1995. Kluwer Academic Publishers.
23. Helmrot E, Alm Carlsson G. Measurement of radiation dose in dental radiology. Proceedings of second Malmö conference on medical X-ray imaging. Malmö. Rad Prot Dosim. 2004:
24. Visser H, Hermann KP, Bredemeier S, Kohler B. [Dose measurements comparing conventional and digital panoramic radiography]. Mund Kiefer Gesichtschir 2000; 4: 213-216.
25. Hayakawa Y, Kobayashi N, Kuroyanagi K, Nishizawa K. Paediatric absorbed doses from rotational panoramic radiography. Dentomaxillofac Radiol 2001; 30: 285-292.
26. Hallikainen D, Gröndahl H-G, Kanerva H, Tammisalo H E. The Scanora concept. Optimized sequential dentomaxillofacial radiography. edn. Helsinki, 1992.