Subjective image quality assessment of cross sectional imaging methods for the symphyseal region of the mandible prior to dental implant placement

Subjective image quality assessment of cross sectional imaging methods for the symphyseal region of the mandible prior to dental implant placement

Journal of Dentistry, 2011-11-01, Volume 39, Issue 11, Pages 764-770, Copyright © 2011 Elsevier Ltd

Abstract

Objectives

A three dimensional appreciation of the form of the anterior mandible is required to place dental implants safely in the region. This study compared the subjective image quality of four available methods of cross sectional imaging of the symphyseal region, the lateral cephalometric view, cone beam computed tomography (CBCT), spiral tomography and the transymphyseal X-ray view.

Methods

An experimental water phantom was developed to reproduce human soft tissue around the mandible. Images were taken of four mandibles by the four X-ray techniques. Three different CBCT machines were included.

The source of each image was disguised by displaying all images in the same format on the same computer screen. A protocol was developed to process the images for viewing whilst preserving their image quality.

A panel of observers of ten dentists viewed the images and rated their image quality by recording their agreement with six statements on a five point Likert scale.

Results

The results showed a statistically significant difference in image quality between imaging methods. There were clear differences in the ratings between the three cone beam computed tomography machines. Small volume, high resolution CBCT provided images with the highest scores for subjective image quality.

Conclusions

Within the limitations of this study, significant differences in subjective image quality were found between imaging systems used for cross sectional imaging for the symphyseal region of the mandible prior to dental implant placement.

Introduction

Implant supported overdentures in the edentulous anterior mandible are considered the treatment of choice in many cases of severe or moderate alveolar resorption. For example, the provision of two implants in the lower canine regions with stud attachments can be a relatively simple way of addressing otherwise insoluble denture problems. Hyland et al. demonstrated the functional improvement and increased social confidence that implant supported mandibular overdentures can provide. The form of the anterior mandible varies greatly according to the degree and pattern of resorption. This can result in narrow, shallow or knife edge ridges which can complicate implant placement. Crucially, there may also be a lingual fossa. Perforation of the lingual cortical plate during preparation for placement of dental implants has the potential to cause severe bleeding and a life threatening upper airway obstruction. An awareness of the form of the anterior mandible is required in order to avoid such complications. The importance of the correct positioning of dental implants was highlighted by Javed and Romanos.

Several radiographic techniques are available to give a cross sectional image of the symphyseal region. The lateral cephalogram records the superimposition of the lower left to lower right canine region. Whilst this does not represent a true slice of the canine region, it is considered to be an adequate representation of the form of the bone in the anterior mandible. The lateral cephalogram is one of the recommended techniques prior to implant placement in the edentulous anterior mandible in several published guidelines. The transymphyseal view, described by Shelley and Horner, can be taken using conventional intra oral film (or digital receptor) and holders mounted extra-orally, the intention being to produce a similar view of the anterior mandible as that provided by a lateral cephalogram. Radiographic tomography, either as fairly simple optional “add-ons” to panoramic equipment or as dedicated spiral tomographic units have been available for 20 years. Computed tomography (CT) or cone beam computed tomography (CBCT) are widely used, often involving specialist planning software to give cross sectional images. Such imaging systems differ in important respects: radiation dose to patients, economic cost, availability to dentists and image quality. If adequate quality can be conveniently obtained using lower cost, lower radiation dose techniques, then it would be inappropriate to perform more expensive techniques. Review of the literature did not reveal any research that had looked at image quality for all these methods in the same study. Image quality is a measure of the degradation of an image relative to an “ideal” image. The process of image production introduces distortions, degradation and artefact, any or all of which can be measured objectively using test tools and image analysis methods. Subjective image quality grading is widely used as an alternative to objective assessment as it better reflects the clinical value of an image. The aim of this study, therefore, was to carry out image quality assessment of cross sectional imaging methods for the symphyseal region of the mandible.

Materials and methods

In order to allow multiple radiographic exposures, an experimental phantom was developed. Four containers were manufactured which represented a stylised reproduction of the human neck and lower third of the head ( Fig. 1 ). They were made from high impact polystyrene. The containers were supported by a tripod allowing the phantoms to be posed at convenient heights for all X-ray machines used in the study ( Fig. 2 ). The containers were then filled with water as a soft tissue equivalent. Four mandibles were obtained from private historical collections. Two of these were edentulous and two partially dentate. The partially dentate mandibles were included to represent a terminal dentition prior to extraction.

Diagram of experimental phantom.
Fig. 1
Diagram of experimental phantom.
The experimental phantom in use.
Fig. 2
The experimental phantom in use.

Before using the phantoms, the mandibles were soaked in water for a minimum of 48 h in an attempt to displace air from within the trabecular spaces. The intention was to reproduce soft tissue both within and around the mandibles. There were six imaging techniques included in the study as listed in Table 1 .

Table 1
Imaging techniques included in the study.
Imaging method Equipment type Analogue/digital Field of view Voxel/pixel size Voltage Current Time
CBCT 3D Accuitomo 170 a Digital 40 × 40 mm diameter × height 0.08 mm 90 kV 3.0 mA 9.0 s
CBCT Newtom VG b Digital 160 × 140 mm diameter × height 0.3 mm 110 kV 2.05 mA 3.6 s
CBCT Kodak 9000 3D c Digital 50 × 37 mm diameter × height 0.076 mm 70 kV 10 mA 10.68 s
Transymphyseal Trophy Atlantis d Analogue 31 × 41 mm width × height N/A 60 kV 7 mA 0.739 s
Lateral cephalogram Soredex Cranex 3 e Digital 115 × 125 mm width × height 0.083 mm 75 kV 10 mA 0.5 s
Spiral tomography Soredex Scanora e Analogue 69 × 97 mm width × height N/A 57 kV 1.6 mA 84 s

a J Morita Mfg Corp, Kyoto, Japan.

b Quantitative Radiology, Verona, Italy.

c Carestream Health, Inc. Rochester, New York, USA.

d Trophy Radiologie, Marne-la-Vallée, France.

e Soredex, Tuusula, Finland.

The transymphyseal and lateral cephalometric techniques give superimposed images of the anterior mandible from around the canine region to canine region. The outline produced is approximately that of the outline of the mandible at the midline. Therefore, in order to give comparable images, CBCT and Scanora images were taken of the midline of the mandible. The radiographers or practitioners were asked in each case to set their machines to the exposure factors that they would normally use to image the cross section of the anterior mandible.

Ideal reproduction of an image on a computer monitor exists when it displays one pixel for each pixel in the image. The intention was therefore to maintain this one to one ratio at each stage of image preparation in order to protect the integrity of the images. The 3D Accuitomo 170, Soredex Lateral Cephalograms and Newtom VG images were set at 1:1 within the manufacturers’ supplied software. A print screen image was then captured using a key combination from the computer keyboard. This captured the image at the resolution of the monitor. Each image was then opened in Adobe Photoshop. a

a Adobe Photoshop – Image manipulation software. Adobe, San José, California, USA.

The Kodak 9000 3D images were supplied as .png b

b .png – Portable network graphics. An image format which employs lossless data compression and thus preserves full image quality.

files and were opened directly in Adobe Photoshop. In Adobe Photoshop the images were cropped to remove extraneous information. This also allowed concealment of the origin of the images in order to reduce observer bias. Each image was saved at 100% with no magnification or reduction. Using this protocol, the resolution of the images displayed for assessment was exactly the same as the resolution of the images displayed in the manufacturers’ software.

This protocol led to the Newtom VG CBCT images being very small. It was felt that an observer would be very likely to use magnification to view the shape of the mandible. Therefore, in an attempt to reproduce real life conditions, the dedicated viewing software (NNT Report viewer v2.05) was used to produce a maximally magnified image. This was included in the study as well as the smaller, full size, image.

Two sets of images were produced on conventional film. These were the transymphyseal views and the Scanora images. No resolution is recognised for film and so an equivalent 1:1 ratio does not exist. In order to present the images in the same format as the digital images, the following protocol was used. Firstly an image size at the median value of the digital images was chosen. Using a scanner with a variable resolution, the images were then scanned at a resolution that produces a 1:1 image pixel to display pixel ratio.

There were seven image types, four mandibles and therefore 28 different images in total. Four images were duplicated in the assessment so that an intra-observer reliability calculation could be made after data collection. In order to do this the images were numbered sequentially. A random number generator was used to select four of these images. c

c http://www.random.org/ .

These four images were duplicated and then became image numbers 29, 30, 31 and 32. The random number generator was used again to randomise the new list of 32 images.

All images were displayed on the same notebook computer, a Sony Viao PGC-7G1M. The images were presented in the form of a slide show in Microsoft Picture and Fax Viewer. This software preserves the image size and displays one pixel in the image to one pixel on the monitor. The images were displayed against a black background. Examples of the images are shown in Fig. 3 . In an attempt to equalise viewing conditions the monitor was surrounded by a hood to reduce extraneous light.

Examples of images presented for assessment.
Fig. 3
Examples of images presented for assessment.
Top, Kodak 9000 CBCT; Middle, 3D Accuitomo 170 CBCT; Bottom, Newtom VG CBCT.

In an effort to standardise brightness and contrast, it was decided not to attempt to enhance the images in advance nor to allow observers to enhance them at the time of viewing. In the light of evidence from previous studies it was considered that this would introduce unnecessary bias. A possible confounding factor was the brightness and contrast setting of the computer monitor itself. It was therefore decided to set the brightness and contrast at the maximum setting for every assessment.

The assessment questionnaire for each image is shown in Fig. 4 . Members of the panel of observers were presented with six statements and asked to record their level of agreement. A panel of 10 members was chosen to reflect a range of experience in both dental implantology and dental radiology.

The questionnaire.
Fig. 4
The questionnaire.

Data were inputted into PASW d

d PASW – Predictive Analytics Software.

statistics 17.0 e

e SPSS Inc. Chicago, Illinois, USA.

(formerly SPSS). Descriptive statistics were prepared followed by statistical analysis. Following the recommendations of Gardner and Altman, analysis of the data was carried out in two ways: hypothesis testing and estimation using confidence intervals. For hypothesis testing, the null hypothesis was stated as, “There is no difference in image quality rating between the seven image types”. Data were analysed using the non-parametric repeated measures ANOVA, Friedman test. Statistical significance was set at p < 0.05. For estimation, the results were analysed by presentation of 95% confidence intervals. 95% confidence levels were set as the threshold for significance.

Results

The results of the Friedman test showed a statistically significant difference in ratings between the image types ( X 2 (6) = 289.78, p < 0.001). These results therefore favour rejection of the null hypothesis. The ratings for subjective image quality are shown by image type for all questions and all mandibles. The maximum, minimum, mean ratings, standard deviations and 95% confidence intervals are shown in Table 2 . A chart of confidence intervals is shown in Fig. 5 .

Table 2
Mean ratings of subjective image quality by image type with 95% confidence intervals.
Mean Standard deviation 95% confidence interval
Lower Upper
NCBCT 2.75 1.28 2.59 2.91
NCBCTmax 3.13 1.14 2.99 3.28
ACBCT 4.16 0.83 4.05 4.26
LC 2.95 1.13 2.80 3.09
Scan 2.71 1.37 2.53 2.88
TS 2.77 1.26 2.61 2.93
KCBCT 3.74 1.05 3.61 3.88
Descriptive statistics: N = 240. Maximum = 5. Minimum = 1.
Chart of mean ratings of subjective image quality by image type showing 95% confidence intervals.
Fig. 5
Chart of mean ratings of subjective image quality by image type showing 95% confidence intervals.

When considering mean ratings for all statements, all mandibles and all observers, the 3D Accuitomo 170 CBCT (ACBCT) images had the highest mean rating of the image types. Mean ACBCT = 4.16 (95% CI: 4.05–4.26). This was significant at the level of the 95% confidence intervals. The Kodak 9000 3D CBCT (KCBCT) images were ranked second. Mean KCBCT = 3.74 (95% CI: 3.61–3.88). The other image types had no significant differences in mean rating at the 95% CI level with the exception of the Newtom VG CBCT at maximum magnification (NCBCTmax) which had a marginally higher rating than the Soredex Scanora (Scan). This was just significant at the 95% CI level. Mean NCBCTmax = 3.13 (95% CI: 2.99–3.28). Mean scan = 2.71 (95% CI: 2.53–2.88).

Of the CBCT machines the Newtom VG CBCT at both full size (NCBCT) and at maximum magnification (NCBCTmax) had the lowest mean rating. Mean NCBCT = 2.75 (95% CI: 2.59–2.91). Mean NCBCTmax = 3.13 (95% CI: 2.99–3.28).

The results for the 4 duplicate images and their original images were compared for all observers and all statements. Intra-observer agreement was calculated using Cohen’s Kappa. There was an overall level of agreement of 0.59. According to the classification of Landis and Koch, a measure of 0.59 represents moderate agreement. Individual agreement for members of the panel of observers ranged from 0.28 (fair agreement) to 0.78 (substantial agreement). To calculate inter-observer agreement, the intra class correlation coefficient was calculated for the 10 members of the panel of observers. There was a measure of agreement of 0.89 (95% CI: 0.819–0.94). A commonly accepted minimum standard for such reliability coefficients is 0.7.

Discussion

The protocol for preparation of the digital images led to the Newtom VG CBCT images being very small. Magnification of an image leads to interpolation and thus loss of image quality. Nevertheless, in the case of the Newtom VG CBCT, it was felt that a magnified image is likely to be used by a clinician in order to be clinically useful. Therefore the dedicated viewing software (NNT Report viewer v2.05) was used to produce a maximally magnified image. This was included in the study as well as the smaller, full size, image.

One limitation of this study is that the images were not displayed in the conditions under which they would normally be viewed. This was so that the sources of the images could be disguised. The digital images would normally be displayed on dedicated viewing software which would enable magnification, reduction and adjustment of brightness and contrast. In the case of the conventional images, optical magnification would be possible. Also conventional images could be scanned and displayed at any convenient size.

The brightness and contrast of the images is an important factor in the perception of the images and in the case of digital images can be changed within the software. Enhancement of images in this way has been carried out by other authors prior to image assessment. Various methods have been used. These include application of a brightness and contrast algorithm, allowing observers to adjust brightness and contrast themselves and presetting the brightness and contrast. In a study by Baksi, the author preset the brightness and contrast himself. This, however, represents only one person’s perception of the ideal brightness and contrast. Other observers may not have the same preference and thus bias may be introduced. Borg and Gröndahl in 1996 allowed observers to change the brightness and contrast themselves. In another study, Borg and Gröndahl reported that allowing observers to adjust brightness and contrast had no effect on the final outcome.

The analysis of results shows significant differences in image quality rating between the four CBCT image types at the level of the 95% confidence interval. The 3D Accuitomo 170 was rated most highly, followed by the Kodak 9000 3D, the Newtom VG at maximum magnification and then the Newtom VG at full size. This finding is consistent with that of Liang et al. who found that the 3D Accuitomo was rated most highly for image quality out of the five CBCT machines in their study.

The reasons for this difference may be explained partly by the resolution of the machines. For CBCT machines the resolution is represented by voxel size, a larger voxel size leading to lower resolution. The Newtom VG has the lowest resolution, or largest voxel size, of the three CBCT machines with a voxel size of 0.3 mm. f

f http://67.192.191.166/uploads/pdf/NewTom/NT_ENTIntlBrochure07.pdf .

The user reported that the Newtom VG has a fixed resolution, which might be interpreted as suggesting some limitations in its efficacy. The later Newtom VG has two resolution options, 0.3 and 0.15 mm voxels. It is possible that had the higher resolution been available to us in this study, the image quality performance may have been better. Nevertheless, this explanation for the disappointing quality ratings for the Newtom VG is at odds with the findings of Liedke et al. These authors conducted a study to investigate the influence of voxel size on the diagnostic ability of CBCT to evaluate external root resorption. The authors used an iCat g

g Imaging Sciences International, Hatfield, Pennsylvania, USA.

CBCT machine which was set at voxel sizes of 0.2 mm, 0.3 mm and 0.4 mm. They concluded that a 0.3 mm voxel size gave the best compromise of diagnostic performance and X-ray exposure. Therefore, the disappointing image quality rating for the Newtom VG CBCT machine may not be fully explained by the larger voxel size. One can speculate that the processing of the images at the scanning centre, or the dedicated viewing software for the Newtom VG (NNT report viewer V2.05), may have had an important effect on observed image quality. Other factors in image quality of CBCT machines include object to recorder distance, signal to noise ratio, kV, mA, image receptor type, filtration, field of view size, artefacts, slice thickness and a number of subject related factors.

The voxel sizes for the 3D Accuitomo 170, h

h http://www.jmoritaeurope.de/productdatabase/server/assets/174/4/3_Accuitomo_170_EN_090904.pdf .

and Kodak 9000 i

i http://www.thedentalimagingcompany.co.uk/acatalog/K90003D_Brochure_EN.pdf .

were almost identical at 0.08 mm and 0.076 mm, respectively. Despite this, the Kodak 9000 images were not rated as highly as the 3D Accuitomo 170 images. There was, however, noticeably more noise with the Kodak 9000 images and noise is another of the key factors in digital image quality. In other respects the images appeared similar and it seems likely that, in large part, this explained the difference in image quality rating between these two machines.

In the lateral cephalometric view and the transymphyseal view, the beam is at 90° to the sagittal plane and records an image of the symphyseal region by superimposition of the lower left to lower right canine region. A comparison of these two techniques is therefore of interest. In respect of all statements combined, the lateral cephalometric technique scored very similarly to the transymphyseal technique (means of 2.95 and 2.77, respectively). The difference in mean rating was not significant at the level of the 95% confidence interval.

These findings suggest that the transymphyseal radiograph may be a useful alternative to the lateral cephalometric view in general dental practice. The technique has a number of advantages. It uses equipment and materials readily available in most dental practices, it is relatively inexpensive and it is not necessary to refer to a specialist centre. Further the radiation exposure is limited only to the area of interest because the X-ray beam passes only through the symphyseal region. No special processing facilities are required.

In this study the images were all presented in the same format. Each was an image of the midline of the mandible. Nevertheless, when making a comparison of imaging methods, the advantages and disadvantages of each should be taken into consideration. For example, the CBCT machines can obtain true cross-sectional images perpendicular to the curve of the jaw, free from superimposition. Therefore if two implants are planned for the lower canine regions, the exact area of the planned placement can be imaged. The transymphyseal and lateral cephalometric views are superimpositions of the anterior mandible which give an indication of the probable cross section of the canine regions but are less precise. On the other hand the radiation dose is an important consideration. The Effective Dose associated with the transymphyseal projection is around 1–2 μSv. This is a little less than a lateral cephalogram at 3 μSv and substantially lower than the CBCTs which are likely to have doses 10 times as great. Cost and availability are also important considerations.

Conclusions

Seven different image types have been used to image the cross section of the midline of the mandible on four different X-ray phantoms. These results show that there was a difference in image quality between these different image types.

These results also show differences in image quality between the three different CBCT machines and between the full size and magnified versions of one of the machines. Therefore it should not be assumed that all CBCT machines will produce images of the same quality. This study rated the Accuitomo 170 CBCT images above the Kodak 9000 CBCT images. These were in turn rated more highly than the Newtom VG images. Nevertheless, many other machines exist and the very rapidly changing field of CBCT technology means that these results will not remain current. Further studies will be necessary.

A limitation of the study was that images from machines which produce true slices such as the CBCT machines were compared with images that were produced by superimposition such as the lateral cephalometric and transymphyseal techniques. The results of this study can therefore only be applied to images of the midline of the mandible. When planning implant placement in the lower canine regions, a true slice of the canine regions is clearly more useful than the midline view. Nonetheless, factors such as the radiation dose, cost and convenience should also be taken into consideration.

Acknowledgements

The authors are grateful to Mr Daniel Jepson of Plastic Formers Ltd, Denton, Manchester, UK for his invaluable help in the manufacture of the experimental phantoms.

References

  • 1. Feine J.S., Carlsson G.E., Awad M.A., Chehade A., Duncan W.J., Gizani S., et. al.: The McGill consensus statement on overdentures. Mandibular two-implant overdentures as first choice standard of care for edentulous patients. Gerodontology 2002; 19: pp. 3-4.
  • 2. Hyland R., Ellis J., Thomason M., El-Feky A., Moynihan P.: A qualitative study on patient perspectives of how conventional and implant-supported dentures affect eating. Journal of Dentistry 2009; 37: pp. 718-723.
  • 3. Cawood J.I., Howell R.A.: A classification of the edentulous jaws. International Journal of Oral and Maxillofacial Surgery 1988; 17: pp. 232-236.
  • 4. Pigadas N., Simoes P., Tuffin J.R.: Massive sublingual haematoma following osseo-integrated implant placement in the anterior mandible. British Dental Journal 2009; 206: pp. 67-68.
  • 5. Kalpidis C.D., Setayesh R.M.: Hemorrhaging associated with endosseous implant placement in the anterior mandible: a review of the literature. Journal of Periodontology 2004; 75: pp. 631-645.
  • 6. Javed F., Romanos G.E.: The role of primary stability for successful immediate loading of dental implants. A literature review. Journal of Dentistry 2010; 38: pp. 612-620.
  • 7. FGDP(UK) : Selection criteria in dental radiography.2nd edn.2004.Faculty of General Dental Practitioners (UK) Royal College of Surgeons of England
  • 8. Horner K.: European guidelines on radiation protection in dental radiology.2004.European Commission
  • 9. Harris D., Buser D., Dula K., Grondahl K., Harris D., Jacobs R., et. al.: E.A.O. guidelines for the use of diagnostic imaging in implant dentistry. A consensus workshop organized by the European Association for Osseointegration in Trinity College Dublin. Clinical Oral Implants Research 2002; 13: pp. 566-570.
  • 10. Tyndall D.A., Brooks S.L.: Selection criteria for dental implant site imaging: a position paper of the American Academy of Oral and Maxillofacial radiology. Oral Surgery Oral Medicine Oral Pathology Oral Radiology and Endodontics 2000; 89: pp. 630-637.
  • 11. Shelley A.M., Horner K.: A transymphyseal X-ray projection to assess the anterior edentulous mandible prior to implant placement. Dental Update 2008; 35: pp. 689-694.
  • 12. Horner K., Drage N., Brettle D.S.: 21st century imaging.2008.Quintessence Publishing Co. Ltd.London
  • 13. Gardner M.J., Altman D.G.: . British Medical Journal (Clinical Research Edition) 1986; 292: pp. 746-750.
  • 14. Cohen J.: A coefficient of agreement for nominal scales. Educational Psychological Measurement 1960; 20: pp. 37-46.
  • 15. Landis J.R., Koch G.G.: The measurement of observer agreement for categorical data. Biometrics 1977; 33: pp. 159-174.
  • 16. Lohr K.N., Aaronson N.K., Alonso J., Burnam M.A., Patrick D.L., Perrin E.B., et. al.: Evaluating quality-of-life and health status instruments: development of scientific review criteria. Clinical Therapeutics 1996; 18: pp. 979-992.
  • 17. Borg E., Attaelmanan A., Grondahl H.G.: Subjective image quality of solid-state and photostimulable phosphor systems for digital intra-oral radiography. Dentomaxillofacial Radiology 2000; 29: pp. 70-75.
  • 18. Borg E., Grondahl H.G.: Endodontic measurements in digital radiographs acquired by a photostimulable, storage phosphor system. Endodontics & Dental Traumatology 1996; 12: pp. 20-24.
  • 19. Baksi B.G.: Measurement accuracy and perceived quality of imaging systems for the evaluation of periodontal structures. Odontology 2008; 96: pp. 55-60.
  • 20. Borg E., Grondahl H.G.: On the dynamic range of different X-ray photon detectors in intra-oral radiography. A comparison of image quality in film, charge-coupled device and storage phosphor systems. Dentomaxillofacial Radiology 1996; 25: pp. 82-88.
  • 21. Liang X., Jacobs R., Hassan B., Li L., Pauwels R., Corpas L., et. al.: A comparative evaluation of Cone Beam Computed Tomography (CBCT) and Multi-Slice CT (MSCT). Part I. On subjective image quality. European Journal of Radiology 2010; 75: pp. 265-269.
  • 22. Liedke G.S., da Silveira H.E., da Silveira H.L., Dutra V., de Figueiredo J.A.: Influence of voxel size in the diagnostic ability of cone beam tomography to evaluate simulated external root resorption. Journal of Endodontics 2009; 35: pp. 233-235.
  • 23. van Daatselaar A.N., van der Stelt P.F., Weenen J.: Effect of number of projections on image quality of local CT. Dentomaxillofacial Radiology 2004; 33: pp. 361-369.
  • 24. The Sedentexct Project [internet]. Manchester: The University of Manchester; 2009. Available from: www.sedentexct.eu .
  • 25. Ludlow J.B., Davies-Ludlow L.E., White S.C.: Patient risk related to common dental radiographic examinations: the impact of 2007 International Commission on Radiological Protection recommendations regarding dose calculation. Journal of the American Dental Association 2008; 139: pp. 1237-1243.

Was this article helpful?

Subjective image quality assessment of cross sectional imaging methods for the symphyseal region of the mandible prior to dental implant placement Andrew Martyn Shelley , Paul Brunton and Keith Horner Journal of Dentistry, 2011-11-01, Volume 39, Issue 11, Pages 764-770, Copyright © 2011 Elsevier Ltd Abstract Objectives A three dimensional appreciation of the form of the anterior mandible is required to place dental implants safely in the region. This study compared the subjective image quality of four available methods of cross sectional imaging of the symphyseal region, the lateral cephalometric view, cone beam computed tomography (CBCT), spiral tomography and the transymphyseal X-ray view. Methods An experimental water phantom was developed to reproduce human soft tissue around the mandible. Images were taken of four mandibles by the four X-ray techniques. Three different CBCT machines were included. The source of each image was disguised by displaying all images in the same format on the same computer screen. A protocol was developed to process the images for viewing whilst preserving their image quality. A panel of observers of ten dentists viewed the images and rated their image quality by recording their agreement with six statements on a five point Likert scale. Results The results showed a statistically significant difference in image quality between imaging methods. There were clear differences in the ratings between the three cone beam computed tomography machines. Small volume, high resolution CBCT provided images with the highest scores for subjective image quality. Conclusions Within the limitations of this study, significant differences in subjective image quality were found between imaging systems used for cross sectional imaging for the symphyseal region of the mandible prior to dental implant placement. 1 Introduction Implant supported overdentures in the edentulous anterior mandible are considered the treatment of choice in many cases of severe or moderate alveolar resorption. For example, the provision of two implants in the lower canine regions with stud attachments can be a relatively simple way of addressing otherwise insoluble denture problems. Hyland et al. demonstrated the functional improvement and increased social confidence that implant supported mandibular overdentures can provide. The form of the anterior mandible varies greatly according to the degree and pattern of resorption. This can result in narrow, shallow or knife edge ridges which can complicate implant placement. Crucially, there may also be a lingual fossa. Perforation of the lingual cortical plate during preparation for placement of dental implants has the potential to cause severe bleeding and a life threatening upper airway obstruction. An awareness of the form of the anterior mandible is required in order to avoid such complications. The importance of the correct positioning of dental implants was highlighted by Javed and Romanos. Several radiographic techniques are available to give a cross sectional image of the symphyseal region. The lateral cephalogram records the superimposition of the lower left to lower right canine region. Whilst this does not represent a true slice of the canine region, it is considered to be an adequate representation of the form of the bone in the anterior mandible. The lateral cephalogram is one of the recommended techniques prior to implant placement in the edentulous anterior mandible in several published guidelines. The transymphyseal view, described by Shelley and Horner, can be taken using conventional intra oral film (or digital receptor) and holders mounted extra-orally, the intention being to produce a similar view of the anterior mandible as that provided by a lateral cephalogram. Radiographic tomography, either as fairly simple optional “add-ons” to panoramic equipment or as dedicated spiral tomographic units have been available for 20 years. Computed tomography (CT) or cone beam computed tomography (CBCT) are widely used, often involving specialist planning software to give cross sectional images. Such imaging systems differ in important respects: radiation dose to patients, economic cost, availability to dentists and image quality. If adequate quality can be conveniently obtained using lower cost, lower radiation dose techniques, then it would be inappropriate to perform more expensive techniques. Review of the literature did not reveal any research that had looked at image quality for all these methods in the same study. Image quality is a measure of the degradation of an image relative to an “ideal” image. The process of image production introduces distortions, degradation and artefact, any or all of which can be measured objectively using test tools and image analysis methods. Subjective image quality grading is widely used as an alternative to objective assessment as it better reflects the clinical value of an image. The aim of this study, therefore, was to carry out image quality assessment of cross sectional imaging methods for the symphyseal region of the mandible. 2 Materials and methods In order to allow multiple radiographic exposures, an experimental phantom was developed. Four containers were manufactured which represented a stylised reproduction of the human neck and lower third of the head ( Fig. 1 ). They were made from high impact polystyrene. The containers were supported by a tripod allowing the phantoms to be posed at convenient heights for all X-ray machines used in the study ( Fig. 2 ). The containers were then filled with water as a soft tissue equivalent. Four mandibles were obtained from private historical collections. Two of these were edentulous and two partially dentate. The partially dentate mandibles were included to represent a terminal dentition prior to extraction. Fig. 1 Diagram of experimental phantom. Fig. 2 The experimental phantom in use. Before using the phantoms, the mandibles were soaked in water for a minimum of 48 h in an attempt to displace air from within the trabecular spaces. The intention was to reproduce soft tissue both within and around the mandibles. There were six imaging techniques included in the study as listed in Table 1 . Table 1 Imaging techniques included in the study. Imaging method Equipment type Analogue/digital Field of view Voxel/pixel size Voltage Current Time CBCT 3D Accuitomo 170 a Digital 40 × 40 mm diameter × height 0.08 mm 90 kV 3.0 mA 9.0 s CBCT Newtom VG b Digital 160 × 140 mm diameter × height 0.3 mm 110 kV 2.05 mA 3.6 s CBCT Kodak 9000 3D c Digital 50 × 37 mm diameter × height 0.076 mm 70 kV 10 mA 10.68 s Transymphyseal Trophy Atlantis d Analogue 31 × 41 mm width × height N/A 60 kV 7 mA 0.739 s Lateral cephalogram Soredex Cranex 3 e Digital 115 × 125 mm width × height 0.083 mm 75 kV 10 mA 0.5 s Spiral tomography Soredex Scanora e Analogue 69 × 97 mm width × height N/A 57 kV 1.6 mA 84 s a J Morita Mfg Corp, Kyoto, Japan. b Quantitative Radiology, Verona, Italy. c Carestream Health, Inc. Rochester, New York, USA. d Trophy Radiologie, Marne-la-Vallée, France. e Soredex, Tuusula, Finland. The transymphyseal and lateral cephalometric techniques give superimposed images of the anterior mandible from around the canine region to canine region. The outline produced is approximately that of the outline of the mandible at the midline. Therefore, in order to give comparable images, CBCT and Scanora images were taken of the midline of the mandible. The radiographers or practitioners were asked in each case to set their machines to the exposure factors that they would normally use to image the cross section of the anterior mandible. Ideal reproduction of an image on a computer monitor exists when it displays one pixel for each pixel in the image. The intention was therefore to maintain this one to one ratio at each stage of image preparation in order to protect the integrity of the images. The 3D Accuitomo 170, Soredex Lateral Cephalograms and Newtom VG images were set at 1:1 within the manufacturers’ supplied software. A print screen image was then captured using a key combination from the computer keyboard. This captured the image at the resolution of the monitor. Each image was then opened in Adobe Photoshop. a a Adobe Photoshop – Image manipulation software. Adobe, San José, California, USA. The Kodak 9000 3D images were supplied as .png b b .png – Portable network graphics. An image format which employs lossless data compression and thus preserves full image quality. files and were opened directly in Adobe Photoshop. In Adobe Photoshop the images were cropped to remove extraneous information. This also allowed concealment of the origin of the images in order to reduce observer bias. Each image was saved at 100% with no magnification or reduction. Using this protocol, the resolution of the images displayed for assessment was exactly the same as the resolution of the images displayed in the manufacturers’ software. This protocol led to the Newtom VG CBCT images being very small. It was felt that an observer would be very likely to use magnification to view the shape of the mandible. Therefore, in an attempt to reproduce real life conditions, the dedicated viewing software (NNT Report viewer v2.05) was used to produce a maximally magnified image. This was included in the study as well as the smaller, full size, image. Two sets of images were produced on conventional film. These were the transymphyseal views and the Scanora images. No resolution is recognised for film and so an equivalent 1:1 ratio does not exist. In order to present the images in the same format as the digital images, the following protocol was used. Firstly an image size at the median value of the digital images was chosen. Using a scanner with a variable resolution, the images were then scanned at a resolution that produces a 1:1 image pixel to display pixel ratio. There were seven image types, four mandibles and therefore 28 different images in total. Four images were duplicated in the assessment so that an intra-observer reliability calculation could be made after data collection. In order to do this the images were numbered sequentially. A random number generator was used to select four of these images. c c http://www.random.org/ . These four images were duplicated and then became image numbers 29, 30, 31 and 32. The random number generator was used again to randomise the new list of 32 images. All images were displayed on the same notebook computer, a Sony Viao PGC-7G1M. The images were presented in the form of a slide show in Microsoft Picture and Fax Viewer. This software preserves the image size and displays one pixel in the image to one pixel on the monitor. The images were displayed against a black background. Examples of the images are shown in Fig. 3 . In an attempt to equalise viewing conditions the monitor was surrounded by a hood to reduce extraneous light. Fig. 3 Examples of images presented for assessment. Top, Kodak 9000 CBCT; Middle, 3D Accuitomo 170 CBCT; Bottom, Newtom VG CBCT. In an effort to standardise brightness and contrast, it was decided not to attempt to enhance the images in advance nor to allow observers to enhance them at the time of viewing. In the light of evidence from previous studies it was considered that this would introduce unnecessary bias. A possible confounding factor was the brightness and contrast setting of the computer monitor itself. It was therefore decided to set the brightness and contrast at the maximum setting for every assessment. The assessment questionnaire for each image is shown in Fig. 4 . Members of the panel of observers were presented with six statements and asked to record their level of agreement. A panel of 10 members was chosen to reflect a range of experience in both dental implantology and dental radiology. Fig. 4 The questionnaire. Data were inputted into PASW d d PASW – Predictive Analytics Software. statistics 17.0 e e SPSS Inc. Chicago, Illinois, USA. (formerly SPSS). Descriptive statistics were prepared followed by statistical analysis. Following the recommendations of Gardner and Altman, analysis of the data was carried out in two ways: hypothesis testing and estimation using confidence intervals. For hypothesis testing, the null hypothesis was stated as, “There is no difference in image quality rating between the seven image types”. Data were analysed using the non-parametric repeated measures ANOVA, Friedman test. Statistical significance was set at p < 0.05. For estimation, the results were analysed by presentation of 95% confidence intervals. 95% confidence levels were set as the threshold for significance. 3 Results The results of the Friedman test showed a statistically significant difference in ratings between the image types ( X 2 (6) = 289.78, p < 0.001). These results therefore favour rejection of the null hypothesis. The ratings for subjective image quality are shown by image type for all questions and all mandibles. The maximum, minimum, mean ratings, standard deviations and 95% confidence intervals are shown in Table 2 . A chart of confidence intervals is shown in Fig. 5 . Table 2 Mean ratings of subjective image quality by image type with 95% confidence intervals. Mean Standard deviation 95% confidence interval Lower Upper NCBCT 2.75 1.28 2.59 2.91 NCBCTmax 3.13 1.14 2.99 3.28 ACBCT 4.16 0.83 4.05 4.26 LC 2.95 1.13 2.80 3.09 Scan 2.71 1.37 2.53 2.88 TS 2.77 1.26 2.61 2.93 KCBCT 3.74 1.05 3.61 3.88 Descriptive statistics: N = 240. Maximum = 5. Minimum = 1. Fig. 5 Chart of mean ratings of subjective image quality by image type showing 95% confidence intervals. When considering mean ratings for all statements, all mandibles and all observers, the 3D Accuitomo 170 CBCT (ACBCT) images had the highest mean rating of the image types. Mean ACBCT = 4.16 (95% CI: 4.05–4.26). This was significant at the level of the 95% confidence intervals. The Kodak 9000 3D CBCT (KCBCT) images were ranked second. Mean KCBCT = 3.74 (95% CI: 3.61–3.88). The other image types had no significant differences in mean rating at the 95% CI level with the exception of the Newtom VG CBCT at maximum magnification (NCBCTmax) which had a marginally higher rating than the Soredex Scanora (Scan). This was just significant at the 95% CI level. Mean NCBCTmax = 3.13 (95% CI: 2.99–3.28). Mean scan = 2.71 (95% CI: 2.53–2.88). Of the CBCT machines the Newtom VG CBCT at both full size (NCBCT) and at maximum magnification (NCBCTmax) had the lowest mean rating. Mean NCBCT = 2.75 (95% CI: 2.59–2.91). Mean NCBCTmax = 3.13 (95% CI: 2.99–3.28). The results for the 4 duplicate images and their original images were compared for all observers and all statements. Intra-observer agreement was calculated using Cohen's Kappa. There was an overall level of agreement of 0.59. According to the classification of Landis and Koch, a measure of 0.59 represents moderate agreement. Individual agreement for members of the panel of observers ranged from 0.28 (fair agreement) to 0.78 (substantial agreement). To calculate inter-observer agreement, the intra class correlation coefficient was calculated for the 10 members of the panel of observers. There was a measure of agreement of 0.89 (95% CI: 0.819–0.94). A commonly accepted minimum standard for such reliability coefficients is 0.7. 4 Discussion The protocol for preparation of the digital images led to the Newtom VG CBCT images being very small. Magnification of an image leads to interpolation and thus loss of image quality. Nevertheless, in the case of the Newtom VG CBCT, it was felt that a magnified image is likely to be used by a clinician in order to be clinically useful. Therefore the dedicated viewing software (NNT Report viewer v2.05) was used to produce a maximally magnified image. This was included in the study as well as the smaller, full size, image. One limitation of this study is that the images were not displayed in the conditions under which they would normally be viewed. This was so that the sources of the images could be disguised. The digital images would normally be displayed on dedicated viewing software which would enable magnification, reduction and adjustment of brightness and contrast. In the case of the conventional images, optical magnification would be possible. Also conventional images could be scanned and displayed at any convenient size. The brightness and contrast of the images is an important factor in the perception of the images and in the case of digital images can be changed within the software. Enhancement of images in this way has been carried out by other authors prior to image assessment. Various methods have been used. These include application of a brightness and contrast algorithm, allowing observers to adjust brightness and contrast themselves and presetting the brightness and contrast. In a study by Baksi, the author preset the brightness and contrast himself. This, however, represents only one person's perception of the ideal brightness and contrast. Other observers may not have the same preference and thus bias may be introduced. Borg and Gröndahl in 1996 allowed observers to change the brightness and contrast themselves. In another study, Borg and Gröndahl reported that allowing observers to adjust brightness and contrast had no effect on the final outcome. The analysis of results shows significant differences in image quality rating between the four CBCT image types at the level of the 95% confidence interval. The 3D Accuitomo 170 was rated most highly, followed by the Kodak 9000 3D, the Newtom VG at maximum magnification and then the Newtom VG at full size. This finding is consistent with that of Liang et al. who found that the 3D Accuitomo was rated most highly for image quality out of the five CBCT machines in their study. The reasons for this difference may be explained partly by the resolution of the machines. For CBCT machines the resolution is represented by voxel size, a larger voxel size leading to lower resolution. The Newtom VG has the lowest resolution, or largest voxel size, of the three CBCT machines with a voxel size of 0.3 mm. f f http://67.192.191.166/uploads/pdf/NewTom/NT_ENTIntlBrochure07.pdf . The user reported that the Newtom VG has a fixed resolution, which might be interpreted as suggesting some limitations in its efficacy. The later Newtom VG has two resolution options, 0.3 and 0.15 mm voxels. It is possible that had the higher resolution been available to us in this study, the image quality performance may have been better. Nevertheless, this explanation for the disappointing quality ratings for the Newtom VG is at odds with the findings of Liedke et al. These authors conducted a study to investigate the influence of voxel size on the diagnostic ability of CBCT to evaluate external root resorption. The authors used an iCat g g Imaging Sciences International, Hatfield, Pennsylvania, USA. CBCT machine which was set at voxel sizes of 0.2 mm, 0.3 mm and 0.4 mm. They concluded that a 0.3 mm voxel size gave the best compromise of diagnostic performance and X-ray exposure. Therefore, the disappointing image quality rating for the Newtom VG CBCT machine may not be fully explained by the larger voxel size. One can speculate that the processing of the images at the scanning centre, or the dedicated viewing software for the Newtom VG (NNT report viewer V2.05), may have had an important effect on observed image quality. Other factors in image quality of CBCT machines include object to recorder distance, signal to noise ratio, kV, mA, image receptor type, filtration, field of view size, artefacts, slice thickness and a number of subject related factors. The voxel sizes for the 3D Accuitomo 170, h h http://www.jmoritaeurope.de/productdatabase/server/assets/174/4/3_Accuitomo_170_EN_090904.pdf . and Kodak 9000 i i http://www.thedentalimagingcompany.co.uk/acatalog/K90003D_Brochure_EN.pdf . were almost identical at 0.08 mm and 0.076 mm, respectively. Despite this, the Kodak 9000 images were not rated as highly as the 3D Accuitomo 170 images. There was, however, noticeably more noise with the Kodak 9000 images and noise is another of the key factors in digital image quality. In other respects the images appeared similar and it seems likely that, in large part, this explained the difference in image quality rating between these two machines. In the lateral cephalometric view and the transymphyseal view, the beam is at 90° to the sagittal plane and records an image of the symphyseal region by superimposition of the lower left to lower right canine region. A comparison of these two techniques is therefore of interest. In respect of all statements combined, the lateral cephalometric technique scored very similarly to the transymphyseal technique (means of 2.95 and 2.77, respectively). The difference in mean rating was not significant at the level of the 95% confidence interval. These findings suggest that the transymphyseal radiograph may be a useful alternative to the lateral cephalometric view in general dental practice. The technique has a number of advantages. It uses equipment and materials readily available in most dental practices, it is relatively inexpensive and it is not necessary to refer to a specialist centre. Further the radiation exposure is limited only to the area of interest because the X-ray beam passes only through the symphyseal region. No special processing facilities are required. In this study the images were all presented in the same format. Each was an image of the midline of the mandible. Nevertheless, when making a comparison of imaging methods, the advantages and disadvantages of each should be taken into consideration. For example, the CBCT machines can obtain true cross-sectional images perpendicular to the curve of the jaw, free from superimposition. Therefore if two implants are planned for the lower canine regions, the exact area of the planned placement can be imaged. The transymphyseal and lateral cephalometric views are superimpositions of the anterior mandible which give an indication of the probable cross section of the canine regions but are less precise. On the other hand the radiation dose is an important consideration. The Effective Dose associated with the transymphyseal projection is around 1–2 μSv. This is a little less than a lateral cephalogram at 3 μSv and substantially lower than the CBCTs which are likely to have doses 10 times as great. Cost and availability are also important considerations. 5 Conclusions Seven different image types have been used to image the cross section of the midline of the mandible on four different X-ray phantoms. These results show that there was a difference in image quality between these different image types. These results also show differences in image quality between the three different CBCT machines and between the full size and magnified versions of one of the machines. Therefore it should not be assumed that all CBCT machines will produce images of the same quality. This study rated the Accuitomo 170 CBCT images above the Kodak 9000 CBCT images. These were in turn rated more highly than the Newtom VG images. Nevertheless, many other machines exist and the very rapidly changing field of CBCT technology means that these results will not remain current. Further studies will be necessary. A limitation of the study was that images from machines which produce true slices such as the CBCT machines were compared with images that were produced by superimposition such as the lateral cephalometric and transymphyseal techniques. The results of this study can therefore only be applied to images of the midline of the mandible. When planning implant placement in the lower canine regions, a true slice of the canine regions is clearly more useful than the midline view. Nonetheless, factors such as the radiation dose, cost and convenience should also be taken into consideration. Acknowledgements The authors are grateful to Mr Daniel Jepson of Plastic Formers Ltd, Denton, Manchester, UK for his invaluable help in the manufacture of the experimental phantoms. References 1. Feine J.S., Carlsson G.E., Awad M.A., Chehade A., Duncan W.J., Gizani S., et. al.: The McGill consensus statement on overdentures. Mandibular two-implant overdentures as first choice standard of care for edentulous patients. Gerodontology 2002; 19: pp. 3-4. 2. Hyland R., Ellis J., Thomason M., El-Feky A., Moynihan P.: A qualitative study on patient perspectives of how conventional and implant-supported dentures affect eating. Journal of Dentistry 2009; 37: pp. 718-723. 3. Cawood J.I., Howell R.A.: A classification of the edentulous jaws. International Journal of Oral and Maxillofacial Surgery 1988; 17: pp. 232-236. 4. Pigadas N., Simoes P., Tuffin J.R.: Massive sublingual haematoma following osseo-integrated implant placement in the anterior mandible. British Dental Journal 2009; 206: pp. 67-68. 5. Kalpidis C.D., Setayesh R.M.: Hemorrhaging associated with endosseous implant placement in the anterior mandible: a review of the literature. Journal of Periodontology 2004; 75: pp. 631-645. 6. Javed F., Romanos G.E.: The role of primary stability for successful immediate loading of dental implants. A literature review. Journal of Dentistry 2010; 38: pp. 612-620. 7. FGDP(UK) : Selection criteria in dental radiography.2nd edn.2004.Faculty of General Dental Practitioners (UK) Royal College of Surgeons of England 8. Horner K.: European guidelines on radiation protection in dental radiology.2004.European Commission 9. Harris D., Buser D., Dula K., Grondahl K., Harris D., Jacobs R., et. al.: E.A.O. guidelines for the use of diagnostic imaging in implant dentistry. A consensus workshop organized by the European Association for Osseointegration in Trinity College Dublin. Clinical Oral Implants Research 2002; 13: pp. 566-570. 10. Tyndall D.A., Brooks S.L.: Selection criteria for dental implant site imaging: a position paper of the American Academy of Oral and Maxillofacial radiology. Oral Surgery Oral Medicine Oral Pathology Oral Radiology and Endodontics 2000; 89: pp. 630-637. 11. Shelley A.M., Horner K.: A transymphyseal X-ray projection to assess the anterior edentulous mandible prior to implant placement. Dental Update 2008; 35: pp. 689-694. 12. Horner K., Drage N., Brettle D.S.: 21st century imaging.2008.Quintessence Publishing Co. Ltd.London 13. Gardner M.J., Altman D.G.: . British Medical Journal (Clinical Research Edition) 1986; 292: pp. 746-750. 14. Cohen J.: A coefficient of agreement for nominal scales. Educational Psychological Measurement 1960; 20: pp. 37-46. 15. Landis J.R., Koch G.G.: The measurement of observer agreement for categorical data. Biometrics 1977; 33: pp. 159-174. 16. Lohr K.N., Aaronson N.K., Alonso J., Burnam M.A., Patrick D.L., Perrin E.B., et. al.: Evaluating quality-of-life and health status instruments: development of scientific review criteria. Clinical Therapeutics 1996; 18: pp. 979-992. 17. Borg E., Attaelmanan A., Grondahl H.G.: Subjective image quality of solid-state and photostimulable phosphor systems for digital intra-oral radiography. Dentomaxillofacial Radiology 2000; 29: pp. 70-75. 18. Borg E., Grondahl H.G.: Endodontic measurements in digital radiographs acquired by a photostimulable, storage phosphor system. Endodontics & Dental Traumatology 1996; 12: pp. 20-24. 19. Baksi B.G.: Measurement accuracy and perceived quality of imaging systems for the evaluation of periodontal structures. Odontology 2008; 96: pp. 55-60. 20. Borg E., Grondahl H.G.: On the dynamic range of different X-ray photon detectors in intra-oral radiography. A comparison of image quality in film, charge-coupled device and storage phosphor systems. Dentomaxillofacial Radiology 1996; 25: pp. 82-88. 21. Liang X., Jacobs R., Hassan B., Li L., Pauwels R., Corpas L., et. al.: A comparative evaluation of Cone Beam Computed Tomography (CBCT) and Multi-Slice CT (MSCT). Part I. On subjective image quality. European Journal of Radiology 2010; 75: pp. 265-269. 22. Liedke G.S., da Silveira H.E., da Silveira H.L., Dutra V., de Figueiredo J.A.: Influence of voxel size in the diagnostic ability of cone beam tomography to evaluate simulated external root resorption. Journal of Endodontics 2009; 35: pp. 233-235. 23. van Daatselaar A.N., van der Stelt P.F., Weenen J.: Effect of number of projections on image quality of local CT. Dentomaxillofacial Radiology 2004; 33: pp. 361-369. 24. The Sedentexct Project [internet]. Manchester: The University of Manchester; 2009. Available from: www.sedentexct.eu . 25. Ludlow J.B., Davies-Ludlow L.E., White S.C.: Patient risk related to common dental radiographic examinations: the impact of 2007 International Commission on Radiological Protection recommendations regarding dose calculation. Journal of the American Dental Association 2008; 139: pp. 1237-1243.

Related Articles

Leave A Comment?

You must be logged in to post a comment.