Radiographic Methods for the Detection

Transcription

1 Radiographic Methods for the Detection of Progressive Alveolar Bone Loss* Marione. Jeffcoat 367 Intraoral transmission radiographs have been the primary diagnostic method for the assessment of bone support as well as for the detection and measurement of osseous changes due to Periodontitis. The purpose of the present paper is three-fold. The first is to review radiographie techniques for the assessment of periodontal disease progression, presenting the strengths and weaknesses of each method while placing special emphasis on digital subtraction radiography. The second purpose is to present data from a recent study that compared the ability of digital subtraction radiography and automated attachment level probing to detect the same active sites. Thirty Periodontitis patients and eight control patients were studied over a 6-month period. The results indicate that when these two sensitive methods for the assessment of progressive Periodontitis were used there was concordance between the presence or absence of probing attachment loss and bone loss in 82.1% of the sites. The final goal of this paper is to present future directions for the quantitative analysis of digital radiographie images. / Periodontol 1992; 63: Key Words: Bone resorption/diagnosis; bone/radiography; periodontal diseases/diagnosis; radiography, digital subtraction; periodontitis/diagnosis; periodontal attachment. The ability to detect lesions in the alveolar bone has greatly improved since the days when the dentist simply interpreted unstandardized conventional radiographs. Interpretive radiography is a relatively crude tool which does not register alveolar bone loss or gain until 30% to 50% of the bone mineral is destroyed. The reasons for the lack of sensitivity of conventional radiography are many. First, foreshortening or elongation of the radiographie image may apparently "grow or destroy" alveolar bone independent of any real change in the alveolar bone support. Second, variations in the contrast and density of a radiograph, caused by poor control of film processing, or variations in KVp or exposure time, may "burn out" the alveolar crest independent of changes in the alveolar bone. The third source of difficulty in detecting osseous change lies in the two-dimensional nature of the conventional transmission radiograph. Since the image is a two-dimensional mapping of the three-dimensional teeth and alveolar bone, many anatomic structures, such as tooth roots or cortical plates, may overlie lesions in the trabecular bone. The resultant radiograph is a complex image and it is most difficult for the clinician to detect bony small changes in the face of the overlying unchanging structures. Over the past two decades many investigators have developed radiographie methods that improve our ability to Department of Periodontics, University of Alabama School of Dentistry, Birmingham, AL. detect and measure changes in the alveolar bone. Each of these methods addresses one or more of the difficulties as described above. The inherent to interpretive radiography goal of this paper is to review these semi-quantitative and quantitative methods for determining bone loss using conventional intraoral transmission radiographie images. The results of a new study comparing disease activity using sequential radiography and probing attachment loss methods are also presented. Finally, future directions in the radiographie assessment of alveolar bone change are discussed. Early attempts to compensate for angular discrepancies include use of a radiographie grid and the Schei ruler. The radio-opaque grid is usually spaced in 1 mm markings and is placed directly on the film. When the radiograph is exposed, the 1 mm grid is superimposed on the radiographie image of the anatomic area of interest providing a reference for measurement of bone height. Although the technique is simple to use, the results can be misleading. Since the grid is placed directly on the film, the image of the grid does not distort when misangulation of the beam results in foreshortening or elongation of the teeth. In effect, the grid is a ruler, placed on the film to facilitate direct measurement of bone height without compensation for angular discrepancies. Furthermore, the grid lines may obscure the area of interest. A satisfactory alternative is a grid printed on transparent film that may be placed over the radiograph to facilitate measurement.

2 J Periodontol 368 April RADIOGRAPHIC DETECTION OF ALVEOLAR BONE LOSS A second alternative is to utilize the radiographie view that produces minimal distortion of the tooth-bone height relationship. The bite-wing radiograph achieves this aim because the central ray is perpendicular to the tooth and the film, resulting in little foreshortening or elongation of the radiographie image. Furthermore, vertical bite-wing films will facilitate visualization of the osseous crest in order to take measurements of bone loss in millimeters from the cemento-enamal junction1 (Fig. 1). In striking contrast, the Schei ruler is a one-dimensional method for the compensation of geometric discrepancies.2 Bone height is expressed as a ratio of the root length to compensate for foreshortening or elongation of the image.3 The classical Schei ruler has 5 or 10 gradations with which the clinician can grade the extent of bone loss. This categorical grading of the bone loss limits the sensitivity of the method. If, for example, a Schei Ruler is used with five gradations, the investigator can only measure bone loss in 20% increments. This categorical analysis can be misleading because a site may be actively losing bone, but remain within a given category. Thus, a simple analysis of bone loss over time may appear to be stable although up to 20% of the bone around a given tooth has been lost. This limitation of the original Schei Ruler approach has been circumvented with the advent of digitizing devices.4 Such digitizers are usually coupled to microcomputers and can measure the length of the tooth root and the bone height with a resolution of one-hundredth of a millimeter. This virtually continuous "Schei ruler" is mechanically and analytically capable of detecting as little as 1.5% bone loss (Fig. 2). The second limitation of the original Schei Ruler apis that it provides a relative measure of the extent of bone loss; absolute measurements in terms of millimeters proach (Supplement) Computerized Schei ruler. A radiograph is digitized and the of the CEJ (A), alveolar crest (B), and root apex (C) determined interactively. The computer calculates the percent bone loss relative to the length of the root surface (A to B/A to C x 100). Figure Figure 1. Vertical bite-wing radiograph. The distance from the CEJ to the alveolar crest (arrows) is measured in millimeters location not provided, since the ratio is the method used to correct for angulation errors. More recent methods have been developed to account in part for angulation errors.5'6 are These algorithms may be simply thought of as two-dimensional Schei Rulers. Reference points in a well angulated radiograph and the radiograph to be corrected are identified and an aphine (planar)5 or rubber sheet6 warp algorithm used to correct the image of the second film. Thus, after correction, the length of the tooth roots of a first (reference) and second film will be nearly identical obviating the need for ratios (Fig. 3). After the warp algorithms are applied, direct measures of bone loss in terms of millimeters may be used. A further advantage of the warp algorithm is evident when the radiographs are entered into an image processing work station. The corrected image may be displayed with the geometry of the reference film facilitating comparison of the two examinations. Furthermore, use of contrast and brightness correction algorithms as described by Ruttimann et al.7 will correct for errors in brightness and contrast of the films that can occur due to variations in voltage or film processing. None of the measurement methods described above deal with one of the major difficulties in interpretive radiography. That is, the clinician must be able to identify the lesion against the background of superimposed anatomic structures including the tooth roots and cortical plates of bone. Subtraction radiography was introduced in the early 1980s as a method to facilitate visualization of the areas of osseous gain or loss. Photographic subtraction radiography has been used in medicine since the 1930s, and several groups introduced subtraction radiography to dentistry.8'9 The concept underlying subtraction radiography is that an image processing computer subtracts all unchanging structures from

3 Volume 63 Number 4 JEFFCOAT Figure 3. Computer correction of radiographie errors due to angulation. Note the length of the mesial root of 30 is 11.6 mm in the upper left radiograph, but 12.8 mm in the elongated upper right radiograph. The computer algorithm produces the image shown in the lower left, which has a root length of mm. taken at two different examinaneutral gray background in the areas that have not changed; by convention, areas of bone loss are shown in dark shades and areas of bone gain are shown in light shades (Fig. 4, lower left panel). The ability of clinicians to detect minute bony changes using subtraction radiography has been demonstrated by Hausmann et al.9 and also Jeffcoat et al.10 using artificially induced lesions in skulls. Clinicians were able to accurately identify lesions in which less than 5% bone has been removed better than 90% of the time.9 In separate experiments clinicians were able to detect the presence of artificial lesions with a sensitivity of 91.3% and were able to rule out lesions with a specificity of 95.7%.10 Further studies have shown that computer enhancement of the areas of bone loss and bone gain using computer generated pseudocolor can further facilitate visualization of the areas of change.11,12 In order to successfully perform subtraction radiography, the two radiographs must be taken with minimal angular discrepancy, and with similar brightness and contrast. Several methods have been successfully developed to standardize image geometry. The first class of methods utilizes custom Stents to physically couple the tooth, film, and xray tube.13 The second class of methods uses a cephalostat to couple the patient to the x-ray tube; the aphine warp algorithm is used when needed to correct for planar geometric errors, such as those that can occur with film tilt.10 The gamma correlation algorithm of Ruttimann et al.7 may be successfully used to correct for discrepancies in brightness and contrast. One of the drawbacks in the clinical interpretation of subtraction images is that in a high quality image reference, landmarks, such as the teeth and root surfaces, have been a set of two radiographs tions. The result is a 369 Figure 4. Digital subtraction radiography. The upper left and right radiographs were taken 6 months apart. The lower left is an unenhanced subtraction image. Bone loss is represented as dark shades of gray (arrows) in eresiai and subcrestal bone. The lower right image is the result of the morphologic method. The computer has detected the sites ofbony change, computed their size, and superimposed the area of change on the original radiograph. In the orginal radiograph the areas of bone loss are shown in shades of red. subtracted and are virtually invisible. Thus, the clinician may have difficulty localizing an area of bone loss or bone gain to a particular tooth site or infrabony defect. This limitation has recently been assessed with the introduction of a morphologic subtraction technique.14 In brief, the area of interest is isolated using an operator-assisted variable binary threshold. A morphologic open filtering operation is then performed to remove background noise from the subtraction image so that the area of change appears as white against a black background. This morphologic binary subtraction image is combined with the original radiograph so that the location of the area of change may be visualized (Fig. 4, lower right panel). Calibration studies in skulls have shown that the morphologic method can be used to detect and correctly localize lesions less than 25 mg in size with a specificity of 97.7% and a sensitivity of 94.4%. One of the concerns about the extreme sensitivity in digital subtraction radiography has been focused on the possibility of detecting changes in bone that are either false positives or that represent physiologic remodelling rather than pathologic alveolar bone résorption. In an attempt to address this problem we have recently completed a clinical trial to address the relationship of bone loss as detected by digital subtraction radiography and periodontal disease progression as measured with a sensitive automated periodontal probe. MATERIALS AND METHODS with evidence of alveolar bone loss were studied. Patients had not received periodontal therapy and Thirty patients

4 370 RADIOGRAPHIC DETECTION OF ALVEOLAR BONE LOSS Table 1. Association Between Subtraction Radiography and Progressive Probing Attachment Loss in Healthy Subjects Subtraction Radiography Automated Probe Inactive Active Inactive 40* 0 Active 0 0 *Number of sites. had pockets of 5 mm or greater at the experimental sites. In addition, eight patients without evidence of Periodontitis were studied. These healthy subjects did not have radiographic evidence of bone loss nor did they have probing attachment loss greater than 1 mm. Patients who had diseases (such as diabetes) or were taking medications (such as antibiotics) that could potentially modify the course of periodontal disease progression were excluded from study. Patients at risk from periodontal probing or who required antibiotic prophylaxis prior to probing were also excluded. This study was approved by our Investigational Review Board and informed consent was obtained in writing. In the Periodontitis patients, five posterior tooth sites with 30% to 50% bone loss and 5 mm or greater probing depth were selected for study. In the control subjects five posterior tooth sites were selected for study. At the start of the study each patient received a supragingival prophylaxis to facilitate attachment level measurements. No root planing was performed at this time. The progression of Periodontitis was followed over a 6- month period. Probing attachment level measurements relative to the cemento-enamel junctions (CEJ) were used to quantify the progression of the loss of soft tissue attachment and digital subtraction radiography was used to detect sites of progressive bone loss. Attachment level measurements were taken using an automated periodontal probe with CEJ.15 In brief, this instrument automatically enters the pocket when the stroke is activated by a foot switch. The instrument retracts when the resistance at the base of the pocket is equal to the preset force (35 gm). The instrument monitors tip acceleration and extension. On board firmware analyzes the tip acceleration so that the probe is able to detect the catch of the CEJ, and probing attachment level relative to the CEJ is displayed. Previous experiments have shown that the repeatability of this instrument is better than 0.2 mm with a bias error of was observed in 0.09 mm This level of repeatability anterior and posterior sites with either deep or shallow probing pocket depths.17 Attachment level measurements were taken in test sites at baseline 0, 2, 4, and 6 months. The cum-sum method18 was used to detect active sites. A site was considered active when the cumulative sum of the change exceeded twice the error of the method (0.2 mm x 2 = 0.4 mm). Standardized radiographs were taken using the cephalostat method at baseline and 6 months as previously described. 10 Radiographs were digitized, corrected for variations in contrast and density, corrected for planar variations in J Periodontol April 1992 (Supplement) Table 2. Association Between Subtraction Radiography and Progressive Probing Attachment Loss in Periodontitis Patients Subtraction Radiography Automated Probe Inactive Active Inactive 73* 18 spec 80.2% Active sens 71.6% *Number of sites. angulation where required, and subtracted. The morphologic subtraction method14 was used to isolate any areas of bone change and superimpose the lesion on the original radiograph. Each of the five test sites was graded as to the presence of bone loss or whether no bone loss was detectable. Statistical Analysis Two by two tables were constructed to view the association of bone loss as measured by subtraction radiography with progressive probing attachment loss measured with the automated probe. Sensitivity and specificity were calculated and the standard error was corrected for the effects of clustering of sites within patients.19 RESULTS During the course of the study one patient was diagnosed with mitral valve prolapse and further probing attachment level measurements were not taken. No other adverse experiences were reported. Table 1 presents a cross tabulation of progressive bone loss and probing attachment loss in the control patients. No site in a control patient lost bone or probing attachment over the 6-month study period. Table 2 presents a cross tabulation of progressive bone loss versus progressive attachment loss in the Periodontitis patients. When bone loss by subtraction is considered, 38% of the sites were active. These sites were found in 76% of the patients. When probing attachment level was considered, 35% of the sites were active and active sites were observed in 79% of the patients. A concordance of the results of the subtraction radiography and automated probing attachment level examinations was found in 82.1% of the sites. In the Periodontitis patients, the sensitivity of subtraction radiography in its association with progressive probing attachment loss was.716 ±.01; the specificity was.804 ±.006. DISCUSSION The present study indicates that when precise methods for the detection of progressive Periodontitis are used, there is a high degree of concordance between the sites of disease activity detected by electronic attachment level probing and by digital subtraction radiography. Possible reasons for lack of concordance include false positive and negative tests, as well as disease progression that was detectable only in either the soft tissue attachment or bone. It was comforting to observe that the sites in the control patients without peri-

5 Volume 63 Number 4 JEFFCOAT 371 odontitis did not pass the threshold criteria for disease progression even when the relatively stringent criteria of 0.4 mm of probing attachment loss or bone loss detectable on subtraction radiography were used. It is important to note that the patient population studied were patients with sites of untreated Periodontitis. This study was not designed to determine the temporal relationship between probing attachment loss and bone loss in early Periodontitis. Possible approaches to answering these questions would include long-term clinical trials in which both bone loss and probing attachment loss are monitored in patients at risk for Periodontitis. Such a study would be lengthy and costly, since many closely-spaced examinations would be required to ascertain the extent of a between the onset of probing attachment loss and bone loss. Clearly radiation safety will be a factor in determining the interval between examinations. With the advent of direct digital radiographie machines, the radiation dose to the patient can be reduced between 80% and 96%. Such machines eliminate the need for radiographie film and can eliminate the errors inherent to film processing. Furthermore, when such systems are coupled to image processing computers and mass storage devices, such as optical disks, the comparison of multiple radiographie examinations will be facilitated. The future of digital radiography lies in thorough exploitation of the information contained within the gray level patterns within the subtraction image. Interpretive subtraction radiography relies on an experienced clinician to "read" the image and detect sites of bony change. The morphologic method14 improves on this by isolating the area of change from the background noise of the image and superimposing the area of change on the original radiograph. This method does require use of the gray level information within the image. The first practical use of the gray level technique was found in the CADIA method (computer aided digital image analysis) introduced by Braegger et al.20 This method has been applied to the study of Periodontitis and dental implants. The method requires prior definition of areas of interest and presents bony change in terms of CADIA units which are related to the gray level change. Other papers21"23 have introduced the concept of the use of a reference wedge for the calculation of absolute changes in bony volume or mass. A recently completed study in our laboratory indicated that when the mean gray level in the area of osseous change identified with the morphologic technique is related to a reference wedge, the mass of the lesion may be calculated with a high validity.24 The correlations between actual lesion size and computed lesion size exceed 90%. Digital radiography is a rapidly changing field. The contributions of many research groups to the development of these state-of-the-art methods have been recently reviewed25 time lag and are now being applied to clinical trials and are in use in the evaluation of patient care in several centers. Future studies which marry the power of the digital imaging workstation with the convenience and low radiation dose of the direct digital radiography systems have the potential to allow more precise measurement of bony change with less risk to the patient than was previously possible. Acknowledgment Supported by NIH Grant DE REFERENCES 1. Hausmann E, Allen, Christersson L, Genco R. Effect of x-ray beam vertical angulation on radiographie alveolar crest level measurement. J Periodont Res 1989; 24: Schei O, Waerhaug J, Lovdal A, Arno A. Alveolar bone loss as related to oral hygiene and age. J Periodontol 1959; 30: Lavstedt S. A methodological-roentgenologic investigation on alveolar bone loss. Acta Odont Scand 1973; 33: Jeffcoat MK, Williams RC. Relationship between linear and area measurements of radiographie bone levels using simple computerized techniques. J Periodont Res 1984; 19: Jeffcoat MK, Jeffcoat R, Williams RC. A new method of the comparison of bone loss measurements on non-standardized radiographs. J Periodont Res 1984; 19: Webber RL, Ruttimann UE, Groehuis RAJ. Computer correction of projective distortions in dental radiography. J Dent Res 1984; 63: Ruttimann UE, Webber RL, Schmidt E. A robust digital method for film contrast correction in subtraction radiography. J Periodont Res 1986; 21: Webber RL, Ruttimann UE, Grondhal HG. X-ray image subtraction as a basis for the assessment of periodontal changes. J Periodont Res 1982; 17: Hausmann E, Christersson L, Dunford R, Wiskesjö U, Phylo J, Genco RJ. Usefulness of subtraction radiography in the evaluation of periodontal therapy. J Periodontol 1985; 56(suppl.): Jeffcoat MK, Reddy MS, Webber RL, Williams RC, Ruttimann UE. Extraoral control of geometry for digital subtraction radiography. J Periodont Res 1987; 22: Braegger U, Pasquali L. Color conversion of alveolar bone density changes in digital subtraction images. / Clin Periodont 1989; 16: Reddy MS, Bruch JM, Jeffcoat MK, Williams RC. Contrast enhancement as an aid to interpretation in digital subtraction radiography. Oral Surg Oral Med Oral Pathol 1991; 71: McHenry K, Hausmann E, Wikesjö U, et al. Methodological aspects and quantitative adjuncts to computerized subtraction radiography. J Periodont Res 1987; 22: Jeffcoat MK, Page R, Reddy MS, et al. Use of digital radiography to demonstrate the potential of naproxen as an adjunct in the treatment of rapidly progressive Periodontitis. J Periodont Res 1991; 26: Jeffcoat MK, Jeffcoat R, Captain K. A periodontal probe with automated cemento-enamal junction detection-design and clinical trials. IEEE Trans Biomed Eng 1991; 38: Jeffcoat MK, Reddy MS. Progression of longitudinal change in Periodontitis. J Periodontol 1991; 62: Jeffcoat MK, Jeffcoat RL, Captain K, Reddy M, Williams RC. Attachment level probing with automated CEJ detection: Clinical trials. J Dent Res 1989; 68 (Spec. Issue):236 (Abstr. 440). 18. Aeppli DM, Pihlstrom BL. Detection of longitudinal change in Periodontitis. J Periodont Res 1989; 5: Hujoel PP, Moulton LH, Loesche WJ. Estimation of sensitivity and specificity of site-specific diagnostic tests. J Periodont Res 1990; 25: Braeggar U, Pasquali L, Weber H, Kornman KS. Computerized densitometric image analysis (CADIA) for the assessment of alveolar

6 372 RADIOGRAPHIC DETECTION OF ALVEOLAR BONE LOSS bone density changes in furcations. J Clin Periodontol 1989; 16: Vos, Janssen PTM, Van Aken J, Heethar RM. Quantitative measurement of periodontal bone changes by digital subtraction. / Periodont Res 1986; 21: Ruttimann UE, Webber R. Volumetry of localized bone lesions by subtraction radiography. J Periodont Res 1987; 22: Webber RL, Ruttimann UE, Heaven TJ. Calibration errors in digital subtraction radiography. J Periodont Res 1990; 25: Jeffcoat MK, Reddy MS, Van den Berg HR, Bertens E. Quantitative J Periodontol April 1992 (Supplement) digital subtraction radiography for the assessment of peri-implant bone change. J Clin Implantology Accepted for publication. 25. Hausmann E. A contemporary perspective on techniques for the clinical assessment of alveolar bone. J Periodontol 1990; 61: Send reprint requests to Dr. Marjorie K. Jeffcoat, Department of Periodontics, University of Alabama School of Dentistry, UAB Station, Birmingham, AL