Reliability and validity issues of laser fluorescence measurements in occlusal caries diagnosis

×
Rating : Rate It:
 
Embed :   
Post a comment
    Post Comment on Twitter
Comments:  



  Notes
 
 
Slide 1 : Reliability and validity issues of laser fluorescence measurements in occlusal caries diagnosis* Hanna M. Alwas-Danowska1Alphons J.M. Plasschaert2Stanislaw Suliborski1 Emiel H. Verdonschot21Department of Propaedeutics and Diagnostics in Stomatology, Medical University of Lodz, Poland, 2Department of Cariology and Endodontology, University Medical Centre Nijmegen, The NetherlandsKey words: Occlusal caries; Laser fluorescence; Visual inspection; Diagnosis*Published in: Journal of Dentistry, 30 (2002) 129-134
Slide 2 : 1. Introduction In 1998 a new diagnostic tool for detecting occlusal caries lesions was introduced (DIAGNOdent®, KaVo, Bibberach, Germany). This tool is based on fluorescence caused by porphyrines present in carious tissues. Laser light with a wavelength of 655 nm is directed onto the occlusal surface of a premolar or molar tooth using a fibre-optic probe. The light penetrates the enamel and dentine, and part of it will be transferred into light with a larger wavelength, a phenomenon called fluorescence. The fluorescent light is transported back to the device through specific fibres in the probe and transported to the device. Here, the intensity of the fluorescent light is measured and converted into digits which are displayed on a monitor. A comprehensive description of the DIAGNOdent® can be found elsewhere. It has been recommended that the DIAGNOdent® should be used for monitoring occlusal caries lesions (8, 10). For this purpose it is imperative that devices manufactured in successive years produce the same measurements given the same patients and teeth, irrespective of the depth or size of the caries lesion. In this investigation, the reproducibility of two DIAGNOdent® devices and two observers was established. In addition, the relationship between laser fluorescence measurements and lesion depth and size was evaluated.
Slide 3 : Materials and methods 2.1. In vivo part of the investigation The reproducibility of DIAGNOdent® measurements was established in a clinical study employing 13 students aged 18-25 year, who volunteered to participate. Permission from the local ethical committee was obtained. In each patient one of the investigators selected two to four permanent molar teeth. Only molars which showed no cavitation, restoration or fissure sealant at the occlusal surface were included. The occlusal surfaces were thoroughly cleaned with a rubber cup and a non-fluorescent polishing paste (Nupra fine mint, Dentsply, York PA, USA). At each occlusal surface the same investigator selected one site to be measured. The sites were recorded on two separate data collection sheets, one for each observer. The observers conducted DIAGNOdent® measurements from the indicated sites using two devices, one DIAGNOdent® manufactured in 1998 and one manufactured in 1999.
Slide 4 : The latter device was calibrated against a reference object, and the obtained value was exactly the same as the one mentioned in the documents which were delivered with the device. Such reference object was not customary delivered with the 1998 DIAGNOdent® and, hence, no calibration could be performed.Only one new probe tip (tip A) was used throughout the investigation to avoid variation from the wear of the fibres caused by multiple application. Using compressed air, the teeth were dried prior to each measurement. Following the manufacturer’s instructions, the device was calibrated to the natural fluorescence of a patient’s teeth by conducting a reference measurement at the labial surface of tooth 11. Subsequently, the indicated sites were measured by placing the probe tip perpendicular to the occlusal surface and by making a rotational movement. The DIAGNOdent® measures the intensity of the fluorescent light and converts this measurement into a value ranging from 0 to 99. During measurement the peak value is displayed, and the measurement procedure was continued until the peak value remained stable. For each site the peak value was recorded. All sites were measured independently by two observers using both devices.The reproducibility of DIAGNOdent® measurements from both devices and both observers was expressed in intraclass correlation coefficients.
Slide 5 :  2.2. In vitro part of the investigation For this part of the investigation 50 molar teeth were selected from a large batch. All teeth had been stored in tap water without disinfection fluid for six months. The included teeth were either completely sound or showed uncavitated caries lesions at the occlusal surface upon visual inspection. Using waterproof ink, the buccal and lingual cusps were marked such that the two marks and the site to be measured were on a straight line. The measurements were carried out independently by two observers. They estimated the presence and depth of a caries lesion at the selected sites by visual inspection using an ordinal rating scale: 0 = no caries lesion present; 1 = caries lesion restricted to the enamel; 2 = caries lesion reaching the dentino-enamel junction; 3 = caries lesion in dentine. Subsequently, DIAGNOdent® measurements were obtained from the sites in the same way as described in the in vivo part of this investigation.Because the measurement of the volume of the enamel and dentinal part of a lesion is extremely difficult, the area of the lesion parts measured from hemisections was considered an appropriate estimation of the size of the lesion parts.
Slide 6 : The teeth were embedded in a translucent methacrylate and hemisected in bucco-lingual direction parallel to the marks, immediately adjacent to the measured site. The hemisection showing the largest area of the lesion was used for validation. To calculate the depth and area of the lesion a metal rod of known length (5,87 mm) was placed on the section in the region of the pulp chamber, a colour slide was taken (Figure 1a), and converted to a digital image.Using measuring software (Scion Image, NIH, Bethesda, USA) the number of pixels on the length of the rod was determined which allowed for the calculation of the length and width of a pixel in millimetres. With the dimensions of a pixel known, the depth in millimetres of a lesion could be measured. To determine the area of the enamel part of the lesion the boundary of the lesion in enamel was marked (Figure 1b) and the number of pixels inside the boundary were counted. Given the dimensions of a pixel the area of the enamel part of the lesion was obtained in square millimetres. Likewise, the area of the dentinal part of the lesion was measured (Figure 1c). All measurements were conducted independently by two observers.
Slide 7 : The reproducibility of all diagnostic measurements were expressed in intraclass correlation coefficients. The validity of visual inspection and DIAGNOdent® measurements was expressed in the Pearson correlation between the diagnostic measurements and the depth and area of the lesions measured from the hemisected teeth. To evaluate the ability of both methods to detect dentinal caries, the sensitivity and specificity of visual inspection and DIAGNOdent® measurements were determined. The cut-off for the DIAGNOdent® was between values 20 and 21 (11). The sensitivity is the proportion of dentinal lesions identified by the histological validation that were correctly identified by the diagnostic method. The specificity is the proportion of sound surfaces and surfaces containing caries limited to the enamel identified by the histological validation that were correctly identified by the diagnostic method. A Receiver Operating Characteristic (ROC) analysis was conducted to compare the diagnostic performance of visual inspection to that of the DIAGNOdent® measurements. In an ROC analysis, using varying cut-offs at the diagnostic rating scale, the sensitivity of a diagnostic test is plotted against 1 minus specificity (14, 15). The area under an ROC curve of a diagnostic method represents its performance. An area test (16) was applied to test the significance of the difference between the areas under the ROC curve (Az) from visual inspection and DIAGNOdent® measurements.
Slide 8 : Fig. 1a
Slide 9 : Fig. 1b
Slide 10 : Fig. 1c Figure 1a,b,c.. To determine the actual depth and area of a caries lesion from a hemisected tooth, a digital image was produced containing a metal rod of known length (a). In this image the boundaries of the enamel (b) and dentinal part (c) of the lesion were manually set. The number of pixels and the calculation of the area were accomplished using an image software programme.
Slide 11 : 3. Results 3.1. Results from the in vivo investigationDIAGNOdent® measurements were obtained from 45 molar teeth in 13 patients. The means and standard deviations are contained in table 1. Intraclass correlation coefficients between both DIAGNOdent® devices were 0.90 for observer 1, and 0.88 for observer 2. The intraclass correlation between both observers was 0.85 using the DIAGNOdent® produced in 1998, and 0.90 using the DIAGNOdent® produced in 1999. 
Slide 12 : Table 1. Means and standard deviation from caries diagnostic measurements from occlusal surfaces.  Diagnostic measurement Observer 1 Observer 2 Mean S.d. Mean S.d. in vivo investigationDIAGNOdent® 1998 27.3 25.7 23.1 23.8DIAGNOdent® 1999 17.5 19.9 17.1 22.9in vitro investigationVisual inspection 1.39 0.95 1.55 1.29DIAGNOdent® 1998 38.6 30.3 37.5 28.1DIAGNOdent® 1999 41.8 31.2 44.6 32.6Lesion depth (mm) 1.40 1.11 1.35 1.02Enamel lesion area (mm2) 1.39 1.41 1.33 1.52Dentinal lesion area (mm2) 0.57 1.45 0.55 1.42
Slide 13 : 3.2. Results from the in vitro investigationDuring the study remnants of a fissure sealant were discovered in one molar tooth, which was excluded from the sample. The results were therefore based on measurements from 49 teeth. The means and standard deviations are contained in table 1.The interobserver agreement of visual inspection, expressed in the intraclass correlation coefficient, was 0.59. The intraclass correlation between both DIAGNOdent® devices was 0.97 for observer 1, and 0.94 for observer 2. The correlation between both observers was 0.96 using the 1998 DIAGNOdent®, and 0.89 using the 1999 device. The intraclass correlation coefficients between both observers for measuring the depth of the caries lesions from the sections, the area in enamel and the area in dentine were 0.96, 0.97 and 0.99, respectively.
Slide 14 : The specificity of visual inspection was higher than that of the DIAGNOdent® devices, whereas the sensitivity of the DIAGNOdent® devices exceeded that of visual inspection (table 2). The area under ROC was highest for the visual inspection, with Az-values of 0.91 and 0.81 for observers 1 and 2, respectively. The Az-values for the DIAGNOdent® measurements were in a small range of 0.80 – 0.82 for both observers and both devices. ROC area tests revealed no statistically significant differences between visual inspection and DIAGNOdent® measurements for both observers (P>0.05).Table 3 contains the Pearson correlation coefficients between the diagnostic measurements, and the depth of the caries lesions and area of the enamel and dentinal part of the lesion. The largest correlation was obtained for visual inspection by observer 1 (r=0.72). The correlation coefficients between DIAGNOdent® measurements and the area of the enamel part of the lesion systematically exceeded those between the DIAGNOdent® measurements and the dentinal part of the lesion, indicating that the size of small caries lesions can be better predicted by laser fluorescence measurements than the size of the larger, dentinal lesions.
Slide 15 : Table 2. The performance of two DIAGNOdent® devices and visual inspection by two observers, expressed in sensitivity, specificity and the area under the ROC-curve (Az), in diagnosing occlusal caries lesions affecting the dentine.  Diagnostic method Observer Sensitivity Specificity Az   Visual inspection 1 0.40 0.94 0.91 Visual inspection 2 0.60 0.88 0.83DIAGNOdent®-1998 1 0.93 0.59 0.81DIAGNOdent®-1998 2 0.93 0.53 0.81DIAGNOdent®-1999 1 1.00 0.50 0.82DIAGNOdent®-1999 2 0.93 0.47 0.80
Slide 16 : Table 3. Pearson correlation coefficients (r) expressing the association between visual inspection and DIAGNOdent® measurements, and the depth of the caries lesions, the area of the enamel part and the area of the dentinal part of the caries lesions.  Diagnostic methd Observer Depth Area enamel part Area dentinal part r r r Visual inspection 1 0.72 0.64 0.53Visual inspection 2 0.44 0.33 0.36DIAGNOdent®-1998 1 0.53 0.54 0.45DIAGNOdent®-1998 2 0.48 0.47 0.39DIAGNOdent®-1999 1 0.52 0.52 0.40DIAGNOdent®-1999 2 0.50 0.49 0.38
Slide 17 : 4. Discussion A diagnostic tool should be reliable (reproducible) and valid. The interobserver reliability of the DIAGNOdent® measurements was superior to that of visual inspection. The intraclass correlation coefficients (ICC) between both DIAGNOdent® devices and between the two observers (in vivo: 0.88
Slide 18 : This can be explained by the relatively longer distance that the photons from the fluorescent light have to travel when excitated in carious dentine before they reach the tip. The longer the distance the higher the probability that these photons are scattered or absorbed by the dentine and overlying enamel. This indicates that the DIAGNOdent® is particularly useful to detect small, superficial caries lesion, hence lesions restricted to the enamel. In the absence of enamel which conceals the dentinal lesion, i.e. in the case of cavitation, the fluorescence from carious dentine could be improved, which should result in an increased correlation between DIAGNOdent® measurements and dentinal part of the lesion. However, the detection of cavitated lesion at the occlusal surface of (pre)molar teeth is the domain of visual inspection, rather than that of the DIAGNOdent®. Table 2 indicates that a high specificity for visual inspection is combined with a low sensitivity, whereas the reverse is the case for the DIAGNOdent®. The DIAGNOdent® measurements were validated with a threshold between “carious” and “sound” at the dentino-enamel junction, and with the DIAGNOdent® threshold between values 20 and 21. Application of different thresholds will result in other sensitivities and specificities. Compared to the results published by Lussi et al. (8), who placed the cut-off between DIAGNOdent® values 10 and 11, the specificity of the DIAGNOdent® measurements found in our study was higher, but the sensitivity was lower. By adopting a lower threshold, more dentinal caries lesions would be detected at the cost of an increased number of false-positives. A recent multi-centre in vivo study on the performance of the DIAGNOdent® reported a sensitivity of 0.92 and a specificity of 0.86 (13), with the cut-off between 20 and 21 for unaffected and affected dentine, respectively.
Slide 19 : The sensitivity is quite high, indicating that almost all occlusal carious lesions were detected, and the specificity those in most may appear adequate. One should bear in mind, however, that the use of the DIAGNOdent® in a low caries prevalence population, such as Western countries, will lead to high rates of false-positive diagnoses and, subsequently, to overtreatment. If, for example, the prevalence of concealed occlusal caries lesions in 1000 occlusal surfaces is 0.05, indicating that one out of twenty surfaces contains a caries lesion, the DIAGNOdent® with sensitivity 0.92 will correctly identify 46 of the 50 lesions present. Undertreatment is limited to only four lesions. With a specificity of 0.86, 817 non-carious surfaces will be correctly diagnosed as being sound, but still a very large number of 133 non-carious surfaces will be erroneously diagnosed as carious. This example illustrates that a caries diagnostic method should preferably have a high specificity when applied in low prevalence populations. Visual inspection has a high specificity and a low sensitivity. At the cost of some undertreatment, it will hardly cause unnecessary treatment. From this perspective, visual inspection remains the first choice to detect occlusal caries lesions when the prevalence is low. This finding is supported by the high overall validity of visual inspection, as exemplified by the high area under ROC for observer 1 (table 2). To interpret the validity of diagnostic tests in dentistry irrespective of a threshold, ROC analysis can be employed (14, 15). This analysis resulted in areas under ROC for visual inspection and DIAGNOdent® measurements which were not statistically significantly different. Being of equal performance, the DIAGNOdent® could be used as an aid to visual inspection, particularly when small lesions are suspected.
Slide 20 : 5. Conclusions It was concluded from this investigation that the reliability of DIAGNOdent® measurements is excellent, but that the validity of the DIAGNOdent® was not statistically significantly different from that of visual inspection. Due to the higher specificity, visual inspection should be preferred to detect occlusal caries lesions in populations with a low caries prevalence. The DIAGNOdent® seems more suitable for detecting small superficial lesions, rather than deep dentinal lesions. Because of the high reproducibility, dental practitioners who wish to use the DIAGNOdent® for monitoring caries lesions, may replace an old device by a new one.
Slide 21 : References [1] Truin GJ, van ‘t Hof MA, Kalsbeek H . Secular trends of caries prevalence in 6- and 12-year old Dutch children. Community Dentistry and Oral Epidemiology 1993; 21; 249-52.[2] Lussi A. Comparison of different methods for the diagnosis of fissure caries without cavitation. Caries Research 1993; 27; 409-16.[3] Sawle RF, Andlaw RJ. Has occlusal caries become more difficult to diagnose? British Dental Journal, 1988; 164; 209-11.[4] Wenzel A, Verdonschot EH, Truin GJ, König KG. Accuracy of visual inspection, fiber-optic transillumination, and various radiographic image modalities for the detection of occlusal caries in extracted non-cavitated teeth. Journal of Dental Research 1992; 71; 1934-37.[5] Weerheijm KL, Soet JJ de, Amerongen WE van, Graaff J de. Occlusale dentinelaesies onder een klinisch ogenschijnlijk intact glazuuroppervlak. Nederlands Tijdschrift voor Tandheelkunde 1992; 99; 213-16.[6] Ie YL, Verdonschot EH. The performance of diagnostic systems in occlusal caries detection compared. Community Dentistry and Oral Epidemiology 1994; 22; 187-91.[7] Verdonschot EH, Angmar-Mänsson B, ten Bosch JJ, et al. Developments in caries diagnosis and their relationship to treatment decisions and quality of care. Caries Research 1999; 33; 32-40.[8] Lussi A, Imwinkelried S, Pitts NB, et al. Performance and reproducibility of a laser fluorescence system for detection of occlusal caries in vitro. Caries Research. 1999; 33; 216-66.[9] Shi X-Q, Welander U, Angmar-M?nsson B. Occlusal caries detection with KaVo DIAGNOdent® and radiography: an in vitro comparison. Caries Research, 2000; 34; 151-58. [10] Pereira AC, Verdonschot EH, Huysmans MCDNJM. Caries detection methods: can they aid decision making for invasive sealant treatment? Caries Res 2001; 35: 83-89.[11] Attrill DC, Ashley PF. Occlusal caries detection in primary teeth: a comparison of DIAGNOdent® with conventional methods. Br Dent J 2001; 190: 440-443. [12] Shi XQ, Tranaeus S, Angmar-Månsson B. Validation of DIAGNOdent® for quantification of smooth-surface caries: an in vitro study. Acta Odontol Scand 2001b; 59: 74-78.[13] Lussi A, Megert B, Longbottom C, Reich E, Francescut P. Clinical performance of a laser fluorescence device for detection of occlusal caries lesions. Eur J Oral Sci 2001; 109: 14-19.[14] Verdonschot EH, Wenzel A, Bronkhorst EM. Assessment of diagnostic accuracy in caries detection: an analysis of two methods. Community Dentistry and Oral Epidemiology 1993a; 21; 203-8.[15] Verdonschot EH, Wenzel A, Bronkhorst EM. Applicability of Receiver Operating Characteristic (ROC) analysis on discrete caries depth ratings. Community Dentistry and Oral Epidemiology 1993b; 21; 269-72.[16] Metz CE, Kronman HB. Statistical significance test for binormal ROC curves. Journal of Mathematical Psychology 1980; 22; 218-43.

 


Related 

 
Free Powerpoint Templates
Add as Friend alwas@mnc.pl     4 Years ago.
1014 Views, 0 favourite
PowerPoint Presentation on Reliability and validity issues of laser fluorescence measurements in oc    more
More By User

Flag as inappropriate





Browse | Powerpoint Templates | Tags | Contact | About Us | Privacy | FAQ | Blog

© Slideworld