Einfluss der digitalen Bildbearbeitung auf das Beurteilungsergebnis des International Caries Detection and Assessment System (ICDAS) in der studentischen Ausbildung

Influence of digital image processing on the evaluation results of the ICDAS method of visual caries detection by students Objectives: Main object in dental education is to transfer deeper knowlege for detection of carious lesions. The most important tool in daily use is the visual inspection of teeth. The International Caries Detection and Assessment System (ICDAS) has been proven to be a valid and multi-purposed visual diagnostic-protocol. A standardization of diagnostic process has been given and it is useful to point out conformable therapy concepts for different stages of disease. Digital-visual examinations of teeth-images seem to be a suitable didactical method to teach ICDAS to students. Aim: This study aims to evaluate the use of ICDAS applied by undergraduate students when digitally optimized macrophotography images of extracted teeth are used. Two different methods have been used to produce significant images. Another aim was to prove influence of humidity on teeth-surface to results of digital-visual image examinations. Test subjects were undergraduate third- and fifth-year-students respectively. It has been analyzed if a different stage of clinical experience had effects on reproducibility and diagnostical quality of the present examination. Material and method: 60 extracted human teeth were used for this study (5 incisors, 1 canine, 8 premolars, 46 molars). Prior to extraction informed consent had been obtained from the patients about the use of the teeth for research purposes. Teeth with various stages of carious lesions have been used (ICDAS codes 0-6). Teeth had been cleaned and checked thoroughly by two approbated examiners for reference (A: longstanding expert knowledge; B: postgraduate assistant). An ICDAS-reference-score was determined for every tooth (consent diagnosis). Photocopy of teeth and image-generating followed a standardized setting. Pictures had been taken under dried and humidified surface conditions: Foto konv (single shot, focussing on spot MP, aperture f/22); Foto comp (digitally stitched image from 10-15 single shots, focussing on different levels of image definition between spot HP and TP, aperture f/5,6). The digital stitch of the single shots was made with Helicon Focus Version 5.3.3 (image editing program). Images were digitally delivered to subject groups A (third-year-students, no clinical experience of using ICDAS) and B (fifth-year-students, moderate clinical experience of using ICDAS). Group A and B had to assess the images twice in one week. During the second cycle of evaluation 2/3 of images were reassessed (randomized). Statistical analysis was performed using MedCalc 12.4.0. Examiner reproducibility was calculated using weighted kappa-values. Sensitivity and specifity of examinations were determined on D1- (enamel and dentine-caries) and D3-level of caries detection (dentine-caries). Diagnostic performance (area under the ROC-curve) was calculated using ICDAS consensus scores of reference-examiners A/B (ICDAS-reference-scores). Results: Intraexaminer-reproducibility for group A and B was between κw=0.531-0.898 (moderate to nearly perfect reproducibility). Inter-examiner-reproducibility was between κw=0.4320.730 (moderate to substantial reproducibility). Differences between groups were statistically non-significant (t-test; p=0.231 respectively p=0.969). At D1-level of caries detection sensitivity was between 40-100%, specifity was 30-100%. At D3-level sensitivity was 3-91%, specifity was 67-100%. Diagnostic perfomance of both groups was high (AUC group A: 0.809-0.933; group B: 0.801-0.961). At D1-level of caries detection students of group B had significantly better AUCs when digitally stitched images (Foto comp), images with dried tooth-surface (Foto trocken) and parametercombination (Foto comp/trocken) were used (non-parametrical test; p=<0.001-0.004). At D3-level of caries detection differences in diagnostic performance between both groups were statistically non-significant. Conclusion: Basically tooth-images are suitable to introduce students to visual caries inspection. Third- and fifth-year students alike were able to use ICDAS caries-score to evaluate tooth-images digital-visually. Clinically experienced students notably benefit from using digitally stitched images and dried surface to detect enamel carious lesions. The additional use of high-quality images can be recommended especially for senior dental students in advanced training and/or further education respectively.