IADR Abstract Archives
Occlusal Caries Detection in Vivo and on 3D Digital Models
Objectives: To evaluate the diagnostic accuracy of visual caries detection performed on 3D digital models and compare it to in vivo visual examination.
Methods: Fifty-two molars and premolars scheduled for extraction were included in this study. One to three independent examination sites were defined on the occlusal surface of each tooth and were clinically inspected using ICDAS criteria. Afterwards, the examined teeth were scanned intraorally with TRIOS 4 intraoral scanner (3Shape TRIOS A/S, Denmark) using white light to obtain 3D color models, followed by illumination with light 415nm wavelength. The latter excites fluorescence by the dental tissues that were mapped on the 3D models. On the same visit, the teeth were extracted and stored until histological assessment could be conducted by an independent examiner. The examination sites were annotated on the 3D models and after six months, the principal examiner visually inspected and scored all examination sites on the high-resolution 3D models visualized with specific software (3Shape A/S, Denmark) using ICDAS criteria. Caries scoring was done first on the color models and thereafter on the models with the fluorescence texture.
Results: At the examined histological levels, all methods showed comparable diagnostic performance, with no significant differences between the areas under the ROC curves (Az>0.65, p>0.05). All methods showed similar diagnostic accuracy ranging from sufficient to good at more initial caries lesion stages (ACC > 59), and from very good to excellent at more extensive caries stages (ACC > 77). The intra-examiner reliability was perfect for all methods (kappa > 0.80).
Conclusions: Visual inspection of 3D digital color models or models with a fluorescence overlay can help detect occlusal caries lesions with accuracy equivalent to direct clinical visual inspection. Caries detection on 3D models has the potential to be utilized for teledentistry purposes.
Methods: Fifty-two molars and premolars scheduled for extraction were included in this study. One to three independent examination sites were defined on the occlusal surface of each tooth and were clinically inspected using ICDAS criteria. Afterwards, the examined teeth were scanned intraorally with TRIOS 4 intraoral scanner (3Shape TRIOS A/S, Denmark) using white light to obtain 3D color models, followed by illumination with light 415nm wavelength. The latter excites fluorescence by the dental tissues that were mapped on the 3D models. On the same visit, the teeth were extracted and stored until histological assessment could be conducted by an independent examiner. The examination sites were annotated on the 3D models and after six months, the principal examiner visually inspected and scored all examination sites on the high-resolution 3D models visualized with specific software (3Shape A/S, Denmark) using ICDAS criteria. Caries scoring was done first on the color models and thereafter on the models with the fluorescence texture.
Results: At the examined histological levels, all methods showed comparable diagnostic performance, with no significant differences between the areas under the ROC curves (Az>0.65, p>0.05). All methods showed similar diagnostic accuracy ranging from sufficient to good at more initial caries lesion stages (ACC > 59), and from very good to excellent at more extensive caries stages (ACC > 77). The intra-examiner reliability was perfect for all methods (kappa > 0.80).
Conclusions: Visual inspection of 3D digital color models or models with a fluorescence overlay can help detect occlusal caries lesions with accuracy equivalent to direct clinical visual inspection. Caries detection on 3D models has the potential to be utilized for teledentistry purposes.