Análise fractal em imagens odontológicas: estudos comparativos entre modalidades de imagem e parâmetros de aquisição

Abstract

Studies on fractal analysis have demonstrated its potential applicability across different areas of Dentistry. Although the structured protocol for its use in dental imaging was proposed more than twenty years ago, important gaps remain regarding the influence of imaging modality, acquisition parameters, image-processing methods, and the selection of regions of interest. In this context, the present dissertation had two objectives. The first was to compare different imaging modalities (panoramic radiography and panoramic reconstructions obtained from cone beam computed tomography with varying thicknesses) in the assessment of trabecular bone structure using fractal analysis. The second objective was to investigate the influence of milliampere settings, image-processing methods for fractal analysis, and regions of interest on fractal dimension values obtained from cone-beam computed tomography images, as well as to evaluate whether variations in these factors affect fractal dimension values. The first study evaluated images from 20 patients, including panoramic radiographs and panoramic reconstructions with different thicknesses (160 µm; 480 µm; 800 µm; 1.1 mm; 2.1 mm; 3 mm; 4 mm; 5 mm; 10 mm; 14.9 mm; 20 mm; 29.9 mm; 40.2 mm; and 50.1 mm). In each image, bilateral regions of interest were selected 2 mm superior, medial, and inferior to the mental foramen, and fractal dimension values were obtained using the White and Rudolph method. The results demonstrated differences between fractal dimension values obtained from panoramic radiographs and panoramic reconstructions, as well as between thinner reconstructions (160 µm and 480 µm) and thicker reconstructions (29.9 mm, 40.2 mm, and 50.1 mm). In the second study, eight edentulous human mandibles were evaluated using cone beam computed tomography with seven different milliampere settings (2, 4, 6.3, 8, 10, 12, and 15). Cross-sectional slices were selected in five regions of interest bilaterally (incisor, canine, premolar, first molar, and second molar), and fractal analysis was performed using three different image-processing methods (White and Rudolph, revised White and Rudolph, and the gray-scale method). The results indicated that fractal dimension values in tomographic images are significantly influenced by the image processing method used for fractal analysis but remain stable across variations in milliampere settings and regions of interest within the mandibular body. Overall, the findings of this dissertation demonstrate that fractal dimension is sensitive to specific methodological factors, reinforcing the need for standardization of the imaging modality and processing protocol for fractal analysis in longitudinal follow-up and when comparing patient data.