The aim of this study will be to use the texture analysis (TA) technique in cone beam computed tomography (CBCT) images as a tool to improve the differential diagnosis between odontogenic (OS) and non-odontogenic sinuses (NOS), in the maxillary sinuses. From the database of CBCT exams in the Radiology Clinic file, 40 CBCT exams (40 maxillary sinuses) will be selected by a previously calibrated evaluator, being 20 OS and 20 NOS, according to the criteria of visual analysis based on those proposed by Maillet et al. (2011). The exams will be exported to the OnDemand3D software (Cybermed, Seoul, South Korea), where 03 cross-sections will be selected, covering different regions of the image corresponding to the sinusitis, they will be individually exported and identified in bitmap format (.bmp). Each cross-section will be exported to the MaZda 4.60 software (Institute of Electronics, Technical Universityof Lodz, Poland) and then, with the "draw rectangle" tool, a region of interest (ROI), of fixed dimensions (5.0x3.0mm), will be determined for all images, positioned in the central region of sinus opacification. The texture parameters will then be calculated using the co-occurrence matrix (COM). The results will be tabulated and, for each TA parameter obtained, an average of the three cuts analyzed will be made. Spearman's correlation coefficient will be used to assess the correlation between texture variables and select variables for analysis. The groups (OS and NOS) will be compared using the Mann-Whitney test and the p-values will be adjusted using the FDR (False Discovery Rate) method. The ROC (Receiver Operating Characteristic) curve will be used to assess the sensitivity and specificity of the texture variables to differentiate the two sinus groups (OS and NOS). The cluster analysis will be used to form the two groups of sinusitis based on the texture data and assess whether these two groups formed are equivalent to the two study groups (OS and SOS). The level and significance adopted will be 5%.
News published in Agência FAPESP Newsletter about the scholarship: