Análise de textura em imagens de tomografia computadorizada de feixe cônico de regiões da mandíbula afetadas por displasia

Abstract

Background: Bone dysplasia is a type of lesion of the human jaw, as determined by visual inspection of computed tomography images. The limited ability of the human eye to distinguish gray levels, coupled with visual fatigue and tiredness, can result in misdiagnosis and, consequently, inappropriate treatment. Objective: The objective of this study was to quantify visual characteristics in computed tomography images of lesions caused by dysplasia. Materials and Methods: Seven cone-beam computed tomography images of patients diagnosed with dysplasia were retrospectively screened. Regions of interest (ROIs) were selected by a radiologist and subsequently processed using conventional and adaptive contrast enhancement techniques. First-order (mean, standard deviation, skewness, kurtosis, entropy) and second-order (contrast, brightness, homogeneity, energy) texture descriptors were applied to the processed ROIs to quantify radiographic aspects of dysplasia lesions. Results: A total of nine texture descriptors were calculated for each region of interest. Comparisons between texture feature values revealed numerical differences between the healthy and dysplasia regions. The mean, asymmetry, and entropy features differed significantly between the healthy and dysplastic regions. Contrast, homogeneity, and energy also exhibited observable differences for the two regions. Conclusions: This study focused on quantifying radiographic changes in computed tomography images of lesions caused by dysplasia in the human jaw bone. Direct comparisons revealed numerical differences for most texture features between the healthy and dysplasia regions. Although the results are numerically distinct, their significance still needs to be investigated in future stages through statistical tests applied to a larger number of images. Novelty: The study's novelty lies in the quantification of radiological alterations in areas of the mandible affected by dysplasia. Clinical Relevance: Unlike qualitative analysis through visual inspection, which is costly and subjective, the use of a more systematic and non-subjective method, such as the one proposed, has the potential to facilitate diagnoses in routine dental clinical practice. Knowledge Advances: There is a gap in the literature regarding the use of texture analysis to characterize subtypes of lesions caused by dysplasia. In this preliminary study, one type of lesion was characterized, and the procedure developed can be reproduced to characterize other lesion subtypes.