Fibrous Dysplasia Characterization Using Lacunarity Analysis

Fibrous dysplasia (FD) is a developmental anomaly in which the normal medullary space of the affected bone is replaced by fibro-osseous tissue. This condition is typically encountered in adolescents and young adults. It affects the maxillofacial region and it can often cause severe deformity and asymmetry. Therefore, accurate diagnosis is critical to determine the appropriate treatment of each case. In this sense, computed tomography (CT) is a relevant resource among the imaging techniques for correct diagnosis of this condition. Thus, in this paper, we propose to analyze fibrous dysplasia through its texture pattern. To accomplish this task, we propose to use lacunarity analysis, a multiscale method for describing patterns of spatial dispersion. Results indicated lower lacunarity values for fibrous dysplasia in comparison to normal bone samples, an indication that their texture images are more homogeneous, and a high separability between the classes when using principal component analysis (PCA) and decision trees for statistical analysis.