Introduction: On the basis of the self-similarity of the human body, fractal dimensions can be calculated mathematically. Fractal analysis can also quantitatively express the density and complexity of a structure as a fractal dimension. We hypothesize that the fractal dimensions of ultrasonographic findings would change in thyroid disease. In this study, we explore the possibility of applying fractal analysis to the quantitative evaluation of ultrasonographic findings in thyroid tissue. Methods: This retrospective study used data on 1205 patients who newly received thyroid ultrasonography between April 2004 and March 2016. Patients who were included had one or more of the following: normal thyroid, Graves' disease, Hashimoto's thyroiditis, subacute thyroiditis, adenomatous goiter, or papillary carcinoma. Furthermore, the fractal dimension of thyroid ultrasonography was calculated in sagittal pictures. The region of interest was set along the contour of the thyroid gland, excluding a film. Fractal analysis was performed using box-counting methods. Results: For the normal thyroid gland, there was a significant decrease in the fractal dimensions of both contour and concentration structures, with significant differences compared with the disease groups. Among the thyroid disease groups, papillary carcinoma had the highest fractal dimension. The fractal dimension of the thyroid gland increased with age in the normal group. There were no differences in fractal dimensions by gender. Conclusion: Fractal analysis makes it possible to objectively determine the presence of chronic inflammation in thyroid tissue, but it is difficult to distinguish papillary carcinoma from benign goiter.
Komatsu et al. (Sun,) studied this question.