Estimation of the Fractal Dimension of Thyroid Nodules as a Complementary Diagnostic Tool for Predicting Malignancy

Author(s): Lothaire P, Uzureau P, Boudjelta Kz, Nagy N, Michiels C, Carletti T

Rationale and objectives: Ultrasound-guided fine-needle aspiration is the most sensitive investigation procedure in the evaluation of patients with thyroid nodules; however, despite the level of achieved precision, it is still impossible to preoperatively discriminate between follicular adenomas and carcinomas. Thus, no current detection tool of thyroid nodule has both high sensitivity and high positive predictive value for the diagnosis of thyroid cancer. Ultrasound images provide information that are currently missed due to visual interpretation, including the roughness of the echotexture. However, such information can be retrieved using the measurement of the fractal dimension (FD).

Materials and methods: This retrospective study was approved by the ethic committee of CHU Charleroi. According to the boxcounting method, we measured the FD of the echotexture of ultrasound images of 13 benign thyroid nodules and 14 malignant thyroid nodules compared to normal thyroid tissue from 10 men and 16 women (mean age 40).

Results: The difference of the FD value between malignant nodules (median FD 2.31) and corresponding normal tissues (median FD 2.184) was significant (p<0.001) while the FD values for the benign nodules (median FD 2.21) and for corresponding normal tissues (median FD 2.231) could not be distinguished (p=0.78).

Conclusion: This preliminary work revealed that FD would be an additional feature to discriminate benign from malignant nodules. A larger sampling is now required to confirm the possible utilization of FD in the thyroid cancer diagnosis in clinical practice.