Quantitative Ul Trasound-Based Detection of Cancerous Thyroid Nodules

摘要

Thyroid cancer is the most-common endocrine malignancy in the world. The objective of this study was to investigate the suitability of quantitative ultrasound (QUS) for detecting thyroid cancer. Radio-frequency (RF) data were acquired from 53 patients using a GE Logiq E9 research ultrasound system equipped with a lO-MHz, ML6-l5 linear probe. QVS-estimate values, such as effective acoustic concentration (EAC), effective scatter diameter (ESD), and spectral parameters (i.e., midband fit (Mf), slope (SS), and intercept (I0)) were derived using a reference-phantom method. Additional QUS estimates (i.e., μ and Ω) were derived from the envelope statistics of a Nakagami distribution. Estimate values were used to train a linear classifier, and performance was assessed using area-under-the-curve (AUC) values obtained from receiver operating characteristic (ROC) analyses based on lO-fold cross validation. Eight patients had pathology-confirmed thyroid cancer. A combination of EAC, I0, and μ produced an AUC value of 0.96. TI-RADS used alone by clinical thyroid sonographic experts produced an AUC of 0.94. A combination of TI-RADS, EAC, andμ produced an AUC value of 0.97. By achieving AUC values of 0.97, our initial results are very encouraging for developing a new objective and non-invasive tool for thyroid-cancer detection. These results suggest that QUS has promising potential to improve non-surgical evaluation of thyroid nodules.