According to an article published ahead-of-print in the April issue of the American Journal of Roentgenology (AJR), a Stanford University team has developed a quantitative framework able to sonographically differentiate between benign and malignant thyroid nodules at a level comparable to that of expert radiologists, which may prove useful for establishing a fully automated system of thyroid nodule triage.
Alfiia Galimzianova et al. retrospectively collected ultrasound images of 92 biopsy-confirmed nodules, which were annotated by two expert radiologists using the American College of Radiology's Thyroid Imaging Reporting and Data System (TI-RADS).
In the researchers' framework, nodule features of echogenicity, texture, edge sharpness, and margin curvature properties were analyzed in a regularized logistic regression model to predict nodule malignancy. Authenticating their method with leave-one-out cross-validation, the Stanford team used ROC AUC, sensitivity, and specificity to compare the framework's results with those obtained by six expert annotation-based classifiers.
The AUC of the proposed framework measured 0.828 (95% CI, 0.715-0.942)--"greater than or comparable," Galimzianova noted, "to that of the expert classifiers"--whose AUC values ranged from 0.299 to 0.829 (p = 0.99).
Additionally, in a curative strategy at sensitivity of 1, use of the framework could have avoided biopsy in 20 of 46 benign nodules--statistically significantly higher than three expert classifiers. In a conservative strategy at specificity of 1, the framework could have helped to identify 10 of 46 malignancies--statistically significantly higher than five expert classifiers.
"Our results confirm the ultimate feasibility of computer-aided diagnostic systems for thyroid cancer risk estimation," concluded Galimzianova. "Such systems could provide second-opinion malignancy risk estimation to clinicians and ultimately help decrease the number of unnecessary biopsies and surgical procedures."
Founded in 1900, the American Roentgen Ray Society (ARRS) is the first and oldest radiology society in North America, dedicated to the advancement of medicine through the profession of radiology and its allied sciences. An international forum for progress since the discovery of the x-ray, ARRS maintains its mission of improving health through a community committed to advancing knowledge and skills with an annual scientific meeting, monthly publication of the peer-reviewed American Journal of Roentgenology (AJR), quarterly issues of InPractice magazine, AJR Live Webinars and Podcasts, topical symposia, print and online educational materials, as well as awarding scholarships via The Roentgen Fund®.
American Journal of Roentgenology