Binary logistic regression analysis of solid thyroid nodules imaged by high-frequency ultrasonography, acoustic radiation force impulse, and contrast-enhanced ultrasonography

X.-N. Liang, R.-J. Guo, S. Li, Z.-M. Zheng, H.-D. Liang

Department of Ultrasound Medicine, Beijing Chaoyang Hospital, Capital Medical University, China. guodaoshi@126.com

---

• Endocrinology

OBJECTIVE: The common clinical techniques used for examining thyroid tumors include palpation, imaging, immunoassays and tissue biopsy. Ultrasonography is easy, non-invasive, non-radioactive and highly reproducible imaging technique; however, due to the disease polytropism, diagnosis may become difficult sometimes. Ultrasound elastography, particularly acoustic radiation force impulse (ARFI) imaging and contrast-enhanced ultrasonography (CEUS) have been successfully used to diagnose the thyroid tumors. The objective of this retrospective study was to analyze and compare the solid thyroid nodules imaged by high-frequency ultrasonography (HFUS), ARFI imaging, and CEUS.

PATIENTS AND METHODS: For this purpose, images of the 80 solid thyroid nodules (58 benign and 22 malignant) with surgical pathology were obtained and data were compared using binary logistic regression analysis.

RESULTS: Morphology (p < 0.001), and internal calcification (p = 0.007) were statistically different. The mean shear wave velocity (SWV) measured by ARFI was significantly different (p = 0.029). Three sets of comparison on CEUS (p = 0.019) and time to peak (TTP) of CEUS were significantly different (p = 0.001). The logistic regression analysis indicated that the morphology, mean SWV of ARFI and TTP were independent risk factors for malignancy. The diagnostic accuracy for solid thyroid nodules was 85.1% (68/80) and the area under the receiver operating characteristic (ROC) curve was 0.945±0.033.

CONCLUSIONS: Logistic regression analysis can effectively screen significant parameters for the differential diagnosis of solid thyroid nodules imaged by ultrasonography.

Free PDF Download