Nowadays, molecular analysis is usually applied to re-stratify the risk of malignancy of thyroid nodules with indeterminate cytology readings. Its high costs impair the widespread application: currently, the molecular testing for thyroid nodules is not reimbursed by the Italian health system.
Preliminary data shows that there is an association between some sonographic features and BRAF or TERT promoter mutations.
The main aim of the study is to identify imaging features or pattern predictive of specific molecular alterations. This may be useful to: a) avoid expensive pre-operative molecular testing in nodules whose imaging features reliably predict molecular alterations and diagnosis; b) derive prognostic data from both cytology and "radiomics biopsy" and compare their prognostic performance. Finally, the results may allow to limit the application of actual molecular testing to the cases in which it has better discriminance, improving its overall cost-effectiveness.
Images will be prospectively collected from patients who undergo fine-needle aspiration biopsy with a cytological diagnosis. Each image will be de-identified, anonymized and labeled as "benign", "low-risk indeterminate", "high-risk indeterminate", or "malignant". The following data will be collected: traditional interpretation according to sonographic risk stratification system, quantitative feature extraction using radiomics and computer vision approaches. Part of cytological sample will be used to extract nucleic acids to be submitted to targeted Next-Generation Sequencing analysis. Each single human-derived sonographic feature, sonographic classification, and radiomics-derived parameter will be then evaluated to estimate its association with specific molecular alterations.
The data will demonstrate if "radiomics biopsy" would be able to reduce the number of performed cytologies, molecular testing, and thyroid surgeries, improving cost-effectiveness of the whole diagnostic algorithm.
Several approaches have already been proposed to improve the diagnostic accuracy and the reproducibility of US evaluation of thyroid nodules, such as systematic reporting schemas (Su, et al. 2014) and quantitative evaluation of echogenicity (Grani, et al. 2015). According to some evidence, thyroid computer-aided diagnosis (CAD) using artificial intelligence may further improve diagnosis reliability. It was reported that the use of thyroid CAD to differentiate malignant from benign nodules showed accuracy similar to that obtained by radiologists (Chang, et al. 2016; Choi, et al. 2017) and may reduce intra- and inter-observer variability.
In this context, we will check if these techniques [classification according to sonographic stratification systems; quantitative feature extraction, so called "radiomics", with the application of "computer vision" and machine learning approaches] are able to further stratify the malignancy risk of thyroid nodules, predicting the presence of various molecular alterations. In this case, "radiomics biopsy" will be able to reduce the number of performed cytologies, molecular testing, and thyroid surgeries, improving cost-effectiveness of the whole diagnostic algorithm.
References
Chang, Y., Paul, A. K., Kim, N., Baek, J. H., Choi, Y. J., Ha, E. J., Lee, K. D., Lee, H. S., Shin, D., & Kim, N. (2016). Computer-aided diagnosis for classifying benign versus malignant thyroid nodules based on ultrasound images: A comparison with radiologist-based assessments. Medical physics, 43(1), 554. doi:10.1118/1.4939060
Choi, Y. J., Baek, J. H., Park, H. S., Shim, W. H., Kim, T. Y., Shong, Y. K., & Lee, J. H. (2017). A Computer-Aided Diagnosis System Using Artificial Intelligence for the Diagnosis and Characterization of Thyroid Nodules on Ultrasound: Initial Clinical Assessment. Thyroid, 27(4), 546-552. doi: 10.1089/thy.2016.0372.
Grani, G., D'Alessandri, M., Carbotta, G., Nesca, A., Del Sordo, M., Alessandrini, S., Coccaro, C., Rendina, R., Bianchini, M., Prinzi, N., & Fumarola, A. (2015). Grey-Scale Analysis Improves the Ultrasonographic Evaluation of Thyroid Nodules. Medicine, 94(27), e1129. doi:10.1097/MD.0000000000001129
Su, H. K., Dos Reis, L. L., Lupo, M. A., Milas, M., Orloff, L. A., Langer, J. E., Brett, E. M., Kazam, E., Lee, S. L., Minkowitz, G., Alpert, E. H., Dewey, E. H., & Urken, M. L. (2014). Striving toward standardization of reporting of ultrasound features of thyroid nodules and lymph nodes: a multidisciplinary consensus statement. Thyroid, 24(9), 1341-1349. doi:10.1089/thy.2014.0110