- Research note
- Open Access
Diagnostic accuracy of ultrasound characteristics in the identification of malignant thyroid nodules
BMC Research Notes volume 12, Article number: 193 (2019)
This study was aimed at determining the diagnostic accuracy of the ultrasound characteristics in the identification of malignant thyroid nodules in patients attending the surgical clinic at a tertiary care hospital in Sri Lanka.
This prospective validation study was conducted among 263 patients attending the surgical clinic at a tertiary care hospital, Sri Lanka. The ultrasound characteristics having statistically significant associations with thyroid malignancies were identified by employing a multivariable analysis. These ultrasound characteristics were compared with Fine Needle Aspiration Cytology results and indicators of diagnostic accuracy were computed. The study sample consisted of 33 malignant and 230 benign nodules. Internal vascularity, hypoechogenicity and microcalcification showed statistically significant positive associations with thyroid malignancy (p < 0.05). The highest positive likelihood ratio was observed for the presence of microcalcifications (10.4, 95% CI 4.6–23.7) with a specificity of 96.5% (95% CI 93.3–98.5%). Hypoechogenicity had a sensitivity of 66.7% (95% CI 48.2–82.0%) and a specificity of 74.8% (95% CI 68.6–80.3%). The presence of at least one positive ultrasound characteristic yielded the highest sensitivity (87.9%, 95% CI 71.8–96.6%), whereas, the presence of all three characteristics yielded a near perfect specificity (99.6%, 95% CI 97.6–99.9%). Hence, the presence of microcalcifications was found to be the single most useful criterion in predicting thyroid malignancy.
Globally, thyroid malignancy is the most common endocrine cancer and there has been a steady increase in the global incidence of thyroid cancer over the last few decades [1, 2]. Since there is a 5–15% probability of malignancy in any given thyroid nodule [3, 4], current thyroid guideline recommendations call for imaging modalities to detection of thyroid malignancies early leading to a better prognosis . Ultrasound scanning is widely cited as the imaging modality of choice for early diagnosis owing to its wide availability, relative simplicity and non-exposure to ionizing radiation [6,7,8,9,10].
Despite the highlighted advantages in ultrasound imaging, the evidence is not conclusive in predicting thyroid malignancies [11, 12]. Hence, the diagnostic accuracy of the ultrasound scanning needs to be evaluated by comparing with the results of a gold standard test. Fine needle aspiration cytology (FNAC) is considered as the most reliable cost effective method for definitive evaluation of thyroid nodules [13,14,15,16,17,18,19,20]. Thus, thyroid cancer management guidelines recommend conducting FNAC on any thyroid nodule, which is suspected to be malignant [5, 8,9,10]. Global evidence highlights a multitude of ultrasound characteristics that are indicative of thyroid malignancies [3, 7,8,9, 13, 14, 16, 19, 21,22,23,24,25,26,27,28,29,30]. However, collective global evidence base suggests inconsistent results across studies on predictive abilities.
Thyroid nodules are common clinical presentations to the surgical clinics in Sri Lanka ; hence, accurate identification of potential malignancy will have a great impact on patient management. In this background, it is important to utilise the ultrasound imaging techniques optimally, especially to decide on cases necessitating urgent surgical interventions. Despite the significance, only a limited number of studies have been conducted in Sri Lanka to evaluate the diagnostic accuracy of the thyroid ultrasound imaging . In this research vacuum, the present study was conducted to determine the diagnostic accuracy of the ultrasound characteristics in identification of malignant thyroid nodules.
The findings are reported following the Standards for Reporting of Diagnostic Accuracy Studies (STARDS 2015) guidelines .
Study design and setting
This prospective validation study was carried out at the Radiology Department in one of the largest tertiary care centres in the Central Province, Sri Lanka for a period of 18 months from January 2014 to June 2015.
In this single-gate study, all patients referred to the Radiology Department during the study period for thyroid ultrasound scanning were considered as the study population, excluding patients with known bleeding diathesis. All patients who fulfilled the eligibility criteria were recruited for the study upon the informed written consent.
Ultrasound scan of the thyroid
Ultrasound scanning was done by an experienced Consultant Radiologist, who has over 15 years experience in thyroid ultrasound scanning. Following criteria for the determination of ultrasound characteristics suggestive of the thyroid malignancies were developed based on the findings of an extensive literature review. The Thyroid Imaging Reporting and Data System (TI-RADS) Committee of the American College of Radiology has presented a risk-stratification system for classifying thyroid nodules on the basis of their appearance at ultrasonography  and the five ultrasound characteristics included in this classification were also included in this study.
All 263 participants have undergone the ultrasound scanning of the thyroid nodules and the above-mentioned ultrasound characteristics were recorded in each participant.
Ultrasound guided FNAC of thyroid nodules
Following the ultrasound scanning, participants were subjected to ultrasound guided FNAC, upon their informed written consent. The FNAC of the thyroid nodules were carried out by the same Consultant Radiologist, who performed the ultrasound scanning, using the same ultrasound scanner used to detect the nodule. The procedure was done under strict aseptic conditions using a 23G needle and three samples were taken from each nodule. Prepared slides were immersed immediately in 95% ethyl alcohol solution, before transporting them to the laboratory.
Cytology of thyroid nodules
The thyroid cytopathology report was prepared according to the Bethesda classification . An experienced Consultant Pathologist (with more than 25 years experience in histopathology), who was blinded to the ultrasound scan findings, evaluated all the samples.
Data analysis was carried out by using Statistical Package for the Social Sciences (SPSS) analytic software version 25.0. Based on FNAC findings, the sample was categorised as benign and malignant.
To identify the ultrasound characteristics with significant associations with thyroid malignancy, a bivariate analysis was conducted and the crude odds ratios (OR) were computed with 95% confidence intervals (95% CI). A binomial multiple logistic regression model was used to control for confounding. Ultrasound characteristics that showed statistical significance (p < 0.05) in the bivariate analysis were included in the multivariable analysis model using backward stepwise elimination method and the model produced adjusted odds ratios (AOR) and 95% CI.
The statistically significant ultrasound characteristics were compared with FNAC results separately and in combination. Performance characteristics of each ultrasound characteristic and combined ultrasound characteristics, namely, sensitivity, specificity, positive and negative predictive values and the positive likelihood ratios (LR+) were computed. Point estimates for the performance characteristics were supported by the 95% CI estimates.
Characteristics of the sample
The study sample consisted of 263 participants and all participants have undergone both ultrasound scanning and ultrasound guided FNAC. The majority of participants were females (n = 247, 93.9%). The age range of the sample was 16 years to 74 years (mean = 46.1 years, SD = 12.7 years).
Diagnosis of the ultrasound guided FNAC
Based on the cytological diagnosis of the thyroid nodules, the sample was categorised as benign and malignant. According to the FNAC results, Thy2 lesions were categorised as benign (n = 163). In addition, the benign category consisted of the Thy3 and Thy4 lesions (n = 79), which underwent surgery (n = 58) and histopathologically confirmed as benign (n = 46). Further, Thy3 lesions, which showed no interval changes after 6 months, were also included in this category (n = 21). The malignant group consisted of all the histological proven malignancies in Thy3 and Thy4 group (n = 12) and those that were Thy5 (n = 10) and Thy6 (n = 11) from the FNAC results. Thus, there were 33 proven malignant thyroid nodules and 230 benign nodules.
Association between the ultrasound characteristics and thyroid malignancy
Table 1 summarises the results of the bivariate analysis of the association between ultrasound characteristics and thyroid malignancy and out of the eight ultrasound characteristics assessed, except for the size and consistency, other characteristics showed statistically significant associations with thyroid malignancy at p < 0.05.
Table 2 illustrates the results of the analysis of binary logistic regression of the ultrasound characteristics retained in the final model. Out of the five factors retained in the final model, three factors made unique statistically significant contributions at p < 0.05. When controlled for other factors included in the logistic regression analysis, internal vascularity, hypoechogenecity and microcalcification showed statistically significant positive associations with thyroid malignancy.
Diagnostic accuracy of ultrasound characteristics in the identification of thyroid malignancies
The performance characteristics of individual ultrasound characteristic evaluated against the final diagnosis are summarised in the Table 3.
Amongst the single ultrasound characteristics assessed, the highest LR+ was observed for the presence of microcalcifications and it had a very high specificity, but the sensitivity was relatively low. On the contrary, hypoechogenicity had a relatively higher sensitivity and specificity values.
Amongst the combined ultrasound characteristics, the presence of at least one positive characteristic yielded the highest sensitivity, whereas, the sensitivity values were substantially lower when more than one characteristic was combined. On the contrary, presence of all three characteristics yielded a near perfect specificity.
Since the existing literature on diagnostic accuracy of ultrasound characteristics in predicting thyroid malignancies fails to provide conclusive evidence, we considered a number of important ultrasound characteristics cited in the literature for the present study.
Since experienced consultants in the relevant fields interpreted the findings of the index test and the reference standard separately based on clear, explicit guidelines, the risk of misclassification bias is minimised in the study. Furthermore, given that a single consultant carried out interpretation of ultrasound imaging and cytology separately, there was no inter-observer bias in gathering information. Since the Consultant Pathologist was blinded to the results of the ultrasound scan findings of the participants, there was no risk of bias in interpreting the reference standard test.
In the present study, ultrasound scanning was considered as the index test and FNAC diagnosis, which is cited as the gold standard diagnosis in thyroid malignancies [14, 15, 17,18,19,20] was used as the reference standard.
Rather than evaluating all ultrasound characteristics for their diagnostic accuracy, statistically significant characteristics were identified by employing both bivariate analysis followed by the multivariable analysis. This analysis was important in addressing the potential confounding effect of ultrasound characteristics on each other and has aided in determining the most significant characteristics to be used in diagnostic accuracy evaluation. The analysis has yielded three significant ultrasound characteristics, namely, microcalcifications, hypoechogenicity and internal vascularity. These findings are consistent with most of the existing literature, as many studies have revealed satisfactory indicators of diagnostic accuracy of microcalcification [7,8,9, 14, 16, 19, 21, 29], hypoechogenicity [8, 9, 14, 16, 19, 21, 24] and internal vascularity [7, 13, 16, 19, 28,29,30] in predicting thyroid malignancy.
The presence of microcalcifications is associated with a LR+ of 10.4 (95% CI 4.6–23.7), indicating that the likelihood of having thyroid malignancy has increased by approximately ten-fold given the positive ultrasound characteristic. However, this characteristic was found to have a low sensitivity of 36.4%, which is consistent with low sensitivity (26.1–59.1%) values reported in the literature [3, 14, 16, 21, 29]. In contrast, the specificity of this ultrasound characteristic was very high (96.5%), which is consistent with previously reported values of 90.8%  and 87.8% . The strong association of microcalcifications in thyroid malignancies is thought to be due to the fact that microcalcifications correspond pathologically to calcified ‘psammoma bodies’ that are typical of papillary cancer [21, 36, 37]. In contrast, macrocalcifications or coarse calcifications are related to fibrosis and degeneration [38, 39] and benign nodules of long duration are mostly associated with coarse calcifications [40, 41].
In the present study, hypoechogenicity was shown to have a sensitivity of 66.7% and a specificity of 74.8%. These findings are congruent with existing literature, in which the sensitivity was reported as 62.7% in a meta-analysis . However, literature suggests that the sensitivity values range from 26.5%  to 87.2% . In a similar vein, even though the meta-analysis findings reveal a specificity of 62.3% , the values range from 58.5%  to as high as 94.3% . Furthermore, the LR+ reported in the present study is similar to that in the meta-analysis .
The study findings reveals that internal vascularity had a sensitivity of 48.5%, specificity of 79.1% and a LR+ of 2.3, corresponding with the meta-analysis findings of 45.9%, 78.0% and 2.1 of sensitivity, specificity and LR+ respectively . On the contrary, some studies have concluded that the intra-nodular colour flow alone has a lower predictive value in assessing the malignant thyroid nodules [13, 42].
In relation to the combined ultrasound characteristics, the presence of at least one positive characteristic yielded the highest sensitivity, whereas, the presence of all three characteristics yielded a near perfect specificity. Furthermore, when considering multiple ultrasound characteristics in combination, the LR+ of detecting malignant thyroid nodules increased. These findings are congruent with the findings of the existing literature [16, 32].
Our study showed that the presence of microcalcifications and hypoechogenicity are the most significant criteria in predicting thyroid malignancies, which are also included in the TI-RADS classification. In addition to the characteristics included in the TI-RADS classification, internal vascularity has shown a statistically significant association in our study. Thus, further studies are needed to explore the role of vascularity in predicting thyroid malignancies.
In this study, all the subjects with benign FNAC results did not attend the follow-up ultrasound scan in 6 months. Of Thy3 and Thy4 lesions, 58 subjects underwent surgical removal and the final diagnosis was obtained from the histopathological report. Twenty-one subjects with Thy3 lesion did not undergo surgery, but attended follow-up ultrasound scan.
adjusted odds ratio
fine needle aspiration cytology
positive likelihood ratio
thyroid imaging reporting and data system
- 95% CI:
95% confidence interval
Pellegriti G, Frasca F, Regalbuto C, Squatrito S, Vigneri R. Worldwide increasing incidence of thyroid cancer: update on epidemiology and risk factors. J Cancer Epidemiol. 2013;2013:965212. https://doi.org/10.1155/2013/965212.
La Vecchia C, Malvezzi M, Bosetti C, Garavello W, Bertuccio P, Levi F, et al. Thyroid cancer mortality and incidence: a global overview. Int J Cancer. 2015;136:2187–95. https://doi.org/10.1002/ijc.29251.
Brito JP, Gionfriddo MR, Al Nofal A, Boehmer KR, Leppin AL, Reading C, et al. The accuracy of thyroid nodule ultrasound to predict thyroid cancer: systematic review and meta-analysis. J Clin Endocrinol Metab. 2014;99:1253–63. https://doi.org/10.1210/jc.2013-2928.
Burman KD, Wartofsky L. Thyroid nodules—NEJM. Nejm. 2015;373:2347–56. https://doi.org/10.1056/NEJMcp1415786.
Cooper DS, Doherty GM, Haugen BR, Hauger BR, Kloos RT, Lee SL, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. 2009;19:1167–214. https://doi.org/10.1089/thy.2009.0110.
Smith-Bindman R, Lebda P, Feldstein VA, Sellami D, Goldstein RB, Brasic N, et al. Risk of thyroid cancer based on thyroid ultrasound imaging characteristics: results of a population-based study. JAMA Intern Med. 2013;173:1788–96. https://doi.org/10.1001/jamainternmed.2013.9245.
Woliński K, Szkudlarek M, Szczepanek-Parulska E, Ruchała M. Usefulness of different ultrasound features of malignancy in predicting the type of thyroid lesions: a meta-analysis of prospective studies. Pol Arch Med Wewn. 2014;124:97–104. https://doi.org/10.20452/pamw.2132.
Alam T, Khattak YJ, Beg M, Raouf A, Azeemuddin M, Khan AA. Diagnostic accuracy of ultrasonography in differentiating benign and malignant thyroid nodules using fine needle aspiration cytology as the reference standard. Asian Pac J Cancer Prev. 2014;15(22):10039–43.
Nachiappan AC, Metwalli ZA, Hailey BS, Patel RA, Ostrowski ML, Wynne DM. The thyroid: review of imaging features and biopsy techniques with radiologic–pathologic correlation. Radiographics. 2014;34(2):276–93.
McQueen AS, Bhatia KSS. Thyroid nodule ultrasound: technical advances and future horizons. Insights Imaging. 2015;6:173–88. https://doi.org/10.1007/s13244-015-0398-9.
Peccin S, De Castro JAS, Furlanetto TW, Furtado APA, Brasil BA, Czepielewski MA. Ultrasonography: is it useful in the diagnosis of cancer in thyroid nodules? J Endocrinol Invest. 2002;25(1):39–43.
Baier ND, Hahn PF, Gervais DA, Samir A, Halpern EF, Mueller PR, et al. Fine-needle aspiration biopsy of thyroid nodules: experience in a cohort of 944 patients. Am J Roentgenol. 2009;193:1175–9. https://doi.org/10.2214/AJR.08.1840.
Rago T, Vitti P, Chiovato L, Mazzeo S, De Liperi A, Miccoli P, et al. Role of conventional ultrasonography and color flow-doppler sonography in predicting malignancy in “cold” thyroid nodules. Eur J Endocrinol. 1998;138:41–6. https://doi.org/10.1530/eje.0.1380041.
Kim EK, Cheong SP, Woung YC, Ki KO, Dong IK, Jong TL, et al. New sonographic criteria for recommending fine-needle aspiration biopsy of nonpalpable solid nodules of the thyroid. Am J Roentgenol. 2002;178:687–91. https://doi.org/10.2214/ajr.178.3.1780687.
Castro MR, Gharib H. Thyroid fine-needle aspiration biopsy: progress, practice, and pitfalls. Endocr Pract. 2003;9:128–36.
Frates MC, Benson CB, Charboneau JW, Cibas ES, Clark OH, Coleman BG, et al. Management of thyroid nodules detected at US: Society of Radiologists in Ultrasound consensus conference statement. Radiology. 2005;237(3):794–800.
Cooper DS, Doherty GM, Haugen BR, Kloos RT, Lee SL, Mandel SJ, et al. Management guidelines for patients with thyroid nodules and differentiated thyroid cancer: the American Thyroid Association Guidelines Taskforce. Thyroid. 2006;16(2):109–42.
Gupta M, Gupta S, Gupta VB. Correlation of Fine Needle Aspiration Cytology with Histopathology in the Diagnosis of Solitary Thyroid Nodule. J Thyroid Res. 2010;2010:1–5. https://doi.org/10.4061/2010/379051.
Remonti LR, Kramer CK, Leitão CB, Pinto LCF, Gross JL. Thyroid ultrasound features and risk of carcinoma: a systematic review and meta-analysis of observational studies. Thyroid. 2015;25:538–50. https://doi.org/10.1089/thy.2014.0353.
Zayadeen AR, Abu-Yousef M, Berbaum K. Retrospective evaluation of ultrasound features of thyroid nodules to assess malignancy risk: a step toward tirads. Am J Roentgenol. 2016;207:460–9. https://doi.org/10.2214/AJR.15.15121.
Moon WJ, Jung SL, Lee JH, Na DG, Baek JH, Lee YH, et al. Benign and malignant thyroid nodules: US differentiation-multicenter retrospective study. Radiology. 2008;247:762–70. https://doi.org/10.1148/radiol.2473070944.
Kamran SC, Marqusee E, Kim MI, Frates MC, Ritner J, Peters H, et al. Thyroid nodule size and prediction of cancer. J Clin Endocrinol Metab. 2013;98:564–70. https://doi.org/10.1210/jc.2012-2968.
Mohammadi A, Hajizadeh T. Evaluation of diagnostic efficacy of ultrasound scoring system to select thyroid nodules requiring fine needle aspiration biopsy. Int J Clin Exp Med. 2013;6:641–8.
Papini E, Guglielmi R, Bianchini A, Crescenzi A, Taccogna S, Nardi F, et al. Risk of malignancy in nonpalpable thyroid nodules: predictive value of ultrasound and color-doppler features. J Clin Endocrinol Metab. 2002;87:1941–6. https://doi.org/10.1210/jc.87.5.1941.
Phuttharak W, Somboonporn C, Hongdomnern G. Diagnostic performance of gray-scale versus combined gray-scale with colour doppler ultrasonography in the diagnosis of malignancy in thyroid nodules. Asian Pac J Cancer Prev. 2009;10(5):759–64.
De Nicola H, Szejnfeld J, Logullo AF, Wolosker AM, Souza LR, Chiferi V Jr. Flow pattern and vascular resistive index as predictors of malignancy risk in thyroid follicular neoplasms. J Ultrasound Med. 2005;24:897–904.
Rago T, Vitti P. Role of thyroid ultrasound in the diagnostic evaluation of thyroid nodules. Best Pract Res Clin Endocrinol Metab. 2008;22:913–28. https://doi.org/10.1016/j.beem.2008.09.016.
Frates MC, Benson CB, Doubilet PM, Cibas ES, Marqusee E. Can color Doppler sonography aid in the prediction of malignancy of thyroid nodules? J Ultrasound Med. 2003;22:127–314.
Anil G, Hegde A, Chong FHV. REVIEW Thyroid nodules: risk stratification for malignancy with ultrasound and guided biopsy. Cancer Imaging. 2011;11:209–23. https://doi.org/10.1102/1470-7330.2011.0030.
Iared W, Shigueoka DC, Cristófoli JC, Andriolo R, Atallah AN, Ajzen SA, et al. Use of color Doppler ultrasonography for the prediction of malignancy in follicular thyroid neoplasms: systematic review and meta-analysis. J Ultrasound Med. 2010;29:419–25.
de Silva K. Update on thyroid cancer management and the limitations faced by us. Sri Lanka J Diabetes Endocrinol Metab. 2012;2(1):21–4.
Pathirana AA, Bandara KGMW, Faleel MA, Kuruppumullage SD, Solangarachchi N, Rupasinghe R, et al. A sonographic scoring system to assess the risk of thyroid malignancy. Ceylon Med J. 2016;61(1):32–4.
Cohen JF, Korevaar DA, Altman DG, Bruns DE, Gatsonis CA, Hooft L, et al. STARD 2015 guidelines for reporting diagnostic accuracy studies: explanation and elaboration. BMJ Open. 2016. https://doi.org/10.1136/bmjopen-2016-012799.
Tessler FN, Middleton WD, Grant EG. Thyroid imaging reporting and data system (TI-RADS): a user’s guide. Radiology. 2018;287(1):29–36. https://doi.org/10.1148/radiol.2017171240.
Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopathology. Am J Clin Pathol. 2009;132(5):658–65. https://doi.org/10.1309/AJCPPHLWMI3JV4LA.
Takashima S, Fukuda H, Nomura N, Kishimoto H, Kim T, Kobayashi T. Thyroid nodules: re-evaluation with ultrasound. J Clin Ultrasound. 1995;23:179–84. https://doi.org/10.1002/jcu.1870230306.
Kim BK, Choi YS, Kwon HJ, Lee JS, Heo JJ, Han YJ, et al. Relationship between patterns of calcification in thyroid nodules and histopathologic findings. Endocr J. 2013;60:155–60. https://doi.org/10.1507/endocrj.ej12-0294.
Kakkos SK, Scopa CD, Chalmoukis AK, Karachalios DA, Spiliotis JD, Harkoftakis JG, et al. Relative risk of cancer in sonographically detected thyroid nodules with calcifications. J Clin Ultrasound. 2000;28(7):347–52.
Wang N, Xu Y, Ge C, Guo R, Guo K. Association of sonographically detected calcification with thyroid carcinoma. Head Neck. 2006;28:1077–83. https://doi.org/10.1002/hed.20481.
Kuma K, Matsuzuka F, Kobayashi A, Hirai K, Morita S, Miyauchi A, et al. Outcome of long standing solitary thyroid nodules. World J Surg. 1992;16:583–7. https://doi.org/10.1007/BF02067327.
Solbiati L, Osti V, Cova L, Tonolini M. Ultrasound of thyroid, parathyroid glands and neck lymph nodes. Eur Radiol. 2001;11:2411–24. https://doi.org/10.1007/s00330-001-1163-7.
Argalia G, D’Ambrosio F, Lucarelli F, Mignosi U, Giuseppetti GM, Passarini G, et al. Echo Doppler in the characterization of thyroid nodular disease. Radiol Med. 1995;89(5):651–7.
MCW, SR, NR and NDW were involved in the conception and design of the study. MCW, SR and SR collected data. NDW and MCW analysed and interpreted data. MCW and NDW prepared the manuscript. SR and NR made substantial contribution to revise the manuscript. All authors read and approved the final manuscript.
Authors would like to acknowledge the all the study participants for their support.
The authors declare that they have no competing interests.
Availability of data and materials
The datasets used and analysed during the present study are available from the corresponding author on reasonable request.
Consent for publication
Ethics approval and consent to participate
The study was carried out in accordance with The Code of Ethics of the World Medical Association (Declaration of Helsinki). Ethical clearance to conduct the study was obtained from the Institutional ethics review committee of the Faculty of Medicine, University of Peradeniya, Sri Lanka. Informed written consent form all the participants were obtained prior to data collection. (All the participants were above the age of 16 years.).
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Wettasinghe, M.C., Rosairo, S., Ratnatunga, N. et al. Diagnostic accuracy of ultrasound characteristics in the identification of malignant thyroid nodules. BMC Res Notes 12, 193 (2019). https://doi.org/10.1186/s13104-019-4235-y
- Diagnostic accuracy
- Ultrasound characteristics
- Thyroid malignancy
- Sri Lanka