Meta-analyses

Thyroid nodule ultrasound accuracy in predicting

thyroid malignancy based on TIRADS system
Wanlu Nie1,A, Lili Zhu2,C, Ping Yan1,B, Jie Sun1,E,F
1
Department of Ultrasound, Penglai People’s Hospital, Yantai, China
2
Deptartment of Endocrinology, Penglai People’s Hospital, Yantai, China

A – research concept and design; B – collection and/or assembly of data; C – data analysis and interpretation;
D – writing the article; E – critical revision of the article; F – final approval of the article

Advances in Clinical and Experimental Medicine, ISSN 1899–5276 (print), ISSN 2451–2680 (online) Adv Clin Exp Med. 2022;31(6):597–606

Address for correspondence

Wanlu Nie
Abstract
E-mail: [email protected] Background. A frequent prevalence of thyroid nodules in patients prioritizes the need for an accurate method
that characterizes them as benign or malignant. Fine-needle aspiration biopsy (FNAB) and thyroid ultraso-
Funding sources
None declared
nography (USG) are currently used for this purpose. However, since FNAB is complicated, time-consuming
and expensive, thyroid USG, a fast and highly sensitive method, is preferably used. Although USG is reported
Conflict of interest as a suitable method for characterization of thyroid nodules, there are some contrasting studies available
None declared which report its limited use in the differentiation of benign and malignant thyroid nodules.
Objectives. This meta-analysis aims to assess the accuracy of ultrasound in predicting thyroid cancer in terms
Received on December 21, 2021
of sensitivity, specificity and diagnostic odds ratios (ORs) for positive and negative results.
Reviewed on December 30, 2021
Accepted on February 17, 2022
Materials and methods. Systematic and extensive literature search on the use of ultrasound (US) to pre-
dict thyroid cancer was conducted in the databases of Scopus, CINAHL (via EBSCO), MEDLINE (via PubMed),
Published online on May 4, 2022 and Web of Science, covering the period from 2010 till 2021. The morphological features of thyroid nodules
observed during the USG were analyzed based on Thyroid Imaging Reporting And Data System (TIRADS)
guidelines. The accuracy of thyroid US was determined using parameters such as sensitivity, specificity, positive
likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic ORs. Moreover, the respective forest plot
and hierarchical summary receiver operating characteristics (HSROC) curve were plotted.
Results. A total of 2765 reference studies were examined, and among them, 15 relevant references were
selected. The selected studies were heterogeneous and included retrospective and prospective studies.
The risk of publication bias is low as the p-value for both Egger’s and Begg’s tests is >0.05. The overall
sensitivity of 92.53% (95% confidence interval (95% CI): [84.55%; 96.33%]), specificity of 33.88% (95% CI:
[23.16%; 45.53%]) and diagnostic OR of 12.36 (95% CI: [3.90%; 54.11%]) are achieved. These results were
statistically significant with a p-value < 0.001 and are predictive of US accuracy in detecting cancer.
Conclusions. The present meta-analysis, on the basis of statistically significant results, demonstrated
Cite as the high accuracy of thyroid ultrasound in detection of malignant nature of nodules in patients suspected
Nie W, Zhu L, Yan P, Sun J. Thyroid nodule ultrasound with a worrisome thyroid nodule.
accuracy in predicting thyroid malignancy based
on TIRADS system. Adv Clin Exp Med. 2022;31(6):597–606. Key words: ultrasound, thyroid nodule, fine-needle aspiration biopsy (FNAB), thyroid imaging – reporting
doi:10.17219/acem/146776
and data system (TIRADS), benign and malignant nodule
DOI
10.17219/acem/146776

Copyright
Copyright by Author(s)
This is an article distributed under the terms of the
Creative Commons Attribution 3.0 Unported (CC BY 3.0)
(https://creativecommons.org/licenses/by/3.0/)
598 W. Nie et al. Thyroid nodule and USG

Introduction routine US follow-up. A surgical procedure is considered

based on both US and FNAB results with cytology (FNAC).
All cancers are tumors, but not all tumors are cancer- Currently, different TIRADS guidelines4–8 are available,
ous. Therefore, it is imperative to have a concrete diag- including American College of Radiology (ACR) – TIRADS,
nosis in patients with thyroid nodules in order to dif- American Thyroid Association (ATA) – TIRADS, Amer-
ferentiate between benign and malignant nodules. With ican Association of Clinical Endocrinology/American
a rapid rise in the number of patients with suspicious Clinical Endocrinology/American Medical Endocrinology
thyroid nodules, it is a medical emergency to characterize (AACE/ACE/AME) – TIRADS, Korean Society of Thyroid
the nature of these nodules as either malignant or benign. Radiology (KSR)/(KSThR) – TIRADS, and European Thy-
In order to assess the malignancy and cancerous nature roid Association (ETA)/(EU) – TIRADS. These TIRADS
of the suspected thyroid nodule, fine-needle aspiration provide a list of notable morphological features reported
biopsy (FNAB) is performed.1 However, it is a high cost in the US images as a gold standard to classify a suspected
and time-consuming invasive surgical procedure and nodule as malignant or benign. These features are internal
it is not an easy diagnostic procedure for patients. There- calcifications, hypoechogenicity, vascularity, shape, and
fore, to simplify the nodule diagnosis, current thyroid nodule size.
guidelines (Thyroid Imaging Reporting And Data System As per these guidelines (Fig. 1), if the nodule is round
(TIRADS)) advocate the use of ultrasonography (USG) to ovoid with no solid portion, isoechoic, spongiform,
as a preliminary test for all patients suspected of having grows across the normal tissue plane in a parallel fash-
thyroid nodule.2,3 ion, has smooth margins and increased peripheral blood
Clinical variables and ultrasound (US) findings as per flow, and either has no calcification or egg-shell calcifica-
the TIRADS guidelines are recommended as a primary cri- tion, it is considered as benign. In contrast, if the nodule
terion to assess the benign and malignant nature of a nod- is solid, non-oval, taller than wider cells with irregular
ule. Based on these results, the clinician further advises margins, and has an increased central blood flow with
an additional confirmatory testing (FNAB) or simple marked hypoechogenicity and microcalcifications,

Fig. 1. Thyroid Imaging Reporting and Data System (TIRADS) guidelines

Adv Clin Exp Med. 2022;31(6):597–606 599

it is believed to be malignant and considered for further publication status and prospective or retrospective na-
analysis. ture of the study had no bearing on the study selection.
Among these morphological features of the thyroid, Table 1 presents the demographic features of the studies
USG and solid nodule are the most preferred criteria for included in the MEDLINE database search query along
assessing nodular malignancy. Therefore, many random with the evaluated factors.12–26
clinical trials were performed to assess similar US and The main goal of this study was to evaluate the effective-
FNAB results in predicting nodular malignancy in terms ness of USGat detecting thyroid nodules in people of vari-
of sensitivity and specificity. As a result, the sensitivity and ous ages. In order to assess the efficiency of US for nodule
specificity of various randomized controlled trials (RCTs) examination, patients of varied age groups were studied,
were reported in the range of 98% and 30%, respectively. and statistical parameters such as sensitivity, specificity,
These results align with the sensitivity and specificity val- positive likelihood ratio (PLR), negative likelihood ratio
ues in the range of 98% and 70%, respectively, reported for (NLR), and diagnostic odds ratios (ORs) were calculated
FNAB.9–11 with the help of true positive (TP), false positive (FP), true
Based on these studies, the use of US examination has negative (TN), and false negative (FN) values.
become widely accepted as a significant diagnostic step Two authors (WN and LZ) independently searched
in stratifying the risk of malignancy in the patients; still, the sources for similar studies. Full-text articles were col-
the diagnostic accuracy of several of examined sono- lected, and abstracts were analyzed if sufficient informa-
graphic parameters is a subject of much debate. Therefore, tion could be retrieved. Obsolete references were removed,
this meta-analysis was performed to understand the traits and only valuable studies were included. Data from the in-
of US that help in establishing the diagnosis of a thyroid cluded research were obtained separately by 2 researchers
nodule, either benign or malignant. The positive outcomes (PY and JS).
suggest its accuracy and comparable diagnostic efficiency
to FNAB, and highly recommend its use in the medical Inclusion and exclusion criteria
practice. The US examination enables patients to get their
nodule tested in a short time with no surgical procedures. Studies from the years 2010–2021 that examined the di-
In addition, it could save their money on confirmatory agnostic accuracy of USG for thyroid nodule assessment
tests and thus, have a significant impact on both clinical in the individuals of all ages with a suspected thyroid
practice and guideline recommendations. nodule and subsequent nodule therapy were included
in the study. Only full-text data were included in this analy-
sis; publications with inadequate data, reference standards
Objectives other than US/FNAB report and comparable studies pub-
lished before 2010 were all excluded, as shown in Fig. 2.
The current study is an attempt to analyze thyroid US
results according to TIRADS guidelines and the associated Evaluation of the analytical standard
nodule management, in order to establish this method
as reliable for predicting thyroid cancer in terms of sensi- Two authors (WN and LZ) independently examined
tivity and specificity for positive and negative US results. the methodological validity of the included studies using
the quality evaluation of diagnostic accuracy test assess-
ment instrument to establish their methodological quality
Materials and methods (QUADAS-2). One author (JS) was also in charge of ad-
dressing any issues that arose among other co-authors
This study, with the registration No. SUYP#/IRB/2021/ of this study.
1254, followed the normative requirements of Preferred
Reporting Items for Systematic Reviews and Meta-Anal- Statistical analyses
yses (PRISMA).
A 2 × 2 table was created to determine the pooled sensi-
Search strategy tivity, specificity and diagnostic OR using the DerSimonian
and Laird approach. A higher diagnostic OR number sug-
From 2010 until the end of March 2021, an exhaustive gests that the test is more accurate in its diagnosis. The I2
search was undertaken in MEDLINE (through PubMed), index and the Cochran’s Q statistic were used to determine
CINAHL (via EBSCO), Scopus, and Web of Science da- the heterogeneity of the studies. MedCalc software (Med-
tabases. Keywords like [ultrasonography], [FNAB], [thy- Calc Software Ltd., Ostend, Belgium) was used to create
roid nodules, malignancy], [TIRADS recommendations], the forest plots. The sensitivity and specificity data from
[US-based risk stratification methods] and [diagnostic the various studies are displayed in a hierarchical summary
accuracy] were used in this search. The PRISMA crite- receiver operating characteristics (HSROC) curve with their
ria were used to assess all of the papers. The language, respective 95% confidence intervals (95% CIs).
600 W. Nie et al. Thyroid nodule and USG

Table 1. Demographic summary of included studies with thyroid ultrasound in suspected cases of thyroid malignancy
Study ID Study Total sample Age Gender
Study type Type of US probe
and year duration size [years] M/F
Arpana et al.
cross-sectional 1 year 85 14–70 15/70 NR
201812
Esaote US machine (MyLab™ ClassC, Esaote, Genoa, Italy)
Al-Ghanimi et al.
retrospective 2 years 68 8–82 20/48 and electronically focused near‑field probes with a bandwidth
202013
of 7–12 MHz
Smith-Bindman
retrospective 5 years 11618 30–70 2277/9341 NR
et al. 201314
IU22 device (Philips Medical Systems, Bothell, USA; 5–12 MHz
Liu et al. 201915 retrospective 5 years 1568 18–80 412/1156 linear probe) or the S3000 device (Siemens Medical Solutions,
Mountain View, USA; 5–14 MHz linear probe)
The Mylab™ 90 (Esaote SpA, Genoa, Italy) ultrasound
image system was used for US examination, the L522
Luo et al. 202016 retrospective 2 years 296 30–50 54/168 probe (4–9 MHz; Esaote SpA) for CEUS and the L523 probe
(7.5–13.0 MHz, Esaote SpA) for conventional gray‑scale US,
CDUS and ES.
Kwak et al. 5–12 MHz linear-array transducer (iU22; Philips Medical
retrospective 8 months 1638 11–81 265/1373
201117 Systems.
GE VOLUSON 730 PRO machine (GE Healthcare, Milwaukee,
Srinivas et al.
prospective 4 years 365 18–68 22/334 USA) equipped with a 7.5–12 MHz high‑frequency linear array
201618
transducer with color and power Doppler capability.
GE Logic F8 ultrasound machine with a 6–12 MHz linear array
Mohanty et al. transducer and Samsung HS70A ultrasound machine with
prospective 1 year 50 40–50 10/40
201919 4–18 MHz linear array transducer (Samsung Neurologica Corp.,
Danvers, USA)
Nabahati et al. Samsung H60 ultrasound machine, with a 3–14 MHz linear
cross-sectional 2 years 718 14–83 NR
201920 array transducer (Samsung Neurologica Corp.)
linear array transducer (5–12 MHz)
Ghani et al.
retrospective 2 years 91 27–80 21/83 on ultrasound scanners HD11/HD11 XE/iU22 (Phillips Medical
201821
Systems) or Toshiba Xario200 (Toshiba Corp., Tokyo, Japan)
High frequency linear probe with 7.5 MHz bandwidth
Ram et al. 201522 cross-sectional 2 years 101 15–73 20/81
(models Zario and Nemio; Toshiba Corp.)
Wettasinghe
prospective 1.5 years 263 16–74 16/247 NR
et al. 201923
virtual organ computer-aided analysis; (VOCAL; GE Healthcare)
Azizi et al. 202124 prospective 1 year 355 40–50 45/310 and a 3-D multi-planar display with rendering in HDLive and
HDLive Silhouette (GE Healthcare).
Zayadeen et al. 5–12 MHz linear probe (iU22, Philips Healthcare)
retrospective 3 years 1466 11–96 265/1201
201625 or a 6–15-MHz linear probe (Logiq E9, GE Healthcare)
Richie and
retrospective 2 years 226 18–62 39/187 NR
Mellonie 202126

US – ultrasound; NR – not reported; CEUS – contrast-enhanced ultrasound; CDUS – color Doppler ultrasonography: ES – elastosonography.

Analysis of sensitivity Investigation of sources of heterogeneity

Excluding individuals with equivocal results might cause Meta-regression was used to investigate heterogeneity
diagnostic test accuracy to be overestimated. As a result, of the included experiments, introducing various sources
the sensitivity analysis was carried out, with uninterpre- of heterogeneity as covariates and fitting a bivariate model.
table data factored into the analysis. Finally, we compared To assess the covariate effect on the sensitivity and pre-
the outcomes of the primary analysis, which excluded cision, probability ratio test was used. A p-value <0.05
uninterpretable data, to those of the diagnostic precision was considered statistically significant for any of the sub-
analysis, which included all uninterpretable results. groups. Full-text publications compared to abstracts, high
compared to low risk of bias (RoB) in included studies,
prospective compared to retrospective studies, studies that
Adv Clin Exp Med. 2022;31(6):597–606 601

Fig. 2. Preferred Reporting

Items for Systematic Reviews
Identification of studies via database
and Meta-Analyses (PRISMA)
IDENTIFICATION

diagram
TIRADS – Thyroid Imaging
Reporting And Data
System; 18-FDG PET
relevant records selected from database search
– 18-fluorodeoxyglucose,
(n = 1246) positron emission tomography.

records screened records excluded

SCREENING

(n = 1019) due to invalid titles (n = 227)

reports sought for retrieval reports not retrieved due

(n = 235) to invalid references (n = 784)
ELIGIBILITY

reports assessed reports excluded (n = 194)

for eligibility (n = 41) reason 1 (n = 76) not using 18-F FDG
PET in identification of perineural
spread in head and neck tumor.
reason 2 (n = 64) insufficient data
for 2 × 2 tables.
reason 3 (n = 54) not in the field
EXCLUDED

of interest.
studies included (n = 14)

included only adults compared to mixed adult and pedi- The demographic details of the studies included in this
atric population studies, the proportion of female partici- meta-analysis are shown in Table 1. It describes the authors
pants, the proportion of obese patients, type of US probe, of each included study, year of publication, type of study,
and ultrasonographer experience were among the inves- duration of the study, total sample size, type of US probe
tigated heterogeneity sources. used in the study, age, gender, and the total number of nod-
ules on which US was conducted. In addition, the mor-
phological features of the thyroid nodule, as suggested
Results by TIRADS guidelines (Fig. 1), are set as a gold standard for
its characterization. A total of 18,908 patients were included
Literature search results in all analyzed studies. Four of the studies were prospective,
8 were retrospective and 3 were cross-sectional, and they
Through computerized scanning, a total of 2765 studies were all published as full-text publications. The partici-
were retrieved. We eliminated 745 articles based on their pants’ age ranged from 8 to 80 years, and the information
titles and abstracts, and 1825 papers due to faulty refer- regarding the type of utilized US probe was provided.
ences. Owing to duplicity, about 165 out of the remaining
195 studies were removed. Finally, 30 full-text publications Risk of bias assessment
were screened. Among these, 15 were eliminated due to in-
clusion requirements. As a result, as shown in Fig. 2, this The estimated sensitivity value ranged from 74% to 98%,
meta-analysis included 15 papers that satisfied the inclu- whereas the estimated specificity value ranged from 8%
sion criteria, namely morphological characteristics of thy- to 84%. According to the QUADAS-2 tool, all of the in-
roid USG as per TIRADS standards. The main grounds cluded experiments had a low likelihood of bias, as indi-
for omission were inadequate evidence and improper com- cated in Table 2. Figure 3 shows a Duke funnel plot used
parison criteria needed for creating 2 × 2 tables for review. to assess the possibility of publication bias.
602 W. Nie et al. Thyroid nodule and USG

Meta-analysis results Table 2. Exploration of heterogeneity sources; the impact of sample

subgroups or participant characteristics on overall sensitivity and
specificity
The overall sensitivity of the US scan for thyroid nodule
was 92.53% (95% CI: [84.55%; 96.33%]) and overall specificity Subgroup p-value

was 33.88% (95% CI: [23.16%; 45.53%]) with p < 0.001, indica- Full texts compared to abstracts NA
tive of statistical significance. The value of the overall PLR High compared to low risk of bias NA
was 6.90 (95% CI: [2.66; 23.8]), and the overall NLR value was Prospective compared to retrospective studies 0.024*
0.71 (95% CI: [0.59; 0.85]), as shown in Table 3. These results Adults compared to mixed population 0.924
proved the high accuracy of US scan in detecting only posi- Proportion of female participants 0.05*
tive nodules as malignant. The summary receiver operating
Proportion of obese participants NA
characteristic (SROC) plot showing an estimate of sensitiv-
Type of ultrasound probe 0.034*
ity compared to specificity and area under the SROC curve,
as shown in Fig. 4, indicates its positive efficiency. The box Ultrasonographer experience 0.001*

and whisker plot (Fig. 5) clearly shows that the diagnostic Clinical probability of TC 0.001*
accuracy of thyroid US is high, as the number of TP results TC – thyroid cancer; NA – not available. The details could not be retrieved
is high, similarly to the results of FNAB, while number of FP from the report, or only one party was present; *significant impact
results is low. The diagnostic OR was 12.36 (95% CI: [3.90; of the subgroup on summary results.

54.11]), as shown in Table 4. As reported, the diagnostic

OR higher than 10 indicates the positive outcome of a test, usually the same.9,10 Therefore, early nodule characteriza-
as shown in the forest plot in Fig. 6. Our results are con- tion is of extreme importance since a benign nodule can
gruent with those reports and suggest a greater accuracy be easily cured, but treating a malignant nodule is com-
of thyroid US in diagnosing thyroid cancer. plex and depends on its stage. Therefore, the early detec-
tion of malignant nodules increases a patient’s chances
of treatment and survival rate; otherwise, thyroid cancer
Discussion can be fatal.
Fine-needle aspiration biopsy is the gold standard1,2,9–11
Accurate diagnosis of a worrisome thyroid nodule for for accurately detecting malignancy in thyroid nodule
malignancy has always been challenging because initial patients, with a substantial sensitivity of 98%, according
symptoms of both nodules, either benign or malignant, are to numerous studies.12–26 It is, however, rarely chosen since

Fig. 3. Duke funnel plot test for publication bias

Adv Clin Exp Med. 2022;31(6):597–606 603

Table 3. Sensitivity and specificity of different studies

95% CI upper 95% CI lower 95% CI upper 95% CI
Study ID and year Specificity [%] Sensitivity [%]
limit limit limit lower limit
Kwak et al. 201117 24.06 21.51 26.75 96.66 94.88 97.95
Smith-Bindman et al. 201314 23.61 18.83 28.95 87.94 83.56 91.50
22
Ram et al. 2015 8.06 2.67 17.83 97.50 86.84 99.94
Zayadeen et al. 201625 14.25 11.98 16.78 97.57 96.46 98.41
18
Srinivas et al. 2016 48.15 28.67 68.05 96.45 93.88 98.15
Ghani et al. 201821 23.08 11.13 39.33 93.62 82.46 98.66
Arpana et al. 201812 32.35 17.39 50.53 88.57 73.26 96.80
Wettasinghe et al. 201923 13.68 9.55 18.75 96.55 82.24 99.91
Luo et al. 202016 84.52 77.84 89.82 74.63 62.51 84.47
Liu et al. 201915 57.93 55.12 60.71 84.00 79.89 87.56
20
Nabahati et al. 2019 8.98 6.37 12.21 96.18 94.09 97.68
Mohanty et al. 201919 60.00 36.05 80.88 95.45 77.16 99.88
24
Azizi et al. 2021 14.08 10.21 18.74 92.65 83.67 97.57
Al-Ghanimi et al. 202013 50.00 15.70 84.30 91.67 81.61 97.24
Richie and Mellonie 202126 45.45 24.39 67.79 98.53 95.76 99.70

95% CI – 95% confidence interval.

Fig. 4. Hierarchical summary receiver operating characteristics (HSROC) curve sensitivity compared to specificity
604 W. Nie et al. Thyroid nodule and USG

Table 4. Diagnostic OR of cases studied

Malignant 95% CI 95% CI
Benign nodule Benign nodule Malignant nodule Diagnostic odds
Study ID and year nodule (solid upper lower
(simple cyst) (solid cyst) (simple/mixed cyst) ratio
cyst) limit limit
Kwak et al. 201117 578.00 805.00 255.00 20.00 9.15 5.74 14.61
Bindmann et al. 201314 248.00 220.00 68.00 34.00 2.25 1.44 3.54
22
Ram et al. 2015 39.00 57.00 5.00 1.00 3.42 0.38 30.43
Zayadeen et al. 201625 1043.00 734.00 122.00 26.00 6.67 4.32 10.29
18
Srinivas et al. 2016 326.00 14.00 13.00 12.00 25.23 9.76 65.20
Ghani et al. 201721 44.00 30.00 9.00 3.00 4.40 1.09 17.60
12
Arpana et al. 2018 31.00 23.00 11.00 4.00 3.71 1.05 13.13
Wettasinghe et al. 201923 28.00 202.00 32.00 1.00 4.44 0.58 33.75
16
Luo et al. 2020 50.00 24.00 131.00 17.00 16.05 7.50 32.37
Liu et al. 201915 315.00 517.00 712.00 60.00 7.23 5.36 9.74
20
Nabahati et al. 2019 478.00 365.00 36.00 19.00 2.48 1.40 4.39
Mohanty et al. 201919 21.00 8.00 12.00 1.00 31.50 3.50 283.30
Azizi et al. 202124 63.00 238.00 39.00 5.00 2.06 0.78 5.45
Ghanimi et al. 202113 55.00 4.00 4.00 5.00 11.00 2.08 57.91
Richi et al. 202126 201.00 3.00 10.00 12.00 55.80 13.50 229.90

OR – odds ratio; 95% CI – 95% confidence interval.

The studies covered a wide range of sensitivity, ranging

from 74% to 95% with a 95% CI of [60%; 95%], while in-
cluded studies revealed a wide specificity range ranging
from 8% to 85% (95% CI: [2%; 90%]). Latif et al., in a re-
search similar to ours, evaluated FNAB and USG to di-
agnose benign and malignant thyroid nodules.27 The gold
standard in this investigation was surgery or follow-up.
In this study, the combined sensitivity and specificity
in the adult population were 90% and 77%, respectively,
supporting the use of US in diagnosing thyroid cancer.
Similarly, Ghani et al.21 showed a sensitivity of 100% and
specificity of 91.4%, Wettasinghe et al.23 calculated sen-
sitivity and specificity of 0.13% and 0.95%, respectively,
Luo et al.16 showed a sensitivity and specificity of 0.72%
and 0.83%, respectively, and Richie and Mellonie showed
the high specifcity of 45%.12,23,16,26 All these studies, simi-
larly to the present study, are in support of the applica-
Fig. 5. Box and whisker plot for cumulative positive value (CPV) compared
tion of US imaging for the detection of the malignant
to cumulative negative value (CNV) of samples studied tumor. However, in contrast to the present analysis,
Jiang et al. 28 observed different results and concluded
that the US should not be used for diagnosing thyroid
it is time-consuming, intrusive and costly. The current cancer cases.
meta-analysis is an excellent step in simplifying the nodule Thyroid USG was very accurate when combined with
characterization technique by demonstrating that thyroid FNAB by Salam et al., although, unlike the current study,
USG and FNAB are equally effective in diagnosing prob- that study did not employ any reference standards (mor-
able thyroid cancer in patients of all ages. phological properties of the nodule) to limit the chances
In this meta-analysis, a total of 15 publications were of FN results.4 The positive and negative probability ratios
chosen to predict the specificity, sensitivity, PLR, NLR, were 6.90 (95% CI: [2.66; 23.8]) and 0.71 (95% CI: [2.66;
and diagnostic ORs. Overall sensitivity of 92.53% (95% CI: 23.8]), respectively. The diagnostic OR in this research
[84.55%; 96.33%]) and specificity of 33.88% (95% CI: was 12.36 (95% CI: [3.90; 54.11]), indicating that thyroid
[23.16%; 45.53%]) were found in this study. The diag- USG has a substantial accuracy rate in predicting thyroid
nostic OR was found to be 12.36 (95% CI: [3.90; 54.11]). cancer. The SROC curve of the current study indicates
Adv Clin Exp Med. 2022;31(6):597–606 605

Arpana et al. 201812

Al-Ghanimi et al. 202013

Smith-Bindman et al. 201314

Liu et al. 201915

Luo et al. 202016

Kwak et al. 201117

Srinivas et al. 201618

Mohanty et al. 201919

Nabahati et al. 201920

Ghani et al. 201821

Ram et al. 201522

Wettasinghe et al. 201923

Azizi et al. 202124

Zayadeen et al. 201625

Richie and Mellonie 202126

Total (fixed effects)

Total (random effects)

0.0001 0.001 0.01 0.1 1 10 100

odds ratio
Fig. 6. Forest plot for the diagnostic odds ratio (OR) of case studies100100

the combined effect of sensitivity and specificity, with Limitations

an inclination of the curve towards the upper left, show-
ing high quality of thyroid US diagnostic accuracy. The diversity of US equipment utilized and tests per-
Azizi et al. investigated the hypothesis stating that formed by various sonographers influence the prob-
three-dimensional ultrasound (3-D-US) allows for dis- ability of FN results and are a drawback of this study.
tinguishing benign from malignant thyroid nodules Similar diagnosis with FNAB was not specified in many
with better sensitivity and specificity than two-dimen- types of research, having an influence on the appropriate
sional ultrasound (2-D-US). 24 They used both 2-D-US analysis of the comparability of data. Data from other rel-
and 3-D-US to examine 344 thyroid nodules, followed evant studies that demonstrate the diagnostic accuracy
by a FNAB. Based on the appearance of the margins of US in contrast to other diagnostic imaging modalities
in 3-D-US, TNs were categorized into 4 categories. might also be provided to emphasize its significance.
The researchers employed bivariate and multivariate To clearly distinguish between a benign and a malig-
analyses. In 40 individuals, surgical pathology revealed nant nodule, specific information about a patient’s case
44 thyroid malignancies. In malignant TNs, uneven history, physical examination and pathological testing
margins and microcalcifications (p < 0.001) were more can help to improve the diagnostic accuracy rate of USG
common on 2-D-US. The sensitivity and specificity of ir- in predicting thyroid cancer.
regular margins on 2-D-US were 61.4% and 79.3%, re-
spectively.27 The sensitivity and specificity of irregular
margins on 3-D-US were 86.4% and 83.3%, respectively. Conclusions
Microcalcifications and irregular margins on 2-D-US
had better sensitivity, specificity, as well as positive Ultrasound is a widespread diagnostic investigation
and negative predictive values than irregular margins tool since it is easy to use, inexpensive and efficient, even
on 3-D-US. The 3-D-US evaluation of TN margins had though FNAB has considerably reduced the mortality rate
higher sensitivity and specificity than 2-D-US in distin- owing to its complex surgical technique and high cost.
guishing benign TNs from malignant ones. It is a noninvasive, nonionizing radiation approach for
606 W. Nie et al. Thyroid nodule and USG

nodule identification in patients of all ages and an efficient 14. Smith-Bindman R, Lebda P, Feldstein VA, et al. Risk of thyroid cancer
based on thyroid ultrasound imaging characteristics: Results of a pop-
diagnostic method to decrease the difficulties associated
ulation-based study. JAMA Intern Med. 2013;173(19):1788–1796.
with FNAB. doi:10.1001/jamainternmed.2013.9245
15. Liu T, Guo Q, Lian C, et al. Automated detection and classification
ORCID iDs of thyroid nodules in ultrasound images using clinical-knowledge-
Wanlu Nie  https://orcid.org/0000-0003-4565-7631 guided convolutional neural networks. Med Image Anal. 2019;58:
Lili Zhu  https://orcid.org/0000-0002-6734-4647 101555. doi:10.1016/j.media.2019.101555
Ping Yan  https://orcid.org/0000-0002-3533-9413 16. Luo W, Zhang Y, Yuan J, et al. Differential diagnosis of thyroid nod-
Jie Sun  https://orcid.org/0000-0003-0644-3657 ules through a combination of multiple ultrasonography techniques:
A decision-tree model. Exp Ther Med. 2020;19(6):3675–3683. doi:10.
3892/etm.2020.8621
References 17. Kwak JY, Han KH, Yoon JH, et al. Thyroid imaging reporting and data
1. Chaturvedi R, Kumar A, Balasubramanian B, Sreehari S. A retrospec- system for US features of nodules: A step in establishing better strat-
tive study correlating ultrasound based Thyroid Imaging Reporting ification of cancer risk. Radiology. 2011;260(3):892–899. doi:10.1148/
and Data System (TIRADS) with Bethesda system for thyroid cytopa- radiol.11110206
thology in thyroid nodule risk stratification. NEMJ. 2021;2(2):121–128. 18. Srinivas MN, Amogh VN, Gautam MS, et al. A prospective study to
doi:10.2174/0250688203666210111152307 evaluate the reliability of thyroid imaging reporting and data system
2. Chen H, Ye J, Song J, You Y, Chen W, Liu Y. Comparison of different in differentiation between benign and malignant thyroid lesions.
ultrasound classification systems of thyroid nodules for identify- J Clin Imaging Sci. 2016;6:5. doi:10.4103/2156-7514.177551
ing malignant potential: A cross-sectional study. Clinics (Sao Paulo). 19. Mohanty J, Sanket, Mishra P. Role of ACR-TIRADS in risk stratification
2021;76:e2126. doi:10.6061/clinics/2021/e2126 of thyroid nodules. Int J Res Med Sci. 2019;7(4):1039–1043. doi:10.18203
3. Hahn SY, Shin JH, Oh YL, Park KW. Ultrasound-guided core needle /2320-6012.ijrms20191076
biopsy techniques for intermediate or low suspicion thyroid nodules: 20. Nabahati M, Moazezi Z, Fartookzadeh S, Mehraeen R, Ghaemian N,
Which method is effective for diagnosis? Korean J Radiol. 2019;20(10): Sharbatdaran M. The comparison of accuracy of ultrasonographic
1454–1461. doi:10.3348/kjr.2018.0841 features versus ultrasound-guided fine-needle aspiration cytology
4. Al-Salam S, Sharma C, Abu Sa’a MT, et al. Ultrasound-guided fine in diagnosis of malignant thyroid nodules. J Ultrasound. 2019;22(3):
needle aspiration cytology and ultrasound examination of thyroid 315–321. doi:10.1007/s40477-019-00377-2
nodules in the UAE: A comparison. PLoS One. 2021;16(4):e0247807. 21. Ghani FA, Nurismah MI, Husyairi H, Shahrun Niza AS, Radhika S.
doi:10.1371/journal.pone.0247807 Reliability of the ultrasound classification system of thyroid nod-
5. Kim SC, Kim JH, Choi SH, et al. Off-site evaluation of three-dimen- ules in predicting malignancy. Med J Malaysia. 2018;73(5):263–271.
sional ultrasound for the diagnosis of thyroid nodules: Comparison PMID:30350802.
with two-dimensional ultrasound. Eur Radiol. 2016;26(10):3353–3360. 22. Ram N, Hafeez S, Qamar S, et al. Diagnostic validity of ultrasonog-
doi:10.1007/s00330-015-4193-2 raphy in thyroid nodules. J Pak Med Assoc. 2015;65(8):875–878.
6. Xie C, Cox P, Taylor N, LaPorte S. Ultrasonography of thyroid nod- PMID:26228335.
ules: A pictorial review. Insights Imaging. 2016;7(1):77–86. doi:10.1007/ 23. Wettasinghe MC, Rosairo S, Ratnatunga N, Wickramasinghe ND.
s13244-015-0446-5 Diagnostic accuracy of ultrasound characteristics in the identifica-
7. Tessler FN, Middleton WD, Grant EG, et al. ACR Thyroid Imaging, tion of malignant thyroid nodules. BMC Res Notes. 2019;12(1):193.
Reporting and Data System (TI-RADS): White paper of the ACR TI- doi:10.1186/s13104-019-4235-y
RADS committee. J Am Coll Radiol. 2017;14(5):587–595. doi:10.1016/j. 24. Azizi G, Faust K, Ogden L, et al. 3-D ultrasound and thyroid cancer diag-
jacr.2017.01.046 nosis: A prospective study. Ultrasound Med Biol. 2021;47(5):1299–1309.
8. Nam SJ, Kwak JY, Moon HJ, Yoon JH, Kim EK, Koo JS. Large (≥3cm) doi:10.1016/j.ultrasmedbio.2021.01.010
thyroid nodules with benign cytology: Can Thyroid Imaging Report- 25. Zayadeen AR, Abu-Yousef M, Berbaum K. Retrospective evaluation
ing and Data System (TIRADS) help predict false-negative cytology? of ultrasound features of thyroid nodules to assess malignancy risk:
PLoS One. 2017;12(10):e0186242. doi:10.1371/journal.pone.0186242 A step toward TIRADS. AJR Am J Roentgenol. 2016;207(3):460–469.
9. Russ G, Bonnema SJ, Erdogan MF, Durante C, Ngu R, Leenhardt L. doi:10.2214/AJR.15.15121
European Thyroid Association guidelines for ultrasound malignan- 26. Richie AJ, Mellonie P. Accuracy of thyroid imaging and reporting
cy risk stratification of thyroid nodules in adults: The EU-TIRADS. data systems in risk stratification of thyroid nodules: A retrospective
Eur Thyroid J. 2017;6(5):225–237. doi:10.1159/000478927 observational study. Int J Anat Radio Surg. 2021;10(1):58–61. doi:10.
10. Trimboli P, Durante C. Ultrasound risk stratification systems for thy- 7860/IJARS/2021/47306:2627
roid nodule: Between lights and shadows, we are moving towards 27. Latif MA, El Rakhawy MM, Saleh MF. Diagnostic accuracy of B-mode
a new era. Endocrine. 2020;69(1):1–4. doi:10.1007/s12020-020-02196-6 ultrasound, ultrasound elastography and diffusion weighted MRI
11. Colakoglu B, Yildirim D, Alis D, et al. Elastography in distinguishing in differentiation of thyroid nodules (prospective study). Egypt J Radiol
benign from malignant thyroid nodules. J Clin Imaging Sci. 2016;6:51. Nucl Med. 2021;52:256. doi:10.1186/s43055-021-00640-9
doi:10.4103/2156-7514.197074 28. Jiang D, Zang Y, Jiang D, Zhang X, Zhao C. Value of rapid on-site evalua-
12. Arpana, Panta OB, Gurung G, Pradhan S. Ultrasound findings in thy- tion for ultrasound-guided thyroid fine needle aspiration. J Int Med Res.
roid nodules: A radio-cytopathologic correlation. J Med Ultrasound. 2019;47(2):626–634. doi:10.1177/0300060518807060
2018;26(2):90–93. doi:10.4103/JMU.JMU_7_17
13. Al-Ghanimi IA, Al-Sharydah AM, Al-Mulhim S, et al. Diagnostic accu-
racy of ultrasonography in classifying thyroid nodules compared
with fine-needle aspiration. Saudi J Med Med Sci. 2020;8(1):25–31.
doi:10.4103/sjmms.sjmms_126_18