NEW PROPOSED FORMULA OF TI-RADS CLASSIFICATION BASED ON ULTRASOUND FINDINGS

Abstract

Introduction.

The present study aimed to introduce a new formula for classification of nodules in TI-RADS and describe ultrasonography features of benign and malignant thyroid nodules.

Methods.

This study was conducted on thyroid mass in 1033 patients. The incidence of malignancy for thyroid nodules was determined by selecting malignancy coefficients. Then the patients were first classified using conventional TI-RADS classification criteria and once again according to a new proposed formula.

Results.

Among ultrasonography features of thyroid nodules, the irregular shape (46.7%), unclear margin (47.3%), extension to the capsule (irregular and infiltrative margin) (85%), the marked hypo-echoic nodules (63.8%), micro-calcification (49%), and to have vertical axis (74.0%) were associated with high incidence of malignancy.

Conclusion.

According to the proposed new formula for TI-RADS, there are four coefficients of 7, 3, 1 and 0 for incidence of malignancy of each one of ultrasound findings that help to standardization and unifying of TI-RADS classification. The incidence of malignancy in TI-RADS classification according to the new proposed formula was achieved as follows: group 2: 0.0%, group 3: 0.7%, groups 4a, 4b, 4c: 16.7%, 43.4%, 68.5%, and group 5: 95.2%, respectively.

INTRODUCTION

The high prevalence of thyroid nodules and their malignancies indicate the importance of a timely and accurate diagnosis and differentiation of benign and malignant nodules (3). Fine-needle aspiration biopsy (FNAB) is now considered as a standard method for malignancy diagnosis of thyroid nodules (4); however, its invasive and time-consuming nature, the increased therapeutic costs, the refusal of FNAB by patients due to the fear of possible complications, the high number of non-diagnostic and AUS specimens were limitations of this diagnostic method (4-6). Also, over-diagnosis of thyroid cancer cases was another problem of applied FNAB in treating thyroid nodules, as studies found that in the absence of FNAB in more than a half of thyroid nodules, a thyroid tumor never caused symptoms or death in patients (6). Therefore, it is still necessary to find a non-invasive diagnostic procedure ensuring the necessity of performing FNAB considering the higher probability of thyroid malignancy.

The rate of diagnosed thyroid nodules has increased by two to four times over the past three decades due to advances in imaging techniques (1, 2). Ultrasonography (US) is basically a quick and available method for evaluating thyroid nodules through which the information can be obtained about the anatomical and structural properties of a nodule (7). However, due to the operator-based result of the US leading to a different analysis of the same thyroid nodule image between several sonographers, there is a need for taking measures for the lack of precise standards in the definition and differentiation of benign nodules from malignant ones (7, 8). Therefore, many efforts have been made by various research groups in recent years to find a system for differentiating benign thyroid nodules from malignant ones more accurately (3).

The TI-RADS (Thyroid Imaging Reporting and Data System) is a proposed new classification system for evaluating the malignancy risk of thyroid nodules classifying thyroid nodules into several classes according to their ultrasound shape and appearance (9). Further research is needed to achieve the best criteria for differentiating nodules and to recognize the features that affect the increase in the incidence of malignancy, and alone or in combination with other factors, help to standardization and unifying of TI-RADS classification (10-13).

Based on the FNAB results in sure cases and the pathological report in suspicious cases, the present study aimed to evaluate the incidence of malignancy in each of ultrasonographic findings of applied thyroid nodules in the TI-RADS classification as well as providing a formula for predicting the incidence of malignancy of thyroid nodules.

MATERIAL AND METHODS

Research design and setting

This cross-sectional study was done on all patients with suspected thyroid malignancy referring to the specialized thyroid clinic affiliated to outpatient clinic related to Ghaem and Omid Hospitals of Mashhad University of Medical Science in Iran during 2015-2018. Inclusion criteria included the discovery of thyroid nodules during the initial examination and written consent of patients. There was no age limit as an input criterion, and sampling was done by census. Patients were excluded from the study in the case that the necessary information was not properly collected in each sampling step.

Data collection

Following the initial interview for obtaining the information such as patients’ age, gender and medical history, all patients underwent the US using a digital ultrasound scanner system (Italy, Esaote, class C) and a linear electronic probe with a central frequency of 8 to 16 MHz. All patients were examined by only this ultrasound machine and by an experienced radiologist. Grayscale images and, if necessary, color Doppler of the thyroid and its nodules, and cervical lymph nodes in the supine position with the neck extension were obtained in order to create the maximum exposure in the anterior neck. The main anatomical features of nodules including shapes, margin, structures, echogenicity, heterogeneity, echogenic dot, an extension to the capsule, vertical axis position, and nodule vascularity were evaluated for each thyroid nodule. Doppler study was mainly used to diagnose the relatively common thyroiditis and resulting white knight nodules.

TI-RADS classification by conventional method

Based on reference guidelines (14, 15), the thyroid nodule status of each patient was classified into different TI-RADS groups (Table 1). Following classification of patients according to the TI-RADS, patients in groups 3 to 5 with nodule sizes of greater than 1 cm were subjected to FNAC; and the rest of them were subjected to the follow-up stage.

Table 1.

TI-RADS criteria classification based on ultrasonography features of thyroid nodules

Level		Definition
TI-RADS 1	Normal thyroid	Normal thyroid
TI-RADS 2	Benign lesions	Benign lesions*
TI-RADS 3	Probably benign	Iso/hyperechoic nodule without suspicious feature#
TI-RADS 4a	Slightly suspicious	One suspicious feature
TI-RADS 4b	Suspicious	Two-Three suspicious features
TI-RADS 4c	Highly suspicious	Four suspicious features
TI-RADS 5	Highly suggestive of malignancy	Five or more suspicious features
TI-RADS 6	Proven malignancy	FNA proven malignancy

*: Simple and multinodular Goiters, Adenomatoid goiters (isoechoic spongiform masses), Thyroiditis, white knight nodules in thyroiditis background. Cysts (simple, colloid, spongious, with or without small iso-echoic mass).

# Suspicious features: features with malignancy coefficients equivalents 1until 7 in according to Table 2. (Irregular shape, Unclear or thick or Infiltrative margin, Almost solid, Hypoechogenicity, Micro or macrocalcification, Vertical axis, Extension to nodule capsule and tumoral adenopathy).

Fine Needle Aspiration Cytology (FNAC)

A 25 gauge needle was used for nodal aspiration. Aspirations were directly stained on 4-6 smear glass slides by May-Grunwald-Giemsa. The Bethesda system was used for the cytopathological report. All samples were examined by two skilled cytopathologists. The cytopathology results were reported based on the classification system as either benign nodules including adenomatous goiter, thyroiditis, follicular benign nodules, or malignant nodules including papillary thyroid carcinoma (PTC), suspicious for PTC, follicular thyroid carcinoma (FTC), medullary carcinoma, lymphoma, and anaplastic tumor.

Proposed TI-RADS classification formula

In the present study, a new proposed formula

was proposed for TI-RADS classification, where,

S: Total malignancy coefficients per patient;

i: Appearance features of thyroid nodules;

Xi: Coefficient of presence or absence of thyroid in patients;

(Xi= 1: a patient has appearance features of thyroid nodules; Xi = 0: the patient does not have appearance feature of thyroid nodules);

Wi: Coefficient of malignancy associated with each appearance features of thyroid nodules.

The incidence of malignancy of each thyroid nodule (Pi) was first calculated based on the percentage of patients with malignant nodules in the total patients with that thyroid feature. Then, Wi value was calculated based on the Pi value. If Pi< 7%, then Wi=0; If 7%≤ Pi<20%, then Wi=1; If 20%≤ Pi< 50%, then Wi=3; and if 50%≤ Pi< 90%, then Wi=7. The Pi value>90% was not observed in the study. Table 2 presents all features of thyroid nodule (i), the incidence of malignancy of each thyroid nodule (Pi) and its incidence of malignancy coefficients (Wi). The formula above summarized as follows:

Table 2.

Malignancy coefficients according to ultrasonography features of thyroid nodules

Thyroid nodule	Thyroid nodule features (i)	Number of patients	Patients with malignant nodules	Incidence of malignancy (Pi)	Malignancy coefficient (Wi)
Nodule shape	Regular	881	22	P1= 2.5%	W1=0
	Irregular	152	71	P2= 46.7%	W2=3
Nodule margin	Clear	699	20	P3= 2.8%	W3=0
	Unclear	150	71	P4= 47.3%	W4=3
	Thin halo	158	1	P5= 0.6%	W5=0
	Thick halo	7	1	P6= 14.28%	W6=1
	Infiltrative	20	17	P7= 85%	W7=7
	Marginal calcification	19	0	P8= 0%	W8=0
Nodule structure	Simple cyst	25	0	P9= 0%	W9=0
	Spongy cyst	42	0	P10= 0%	W10=0
	Colloid cyst	26	0	P11= 0%	W11=0
	Cyst with small partial solidity	41	0	P12= 0%	W12=0
	Solid with large cystic areas	143	1	P13= 0.7%	W13=0
	Spongiform mass	224	2	P14= 0.9%	W14=0
	Solid mass with a few cysts	80	8	P15= 10%	W15=1
	Solid without cyst	400	65	P16= 16.3%	W16=1
	White knight	32	0	P17= 0%	W17=0
Nodule Echogenicity	Isoecho	228	4	P18= 1.75%	W18=0
	Hypoecho	306	5	P19= 1.63%	W19=0
	Iso and hyperecho	75	0	P20= 0%	W20=0
	Partial Hypoecho	168	14	P21= 8.4%	W21=1
	Iso and Hypoecho	35	1	P22= 2.8%	W22=0
	Hyper and Hypoecho	20	2	P23= 10%	W23=1
	Iso and hyper and hypoecho	8	0	P24= 0%	W24=0
	Marked Hypoecho	105	67	P25= 63.8%	W25=7
Nodule Heterogeneity	Homogeneous	583	37	P26= 6.3%	W26=0
	Heterogeneous	353	55	P27= 15.6%	W27=1
	Homogeneous/ Heterogeneous	7	1	P28= 14.3%	W28=1
Echogenic particles	No echogenic particles	781	35	P29= 4.48%	W29=0
	Microcalcification	98	48	P30= 49%	W30=3
	Macrocalcification	92	8	P31= 8.7%	W31=1
	Echogenic foci with comet-tail (Colloid)	30	1	P32= 3.33%	W32=0
	Echogenic foci	32	1	P33= 3.13%	W33=0
Extension to nodule capsule	It does not have	1018	80	P34= 7.86%	W34=0
	It has	15	13	P35= 86.7%	W35=7
Vertical axis	It does not have	1010	76	P36= 7.52%	W36=0
	It has	23	16	P37= 74%	W37=7

Table 3.

Results of TI-RADS classifications based on conventional methods and the proposed new formula

TI-RADS Classification	Number of patients		Frequency percentage		Patients with malignant nodules		Possibility of malignancy
	C-M*	P-N-F#	C-M	P-N-F	C-M	P-N-F	C-M	P-N-F
2	224	224	21.7%	21.7%	0	0	0%	0%
3	509	577	49.3%	55.9%	2	4	0.4%	0.7%
4a	180	120	17.4%	11.6%	15	20	8.3%	16.7%
4b	52	53	5%	5%	19	23	36.5%	43.4%
4c	31	38	3%	3.7%	23	26	74.2%	68.5%
5	37	21	3.6%	2%	34	20	91.9%	95.2%

*: Conventional Methods. #: Proposed New Formula

1. All features with non-zero malignancy coefficients are identified in the patient.

2. All malignancy coefficients are summed together and the result is stored in S.

3. Given the S value, the TI-RADS classification is performed according to Table 4.

Table 4.

Relation between total malignancy coefficients (S) and TI-RADS classifications based on the proposed new formula

TI-RADS classification	Total malignancy coefficients (S)
2	0
3	1, 2
4a	3-9
4b	10-16
4c	17-23
5	24-45

Statistical analysis

Statistical analysis was performed using SPSS16 (SPSS Inc. Released in 2009. PASW Statistics for Windows, Chicago: SPSS Inc.). Continuous variables were presented as mean± SD; and categorical variables as numbers and percentages. The incidence of malignancy in each TI-RADS classification was reported based on the conventional method, and a new proposed formula was reported as a frequency table. Diagnostic values of both TI-RADS classification criteria were calculated using Kapa for agreement. Chi-Square test was used to examine relationships between qualitative variables. In all tests, P-value<0.05 was considered as a significance level.

Research ethics

All patients were informed about the project content before the inclusion in the study, and their information was used to carry out the project in the case of signing the consent form. Patients’ privacy and dignity were respected in the project. Patients’ information was entered as codes into statistical analysis programs, and the information was published as a general result. Patients could freely leave the research project at any stage. In the case of any complications during the research project, researchers were responsible for compensation. The present research was an approved research project by Mashhad University of Medical Sciences (Code No. 951257).

RESULTS

Baseline characteristics information

One thousand seventy-six thyroid patients were studied. Among them, 43 patients were excluded from the study due to the lack of information (no FNAC cytology or pathology) and the study was conducted on 1033 patients (133 males and 900 females) with a mean age of 45.8± 14.11 years. The pathological report after FNA was used to confirm the diagnosis in neoplasm or suspicious for malignancy cases. In patients with AUS pathology reports, FNA was repeated after the next six months.

Results of patients’ age and gender and the ultrasonography of thyroid lobes

The incidence of malignancy of thyroid nodules was 12.78% and 9.22% in males and females respectively (P-value= 0.19). The incidence of malignancy was 26.67% in 15 patients with less than 18 years and the incidence of malignancy was 9.43% for the rest of the patients over the age of 18 years (P-value= 0.001). The mean age of patients with malignant nodules was lower than of those with benign nodules (40.5 and 46.6 years, respectively, P-value = 0.001). The incidence of malignancy was higher in patients with less than four nodules, and it was significantly reduced in patients with more than four nodules (P-value= 0.001). The mean diameter of studied nodules was 23± 17 mm in the ultrasonography.

The incidence of malignancy based on ultrasonography features of thyroid nodules

Among ultrasonography features of thyroid nodules, the irregular shape (46.7%), unclear margin (47.3%), extension to the capsule (irregular and infiltrative margin) (85%), the marked hypo-echoic nodules (63.8%), micro-calcification (49%), and to have vertical axis (74.0%) were associated with high incidence of malignancy. Table 2 show the possibility of malignancy and malignancy coefficients according to the ultrasonography features of thyroid nodules.

Pathology results

Based on the results of 1033 studied patients, 933 patients (90.32%) had benign nodules and 100 patients (9.68%) had malignant nodules. The benign nodules were 134 cases of cystic lesions, 367 cases of multiple masses of spongy-like solid with an adenomatous goiter appearance, 32 cases of thyroiditis with or without white knight nodules, and 54 cases of follicular benign nodules. The malignant nodules were seen in 81 patients with PTC, 11 patients with FTC, 5 patients with medullary carcinoma, 2 patients with lymphoma, and 1 patient with an anaplastic tumor. 346 patients did not have surgery (224 patients of class-2 TI-RADS, and 122 patients with nodules smaller than 1cm).

Among patients with suspicious or malignant FNAC, 5 cases had benign lesions. Also, 2 patients had benign FNAC which was found malignant lesions after the surgery (it was due to the patient’s request and the large size of nodule). In 13 cases, FNAC could not diagnose nodule types. Finally, 20 cases (approximately 2%) of FNACs could not correctly diagnose nodule types.

Thyroid nodule classification based on conventional methods

In this study, ultrasonography and classification of thyroid nodules in TI-RADS system was performed by an experienced radiologist. The results of TI-RADS classifications based on conventional methods are presented in Table 3.

Thyroid nodule classification based on the proposed new formula

The sum of malignancy coefficients (Wi) for all thyroid nodule features was 45 in the ultrasonography (Stotal=45). If sum of malignancy coefficients for a patient (S) is divided by 45, the probability of malignancy of the thyroid nodule is obtained. By comparing the probability of malignancy (S/Stotal) and the probability of malignancy in each of the TIRADS classification, S-values are obtained. To simplify the use of the proposed new method, S-values for each group of TI-RADS are calculated and presented in Table 4. S-values are obtained as follows: S= 0 for total malignancy coefficients in TI-RADS = 2 group; S = 1 and 2 for TI-RADS= 3; S = 3 to S = 9 for TI-RADS = 4a; S = 10 to S = 16 for TI-RADS = 4b; S = 17 to S = 23 for TI-RADS= 4c, and S = 24 to S = 45 for TI-RADS = 5. Patients can be put into one of TI-RADS groups according to Table 4 by obtaining the S-value for each patient. The results of TI-RADS classifications based on the proposed new formula are presented in Table 3.

Comparison of TI-RADS classification with conventional method and the proposed method

Based on the conventional method, TI-RADS classification results were equal to: group 2: 0.0%, group 3: 0.4%, groups 4a, 4b, 4c: 8.3%, 36.5%, 74.2%, and group 5: 91.9%, respectively (Table 3). The incidence of malignancy in TI-RADS classification according to the new proposed formula was achieved as follows: group 2: 0.0%, group 3: 0.7%, groups 4a, 4b, 4c: 16.7%, 43.4%, 68.5%, and group 5: 95.2%, respectively (Table 3). The comparison of the TI-RADS classification based on the proposed and conventional method (without formula) showed that it was exactly the same in 832 cases (80.54%). The Kappa test (Kappa = 0.73) was used to compare and match results of Table 3 indicating that the results of the proposed method were exactly consistent with results of the conventional method.

DISCUSSION

The Thyroid Imaging Reporting and Data System (TI-RADS) is an international system for the risk stratification of thyroid nodules, and attempts have been made to improve its parameters over the past years. The ultrasonography structure of a nodule is the most important feature by which the malignancy or benign nodules can be predicted; hence, its various forms were completely examined in the present study. In this study, four features were associated with the highest incidence of malignancy among the anatomical features of the studied thyroid nodule in TI-RADS, as the incidence of malignancy was about 85% in patients with infiltrative margin and extension to the capsule; and 74% in nodules with a vertical axis, and about 64% in marked hypo-echo.

Given the echogenicity of nodules, regardless of marked hypo-echo of nodules, other complete or partial hypo-echogenicity cases were 3% to 8% associated with malignancy. Other studies found that hypo-echogenicity and malignancy of nodules were directly related to each other (2, 3, and 12). Recent studies have also found that hypo-echogenicity is associated with malignancy in more than 50-90% of cases, and hypo-echogenicity increases the chance of malignancy diagnosis from 2.9 to 6 times (16-18).

Shapes and margin of nodules are easily identifiable and crucial and according to findings of the present study, irregular-shaped nodules or unclear margins faced about 50% incidence of malignancy. An irregular shape was reported as non-ovoid or tall nodules in some studies, and other studies considered its high association with malignancy (2, 3, 12, 19-25). Similar studies have shown that the incidence of malignancy in the Taller and non-ovoid nodules is about 20-80% higher; and Taller status compared with the wide shape can increase the probability of malignancy from 2 to more than 10 times (12, 16, 17).

For the echogenic dot, its existence was associated with an increase in the incidence of malignancy in the present study despite the fact that the incidence of malignancy in the presence of micro-calcification, macro-calcification, and echogenic foci with comet-tail was about 49%, 8.7%, and 3.3%, respectively. The reason for more calcification in the malignant nodules can be due to the fact that the malignant tumors are more likely to grow and require more blood supply; however, insufficient blood supply to the tumor leads to the localized fibrosis hyperplasia and calcium deposition (27, 28). Some studies have proven the association of micro-calcification and malignancy of nodules, and other studies have reported the correlation of 30-50% between malignant nodules and micro-calcification (3, 12, 20, 29, 30).

The nodule heterogeneity was the anatomical feature of nodules that increased the risk of malignancy in the present study indicating an increase in the incidence of malignancy in heterogeneous and heterogeneous-homogenous states. Based on our findings, a partial heterogeneity in nodule could lead to an increase up to 15% in malignancy. Other studies have consistently found that heterogeneous thyroid nodules had a higher risk of malignancy than non-heterozygous nodules (3, 31, 32).

The higher percent of cystic component is associated with the lower incidence of malignancy. The highest level of malignancy is found in fully solid (16%) and solid masses with few cysts (10%). Simple, colloidal or spongy cystic masses, or predominantly cystic masses or spongy masses are associated with a malignancy risk of less than 1% which is previously mentioned in other studies (3, 33, 35). Despite the fact that malignant changes were observed in 5% of malignant thyroid nodules according to our experience, all of these patients had other malignant findings such as hypo-echogenicity or micro-calcification, so missing malignancy should be considered in patients with mostly cystic lesions without any other suspicious findings.

Considering zero scores for the ill-defined border is questionable in the ACR-TIRADS classification. In addition, despite the fact that Smooth Rim calcification is not considered in favor of malignancy according to our study and some other studies (17), applying score 1 for it in the ACR-TIRADS classification is questionable. In the present study, the extension to thyroid capsules (irregular and infiltrative margin) was identified as a factor associated with a higher incidence of malignancy, whereas in a few previous studies this factor is considered as a malignancy-related factor, and more studies have to be conducted to generalize the results of this factor. In some studies, the extension to capsules has been reported about 35-95% with malignancy (16-26).

In this study, other variables were studied despite the fact that they were unrelated with anatomical characteristics of thyroid nodules, but their significance in relation to malignancy of thyroid nodules led to interesting results. These variables played significant roles in early diagnosis, for instance the role of the patient’s age is significant. The number of patients with thyroid nodules increased with age. However, by dividing the age by two groups below and above 18 years it was found that the incidence of malignancy was higher in the younger age group. The incidence of nodule malignancy was reported 26.67% in patients younger than 18 years. It was also found that the mean age of patients with malignant nodules was also lower than of patients without malignant nodules (40.5 and 46.6 years, respectively). In the Ulisse study, the median age of patients with malignant nodules was lower than of patients with benign nodules (50 versus 58 years) (12). Another study also found that aging was associated with a reduction in the risk of malignancy in thyroid nodules (36).

Patients’ gender was also investigated, so that the number of women with thyroid nodules was 6.7 times more than men, and it was approximately consistent with the global prevalence of thyroid nodules in women (1, 37 and 38). The incidence of malignancy in thyroid nodules was 3% higher in males than in females, but P= 0.19 showed no correlation between the incidence of malignancy and gender. This finding was studied and confirmed in several studies, so that a study with a limited sample size found no relationship between the patient gender and the incidence of malignancy in thyroid nodules (12).

The present study diagnosed and investigated the correlation between the incidence of malignancy and numbers of diagnosed nodules in each patient indicating the lower rate of malignancy in 4 or more nodules than 1 to 3 nodules because when the number of nodules increased by 4, the probability of adenomatous goiter changes and the resulting spongy mass also increases. However, it seems that no specific relationship can be found between the number of nodules and benign or malignant nodules in the presence of 1, 2, and 3 nodules. It was also observed in Ulisse’s study, finding no correlation between nodularity and malignancy (12).

Another hypothesis of the study was the location of the thyroid nodule, and its relation to the incidence of malignancy. Interestingly, in cases where thyroid nodules were located in both thyroid lobes or both lobes and isthmus, the incidence of malignancy of nodules was reduced 5-25% less than of nodules located in a lobe. According to the obtained result, the incidence of malignancy decreased when a nodule was distributed equally and in balance. In the present study, malignant nodules were reported in about one-third of patients with cervical lymphadenopathy, in other words, 34.82% of patients with cancer had adenopathy at the diagnosis, while it was about 6% in patients without lymphadenopathy.

Fortunately, lymphadenopathies from thyroid carcinoma have increased echogenicity, cystic and micro-calcification changes, and sometimes an increase in sizes although it has not led to favorable results in a few cases of differentiation of malignant or non-malignant lymphadenopathies in patients with thyroid cancer using ultrasonography (39, 40). In the present study, the incidence of malignancy in adenopathy of above 15 mm was 66.67%, which could be important for the sonographers.

Despite the fact that Doppler indices do not play roles in the TI-RADS classifications, Doppler test plays an important role in the diagnosis of a fairly common thyroiditis and white knight nodules resulting in thyroid interno and peripheral vascularity. In addition, different blood sampling of malignant nodules in comparison with the surrounding inflammatory tissue in the thyroiditis is a valuable index for distinguishing secondary PTC to thyroiditis from bloody inflammatory tissues. Except for low intra-nodular vascularity in spongy nodules (adenomatous goiter) and sometimes white knight nodules, other benign and malignant thyroid nodules have central vascularity and the internal blood flow of nodules is not associated with the increased risk of malignancy as also reported in Delfim’s study (2). However, these two types of nodules have a special appearance in gray scale images.

In this study, we sought to change weighting factors associated with the incidence of malignancy of thyroid nodules for the uniformity of the TI-RADS classification. Comparison of scoring through the proposed formula with scoring based on an experienced radiologist indicates that the proposed formula is successful in a grouping. Findings of this study were comparable to those reported in other TI-RADS classifications and were consistent with recent guidelines classifications (2, 14, 15 and 41). The proposed formula would be very practical especially for less experienced ultrasonography specialists since it is easy to use with malignancy coefficients and using it in the formula determines TI-RADS scores and can estimate the incidence of malignancy leading to a better management of patients. This can reduce the disagreement between the sonographers and unify reports in the TI-RADS classification and increase the use of this system. Despite the announcement of different criteria, the TI-RADS classification is almost based on experts’ tastes and experience which can result in different responses in the TI-RADS classification and reduce the application of this system (8, 13, and 42).

As the present study was the first research in West Asia and since the validity of the TI-RADS classification is important in any clinical setting, this study is valuable from this perspective (43, 44). In addition, the sample size of the study was proportional to other studies increasing the generalizability of its results (42). The present study had various limitations. One of the limitations was the lack of histopathological confirmation of all nodules, especially the benign nodules in the cytopathology. This limitation was also found in other studies (2, 10, 25 and 30). Exclusion of some patients with neoplasm cytology result, according to the selection of follow-up approach for these patients, was another limitation of this study. On the other hand, there was no need for FNAB according to the guidelines, and cytological confirmation of benign and malignant lesions was not achieved in many nodules especially those less than 1 cm (5). Considering the acceptance of 2% error for the cytological outcome as well as the relatively common cases of non-diagnostic or AUS cases in cytological reports (8, 13), conducting studies are suggested with the aim to compare cost/error of the ultrasonography with cost/error of the FNA. It seems that the usefulness of FNA with 2% diagnostic error can be questioned in the TIRADS III nodules with 0.4% incidence of malignancy.

In conclusion, according to the proposed new formula for TI-RADS, there are four coefficients of 7, 3, 1 and 0 for incidence of malignancy of each one of ultrasound findings. Based on the conventional method, TI-RADS classification results were equal to: group 2: 0.0%, group 3: 0.4%, groups 4a, 4b, 4c: 8.3%, 36.5%, 74.2%, and group 5: 91.9%, respectively. The incidence of malignancy in TI-RADS classification according to the new proposed formula was achieved as follows: group 2: 0.0%, group 3: 0.7%, groups 4a, 4b, 4c: 16.7%, 43.4%, 68.5%, and group 5: 95.2%, respectively. It seems that the application of these coefficients and formula for evaluating the incidence of malignancy of thyroid nodules can be a useful step towards simplifying and unifying the TI-RADS classification.

Conflict of interest

The authors declare that they have no conflict of interest.

Acknowledgment

The present study was conducted based on findings of a Specialty thesis on Radiology by Dr. Samira Rahemi Karizaki (Code: 951257) at Mashhad University of Medical Sciences. Authors of this paper thank the staff of Omid Hospital in Mashhad for their help and cooperation. This study was conducted by the financial and spiritual support of the Research Department of Mashhad University of Medical Sciences according to an approved proposal by Code 951257.