Diagnostic accuracy of B-mode ultrasound, ultrasound elastography and diffusion weighted MRI in differentiation of thyroid nodules (prospective study)

The incidence of the thyroid nodules and its detection is increasing rapidly. The most precise method for diagnosis of the nodules of the thyroid is FNAC. But, about 10–20% of specimens of FNAC are indeterminate and non-diagnostic. Therefore, there is a demand for another diagnostic method for evaluating thyroid nodules. Thyroid ultrasound elastography may improve the ability to differentiate malignant from benign thyroid nodules. Few articles were published about the results of DW MRI in thyroid nodules, with its results confirmed that malignant nodules have lower mean ADC values than benign nodules. This study aims to investigate and compare the accuracy of B-mode ultrasound, ultrasound elastography and diffusion-weighted MRI in characterization of the nodules of the thyroid. The study included 56 patients with thyroid nodules (36 benign and 20 malignant). Thyroid ultrasound, ultrasound elastography and DWI were done for all patients. Ultrasound-guided FNA Cytological examination (as the gold standard) was done for 48 patients and surgical histopathology was done to 8 patients with non-diagnostic FNAC. The results showed: TIRADS score had sensitivity 90%, specificity 77.8% and accuracy of 82.14%. The elastography score had sensitivity 80%, specificity 88.9% and accuracy 85.7%. The use of the strain ratio had 80% sensitivity, 94.4% specificity and 89.3% accuracy. DWI and ADC value had 100% sensitivity and 94.4% specificity and the accuracy was 96.4% for differentiating malignant from benign thyroid nodules. Multi-parametric analysis by TIRADS and ADC had 100% accuracy. Ultrasound elastography add valuable data over ultrasound TIRADS. But, diffusion weighted MRI and ADC value has more accuracy in differentiating malignant from benign thyroid nodules. The best performance was achieved by the combination of ACR-TIRADS and ADC value.

Ultrasound is the most common imaging modality used to diagnose lesions of the thyroid; however, still there are no dependable criteria to discriminate malignant from benign lesions. Also, in large and multiple nodules, it is not easy to assess the malignant possibility of the nodule [6] (Fig. 4).
It is believed that the most precise method for assessment of the nodules of the thyroid is fine needle aspiration cytology (FNAC). But, about 10-20% of specimens of FNAC are indeterminate and may be non-diagnostic [7]. In addition, the overlap of morphological signs between malignant and benign lesions may lead to difficulty in interpretation of cytology. Therefore, there is a need for another diagnostic method for evaluating thyroid nodules [8] (Fig. 5).
Thyroid ultrasound elastography (USE) is a noninvasive method of evaluating thyroid nodules that can provide complementary information to B-mode US and FNAC. Thyroid USE may improve the differentiation of malignant from benign thyroid nodules leading to decrease the number of needed fine needle aspiration biopsy (FNAB) [9].
There are available many studies discussing the use of ultrasound elastography in diagnosis of the nodules of the thyroid. However, the sensitivity and specificity of this method are sharply variable in the medical literature [10].
Diffusion weighted (DW)-MRI has been proved to have a role in assessment of tumors of the neck, with the quantitative parameter (ADC) value, could be used for differentiation of benign and malignant lesions [6].
Few articles were published about the results of DW MRI in thyroid nodules, with its results confirmed that malignant nodules have lower mean ADC values than benign nodules [11].

Aim of work
The aim of this study was to assess the value, compare the diagnostic accuracy and investigate the multi-parametric analysis of B-mode ultrasound, ultrasound elastography and diffusion-weighted MRI of the thyroid gland in characterization of the nodules of the thyroid (whether it is benign or malignant).  The age of the patients ranged from 17 to 65 years. The major concern of the referring physician was to distinguish malignant from benign nodules of the thyroid. The inclusion criteria were the presence of solitary or multiple nodules in the thyroid gland in patients at different age groups, the nodules were either solid or mixed (containing both solid and cystic parts). In case of patients having multiple nodules, the nodule having suspicious ultrasound features (having TI-RADS score 4 or 5) was selected for further analysis. In addition, patients should have normal bleeding profile. Patients with contraindications of the FNAC procedure (e.g. thrombocytopenia or bleeding disorders), patients with complete shell-calcified nodules that may cause color mapping artifacts or pure cystic nodules without solid parts and patients with recurrent thyroid masses, history of operative procedure, chomo or radiotherapy on the thyroid gland, patients with contraindication to MRI or with bad general condition and patients had nodules with indeterminate or incomplete diagnosis were excluded.
The study had approved by our local committee of research and ethics and all participants signed informed consent before being included.
All patients were subjected to: 1. Thyroid ultrasound and ultrasound elastography using the machine (Toshiba Aplio-500 with linear probe 10-13 MHz). 2. Magnetic resonance imaging examination (mainly DWI) using Philips Ingenia 1.5 Tesla (T) MRI scanner 3. Ultrasound-guided FNA Cytological examination (as the gold standard) for 48 patients. Surgical histopathology was done to 8 patients with non-diagnostic FNAC to obtain final diagnosis.

Imaging techniques Thyroid ultrasound and elastography
The ultrasound examinations included two steps done during the same examination. First step: conventional ultrasound was carried out to define the nodule. Second step: ultrasound elastography with the same transducer using special software for sono-elastography.

Technique
The patient was lying supine with a pillow put under the neck to make his neck hyper-extended and his chin is elevated. Using adequate amount ultrasonic gel, the linear probe was applied to the front of the neck.
First, in B-mode US: thyroid gland nodules were localized and its Thyroid Imaging Reporting and Data System (TI-RADS) categorizations were identified according to ACR-TIRADS, 2017 Tessler et al. [12].
Second, in strain elastography: a region of interest (ROI) was applied, light pressure was done on the neck by the probe and a color box that includes the target nodule was highlighted by the examiner. Repeated compression followed by decompression was done in a vertical manner (perpendicular to the lesion plane). There is an available scale on the elastography machine to ensure that the applied compression was adequate. The strain ratio of the nodule in relation to remaining normal thyroid tissue or to sternomastoid muscle (in case of large nodule with no available normal thyroid parenchyma) was measured and registered.
The available elastogram was presented on the grayscale ultrasound image; in a color scale with the red color represents the softest tissue with great strain and the blue color represents the hardest tissue with no strain. The obtained elastographic images were divided and classified using the 5-points score by Rago's criteria (score1: low stiffness in the whole nodule; the whole nodule has even green color, and also the adjacent thyroid tissue. Score 2: low stiffness in a large part of the nodule; green color is almost completely present in the nodule, with only few spots of blue color. Score 3: most of the nodule has high stiffness; the dominant color of the nodule is blue color with only few spots of green color. Score 4: the whole nodule has high stiffness; the whole nodule has even blue color. Score 5: the whole nodule and adjacent thyroid tissue has high stiffness; blue color presents in the whole nodule and adjacent tissue Rago and Vitti [13].
The thyroid nodules were considered to be suspicious for malignancy, if it had Rago scores of 4 and 5.

The strain index (ratio)
The meaning of strain is the availability of the tissue for bending and deformation; so, soft tissue is expected to have more strain than hard tissue. The strain ratio is valuable in measuring the average stiffness of a lesion in comparison with its adjacent tissues.
The strain index (ratio) of the nodule of the thyroid compared to the normal thyroid tissue or that of the sternomastoid muscle (when no available surrounding normal thyroid tissue in the field for comparison) was measured and calculated. Three strain ratio measurements were taken and the average ratio was used as the representative strain ratio of the nodule. Then the calculated average strain ratio for all examined nodules was correlated and compared to its cytological result, to confirm if it is benign or malignant.

MRI protocol
The MR examination was done using the MRI scanner (Philips Ingenia 1.5 T) for all patients with the same scanning parameters.
Patients were positioned supine, head first, with hips and knees extended, using a circular polarization array head and neck coil.

The interpretation of the images
The MR images of all enrolled cases were presented on the MRI work station of Philips Ingenia 1.5 T MRI scanner. The ADC value for each of the thyroid nodules was measured in the ADC-map image, by drawing a suitable region of interest (ROI). Care should be taken to include the most hypo intense area inside the nodule and to exclude any cystic parts or areas causing artifacts inside the nodule. The average ADC value was calculated after taking three measurements and considered to represent the mean ADC value of the nodule. Then the calculated average ADC value for all examined nodules was correlated and compared to its cytological result, to confirm if it is benign or malignant. All the ultrasound examinations were performed in succession by 2 independent radiologists (M. F. S. and M.A.L.) and all MRIs (mainly DWI) were done in attendance of both. The first radiologist had about 5 years and the second had more than 20 years' experiences in ultrasonic and MRI scanning. In case of controversy, the results were recorded by consensus of both and the aid of the third radiologist (M M EL) with experience more than 30 years in ultrasound and more than 25 years in MRI examinations.

The ultrasound-guided FNAB
The patients were referred to the interventional radiology unit to undergo ultrasound-guided FNAB from the thyroid nodules, provided that there are acceptable bleeding profile tests (INR ˂ 1.4 and platelet count ˃ 50.000/mL). The procedure was done by localization of the suspicious nodules and aspiration by a 20-22G needle under US guidance. The aspirated material was obtained and properly handled in the presence of a cyto-pathologist.
Surgical histopathology was done to the 8 patients with non-diagnostic FNAC to obtain final diagnosis.

Statistical analysis and data interpretation
Data were analyzed using IBM SPSS software, Version 22.0. The qualitative data were presented by number and percent. The quantitative data were presented by median (minimum and maximum) and mean, standard deviation. The threshold of significance was fixed at the (0.05) level.

Data analysis Qualitative data
• Chi-Square test: to compare 2 or more groups • Monte Carlo test as correction for Chi-Square test when more than 25% of cells have count less than 5 in tables (> 2 × 2). • Fischer Exact test was used as correction for Chi-Square test when more than 25% of cells have count less than 5 in 2 × 2tables.

Quantitative data between groups Parametric tests
• Student t-test: to compare 2 groups • One Way ANOVA test: to compare more than 2 groups

Spearman's correlation
The Spearman's correlation: to find out the strength and direction of a linear relationship between two variables.

Diagnostic accuracy
The validity of the test is evaluated using Receiver Operating Characteristic (ROC) curve analysis, and then Sensitivity, Specificity, PPV, NPV and Accuracy were measured.

Results
There were 70 different thyroid nodules in 56 patients (with 10 patients had 2 or more nodules). In patients with multiple nodules, the nodule having TI-RADS score 4 or 5 was chosen for further analysis. So, the final number was 56 patients with 56 nodules, including 20 males and 36 females; with age ranging between (17 and 65 years) and the mean age was a 41.36 ± 14.23 years. The fine needle aspiration was performed for the all 56 patients; it was diagnostic in 48 cases (34 benign and 14 malignant). While 8 cases (2 benign and 6 malignant) needed surgical histopathology for final diagnosis. Finally 36 nodules were benign and 20 were malignant. Table 1 shows that in TIRADS 1 and 2: all nodules (n = 12) were benign. TIRADS 3 score: out of (18) nodules, 16 were benign and 2 were malignant nodules. TIRADS 4 score: out of (18) nodules, 8 were benign and 10 malignant nodules. All TIRADS 5 nodules (8 nodules) were malignant. Table 1 shows that in elastography score 1: all nodules (n = 18) were benign. In elastography score 2 (n = 18): 14 were benign nodules and 4 malignant nodules. In elastography score 3 (n = 10): 4 benign and 6 malignant nodules. All nodules with elastography score 4 (n = 8) and 5 (n = 2) were malignant.
Regarding the elastography score: Nodules had elastography score 1 and 2 (n = 36); 32 of them were benign and 4 were malignant. Nodules with score 3, 4 and 5 (n = 20); 4 of them were benign and 16 were malignant. Sensitivity and specificity were 80% and 88.9% respectively. The accuracy was 85.7% (Table 2). Table 3 shows that strain ratio was significantly lower in benign lesions than in malignant ones (P-value < 0.001). For benign nodules, the strain ratio had a mean 1.2 ± 0.54, median 1.2 and a range (0.27-2.5). For malignant nodules, the strain ratio had a mean 3.79 ± 2.4, median 3.75 and a range (0.7-8.7).
As regard the ADC value: A cut-off value = 1.35 had sensitivity of 100%, specificity 94.4% and accuracy was 96.4% for differentiating benign and malignant nodules (Table 4).
Multi-parametric analysis: leading to increase the diagnostic performance as: Combination of strain ratio to ACR TIRADS: there is increase in sensitivity = 90%, specificity = 94.4% and accuracy = 90% (Table 5).
Combination of TIRADS to ADC: leads to the best performance with sensitivity, specificity and accuracy = 100% (Table 5).

Discussion
To the best of our knowledge, this study is considered the first to use ultrasound ACR-TIRADS categorization, ultrasound elastography and MRI diffusion weighted imaging in a single work to examine the thyroid gland for characterization of malignant nodules, to compare the  diagnostic accuracy of each and also to do multi-parametric analysis of these diagnostic methods in differentiation of malignant and benign thyroid nodules. In the present study; fine needle aspiration (FNA) cytology results were diagnostic in 48/56 (85.7%) of cases, while 8 nodules needed surgical histopathology for final diagnosis. This is also shown in Azab et al. [14] study; where FNA was conclusive in 32/35 (91.4%) of cases and 3 cases required surgical histopathology for final diagnosis.
In the present study; the malignancy risk for TR1, TR2 and TR3 were 0.0%, 0.0% and 11% respectively. These results can cope with the study done by Jabar et al. [19] as malignancy risk for TR1; TR2 and TR3 were 0.0%, 0.0% and 6.9% respectively. Ruan et al. [20] found that malignancy risk for TR1, TR2 and TR3 nodules was 0.0%, 2.1% and 3.1% respectively. While Mohanty et al. [21] reported that none of the nodules with TIRADS score from TR1 to TR3 were malignant.
comparing to the study of Ruan et al. [20] in which the risk of malignancy for TR4 was (40.4%) and for TR5 nodules was (90.6%) and the study of Jabar et al. [19] in which the malignancy risk for TR4 was (29.2%) and for TR5 nodules was (80%).
In this study, the 5-points Rago criteria were used for detection of elastography score. The malignant nodules had higher elasticity scores than benign nodules (P value < 0.001) with 80% sensitivity, 88.9% specificity and 85.7% accuracy. Zhang et al. [22] reported that thyroid elastography score in detection of thyroid malignancy had lower sensitivity (73.0%) and nearly equal specificity (87.8%). Kyriakidou et al. [23] reported higher sensitivity (92%) and nearly equal specificity (90%) for the differentiation of nodules of the thyroid.
The results of this study found that using the cutoff strain ratio of 1.85 for characterization of malignant nodules of the thyroid had area under the curve (AUC) of 0.850 with sensitivity 80.0%, specificity 94.4% and diagnostic accuracy 89.3%. (P value = 0.001).
Idrees et al. [24] used the cutoff strain ratio (2.57) to distinguish between malignant and benign thyroid, it had sensitivity 90.0%, specificity 90.0%, and diagnostic accuracy 90.0%. Tian et al. [25] concluded that using the cutoff strain ratio of 2.52 for differentiation of malignant and benign nodules of the thyroid, had area under the curve (AUC) of 0.861 with 85.7% sensitivity, 90.5% specificity, 85.7% and 88.6% accuracy. Also, Cantisani et al. [26] found that using the cutoff strain ratio of 2.05 to discriminate between benign and malignant nodules of the thyroid, had area under the curve of  The present study showed that the signal intensity in diffusion weighted images was not useful for differentiation of thyroid nodules (P = 0.472). Similarly, Aghaghazvini et al. [27] reported that the signal intensity of the nodule and its border were not valuable for differentiation of thyroid lesions (P-value > 0.05).
The present study showed that malignant lesions had lower mean ADC value than benign ones (P-value < 0.001). Using the ADC value 1.35 × 10 −3 mm 2 /s as a cutoff value for differentiation of malignant and benign nodules of the thyroid had sensitivity 100%, specificity 94.4% and accuracy of 96.4%.
The study done by Pei et al.
[4] concluded that US elastography may be a valuable tool when added to the US TIRADS in differentiation of benign and malignant thyroid nodules, as the combination of US elastography and TIRADS increased the diagnostic performance than elastography or TI-RADS alone. This matches with the results of the current study as strain ratio had more accuracy than ACR-TIRADS (89.3% vs. 82.14% respectively). Adding the strain ratio to the TIRADS leads to increase in sensitivity = 90% and accuracy 90%.
A comparative study between diffusion weighted imaging (ADC) and ultrasound TIRADS was done by Kong et al. [30], they found that the ADC had 92.2specificity and 87.6% accuracy. Ultrasound TIRADS had specificity 80.4 and accuracy of 86.9%. This is also in accordance with the results of this study as ADC had more specificity (94.4%) and accuracy (96.4%) than either ACR-TIRADS (specificity was 77.8% and accuracy was 82.14%) or the strain ratio (specificity 94.4% and accuracy 89.3%.). The reason for the high percent in our study may be due to the small number of patients.
Sasaki et al.
[31] used a special combination of MR time intensity curve and ADC analysis, their results showed sensitivity of 100%, specificity 71% and accuracy of 91% for differentiating benign from malignant thyroid nodules.
Wang et al.
[32] studied the use of multiple MRI parameters for characterization of thyroid malignancy. ADC was the superior in differentiation of benign and malignant nodules, with 90% sensitivity and 91% specificity. When the results of ADC were combined with features associated with thyroid malignancy (such as irregular shape, ring sign and cystic degeneration), the outcome is improved with the sensitivity is increased to 97%, and the specificity was 95%.
The best performance was achieved in the current study by the multi-parametric analysis by adding ACR TIRADS to the diffusion ADC with sensitivity, specificity and accuracy = 100% in differentiation of benign and malignant thyroid nodules. Also, this study was done in a prospective manner in comparison with most of other studies which were retrospective.

Limitation and recommendation
Small number of patients.
Further large scale study is recommended for confirming the results.
In this study, strain elastography was used. Future studies using the shear wave elastography is needed.

Conclusion
Ultrasound elastography and measuring the strain ratio of thyroid nodules add valuable data over ACR-TIRADS categorization. It is recommended as adjunctive tool. However, the diffusion weighted MRI and ADC value had more accuracy in differentiation of malignant and benign thyroid nodules. Multi-parametric analysis leads to increase the diagnostic ability with the best performance achieved by the combination of ACR-TIRADS to diffusion weighted MRI and ADC value.