| Thyroid nodules are the masses of thyroid with abnormal hardness and structure. It's a common disease. The incidence rate increases with age, and is higher in females. The majority of thyroid nodules are benign, and about 2.7%-17% of them being malignant. Imaging examinations of thyroid are used to detect and make an initial differentiation of the lesions. The results are the basis of further treatment, in order to reduce unnecessary surgery and post-surgery complications. However, thyroid carcinoma is a disease with hidden onset. The clinical, imaging and cytology feature of benign and malignant lesions sometimes overlap with each other. Therefore, it's difficult for differentiation before surgery.Traditionally, the primary modalities for disease detection are CT/MRI, radionuclide image, PET, Ultrasound-guided fine-needle aspiration (FNA), and PCR. High-resolution ultrasound is the preferred modality for evaluation of the thyroid gland. It is noninvasive and economical. No contrast media is needed. The internal architecture can be observed clearly. Even the tiny nodules with 1 to 2 mm diameter can be detected. The sensitivity in nodular detection is 100% for ultrasound, which is much higher than CT/MRI and radionuclide image. Conventional ultrasound can assess the characters of thyroid nodules, such as number, diameter, nature and relationship with surrounding tissue. It can detect nodules that can't be felt by palpation. It is beneficial to detect nodules earlier. But it is not easy to discriminate benign and malignancy nodules in thyroid because the characters of thyroid nodules are very complex according to their histology.Ultrasound elastography(UE) is a new and promising imaging technique. It extended the view of differentiated diagnosis of thyroid nodules. It provides another characteristic of the nodules--stiffness. UE has been wildly used in clinic and become an eyespot of ultrasound.UE have been used successfully for many fields, such as breast, thyroid, prostate, abdominal organs and muscle tissue. There are different diagnostic criteria for elastography. Most authors approve Itoh A's 5-point scoring system. The criterion has already been successfully used in differentiation of breast lesions.UE is expected as means for providing novel diagnostic for tissue disease such as cancer since the tissue hardness is closely related to its pathological state. Most malignant nodules of thyroid are harder than benign ones. Therefore, UE can provide this information. Thyroid is a superficial organ, which made it easy to perform UE. Domestic and foreign scholars'studies have shown that UE is a very useful supplement to conventional ultrasound imaging. It can provide additional information in thyroid nodules differential diagnosis. However, there is no set criterion for thyroid nodular elastography. More clinical studies are needed for further research including formulate a diagnosis system.The Logistic regression analysis has been applied in the research of breast ultrasound. But so far as I know, there is no such research concerning about two dimensional gray scale ultrasonography, color Doppler flow imaging, and ultrasonic elastography of thyroid nodules. In this study, thyroid nodules were studied with pathology as golden standard. Part one, the UE images were analyzed according to the criteria proposed by Yu Qing in Zhongshan hospital Fudan university. The elastograms of thyroid nodules were retrospectively analyzed and receiver operating characteristic (ROC) curve was used to assess the role of ultrasound elastography(UE) in differentiating benign and malignant lesions of the thyroid. A cut-off value was also chosen.Part two, the nodules were classified according to the criteria recommended by Hae Kyung and Stacul et.al. Each nodule was scored based on its ultrasonic features and elastogram scores. A Logistic model was obtained, using ultrasonic features and elastogram scores as independent variables and pathologic findings as dependent variable. The purposes were:select out the valuable features in the differential diagnosis of thyroid nodules; construct binary Logistic regression model to analyze ultrasonographic features of thyroid solitary nodules, and improve the diagnosis accuracy of thyroid nodules.1 To evaluate the role of ultrasound elastography (UE) in differentiating benign and malignant lesions in the thyroid and figure out the cut-off value.2 Evaluate the value of ultrasonic features and elastography in the differential diagnosis of thyroid nodules.From June 2009 to December 2009,244 patients with 291 nodules were examined with 2D-US, PDI and UE pre-operation. All the lesions were confirmed by post-operation pathology. Male 61, female 183, age from 7-79 years,mean age (43.38-0.83) years old. The diameter of the nodules ranged from 0.3-3.2 cm,144 noduler in right lobe,141 in left lobe and 6 in isthmus. From January 2010 to February 2010,50 patients with 63 nodules were selected to test the Logistic regression formation. All the lesions were confirmed by post-operation pathology.Hitachi Vision 900 ultrasound system with ultrasonic elastography was used in this study. The frequency of the linear array probe is 6.0-13.0 MHz.2D-US was performed on the thyroid. UE was performed on the detected nodules with 2 to 3 times normal tissues surrounding it. The elastograms were scored by two ultrasonic doctors in blind separately. When the result was contradicted, more doctors were involved to reach an agreement.Part one, the UE images of thyroid nodules were classified into 0-4 patterns(table 1). The elastograms of thyroid nodules were retrospectively analyzed and receiver operating characteristic (ROC) curve was used to assess the role of ultrasound elastography(UE) in differentiating benign and malignant lesions of the thyroid and the cut-off value was chosen.Part two, the nodules were classified according to the criteria recommended by Hae Kyung and Stacul et.al. Each nodule was scored based on its ultrasonic features and elastogram scores. A Logistic model was obtained with its ultrasonic features and elastogram scores as independent variables and pathologic findings as dependent variables.Fifty patients with 63 nodules were selected to test the Logistic regression formation. All the lesions were confirmed by post-operation pathology.Surgical pathology was used as golden diagnostic standard. Statistics analysis: A Logistic model was obtained with ultrasonographic features as independent variables and pathologic findings as dependent variables. The ultrasonographic features of lesions included the shape, boundary, internal echo, posterior echo, aspect ratio and micro-calcification of lesions, the two-dimensional Adler grading of the blood flow, and UE scores. The Waldχ2 test was used to evaluate the estimated value of regression parameters. The likelihood ratio test was used to evaluate the fitting situation of the whole model. The data were analyzed by SPSS 13.0. P value less than 0.05 was considered to be statistically significant.There were 66 nodules in malignant group, including:62 papillary carcinoma, 1 medullar carcinoma,1 follicular carcinoma, and 2 lymphoma.There were 225 nodules in benign group, including:45 thyroid adenoma,175 nodular goiter, and others 5(4 subacute thyroiditis and 1 Hashimoto's thyroiditis). group01234totalZP Adenoma23363145-1.0440.297 goiter114193228175 total1344129259220In benign group,84.9%(191/225)nodules'elastograms scores are equal or less than 2. And in malignant group,89.4%(59/66) nodules'elastograms scores are equal or more than 2. There were significant differences in the elasto graphic scores between benign and malignant nodules(Z=-10.093, P=0.000). Malignant lesions had a higher score than benign ones.The difference of elastographic scores between thyroid adenoma group and nodular goiter group was not significantly (Z=-1.044, P=0.297).The area under ROC curve was 0.883. If the cut-off value of the elastographic scores was equal or more than 3, the sensitivity, specificity and accuracy was 89.4 %,84.5%,85.9% respectively.3. Part two According to the output results of SPSS 13.0, there were four independent variables selected on the final step of Logistic regression analysis, including:the number of nodules, calcification of nodules, echo, the two-dimensional Adler grading of the blood flow and UE scores. The model was:logit (P)=-6.731-0.932X1+0.8828X6+1.392X7+1.388X10The likelihood ratio test was used to evaluate the fitting situation of the whole model. It was statistically significant (χ2=152.507, P=0.000).The model was used to predict the 63 thyroid nodules. The model could distinguish the two kinds of nodules. When the regression value P was more than 0.5, the prediction was malignant tumor; When the regression value P was less than or equal to 0.5, the prediction was benign tumor. The correct rate was 57/63=90.4%. the sensitivity and specificity was 78.6%,93.4%. 1. UE provides more information in differentiating benign and malignant nodules of thyroid. It's helpful in clinical practice.2. Four of ten independent variables were selected by Logistic regression analysis. Conventional ultrasound is the base of the diagnosis of the thyroid nodule. And UE play a more important role in differentiating benign and malignant nodules of thyroid. If both methods were integrated, the accuracy will be improved. |
PDF Full Text Request