Value Of Magnetic Resonance Imaging And Diffusionweighted Images For Differentiating Benign And Malignant Thyroid Nodules

Part I Correlation of MRI Findings and Histopathological Features in Thyroid NodulesObjective To investigate the value of high field magnetic resonance imaging (MRI) for differenting benign from malignant thyroid nodules by comparing their MRI findings with histopathological features.Method Eighty one cases with thyroid nodules (TN) were included in the study. There were twenty six male cases and fifty five female cases who were between14to78years, had a disease course ranging from several days to years. Preoperative MRI scan were observed with the lesion’s location, size, shape, covering, edges, cystic changes, surrounding tissue involvement, enhancing features, lymph node metastasis, and distant metastasis. Then, preoperative MRI characteristics of the nodules and the corresponding pathology were analyzed and discussed.Results Forty six caseswith benign nodules, twenty-ninecaseswith malignant nodules, and six cases coexisted withbenign and malignant noudles were inclued. Patients with benign nodules included forty-one cases of nodular goiter, and five cases of thyroid adenoma. There were a total of nineteen cases of papillary carcinoma (including two cases of papillary thyroid micro-carcinoma (PTMC), five cases of papillary carcinoma coexisted with PTMC),four cases of papillary carcinoma (inculding one PTMC) with Hashimoto’s Thyroiditis, and one case of medullary carcinoma coexisted with papillary carcinoma. Other malignant nodules included two cases of lymphoma and three cases of metastatic cancer from the primary lesion located in the esophagus. Benign coexistedmalignant noudles were considered malignant in statistical analysis, which included five cases of papillary carcinoma (including three PTMC) coexisted with nodular goiter and one case of follicular carcinoma coexisted with nodular goiter.Among twenty-six male patients with thyroid nodules, eleven were benign and fifteen were malignant; and among fifty-five female patients with thyroid nodules, thirty-two were benign and twenty-three were malignant. The gender of patients with benign or malignant thyroid nodules were insignificant differences (P>0.05).67.4%(31/46) of patients with benign nodules were multiple, while57.1%(20/35) of patient with malignant nodules were solitary. Multiple nodules were significantly common in benign thyroid nodules (P<0.05).Eighty-one patients had a total of one hundred thirty-four nodules including eighty-seven benign nodules and forty-seven malignant nodules. MRI findings of the nodules had a significant difference in size, shape, covering, cystic changes, enhancing characteristics of solid component, and lymph node enlargement (P<0.01). bleeding/calcifications in benign or malignant thyroid nodules were insignificant differences (P>0.05). Nodules with irregular shape, ill-defined edges, less fluid, and heterogeneous enhancement of solid content were important indications for the diagnosis of malignant nodules. An incomplete "film-like" low signal surrounding intensity was a characteristic manifestation of malignant nodules. Neck lymph node metastasis and invasion around the thyroid gland and tissue were positive signs of thyroid cancer.There are eleven patients had or coexisted with PTMC in thirty-five malignant cases. Micro-cancers had multifocal pathological appearance with diameters ranging from0.03to0.7cm. MRI scan and contrast enhanced MRI showed no significant lesions in eight patients having a misdiagnosis rate of72.7%(8/11); and only three patients were diagnosed correctly. Conclusion MRI examination is useful in differentiating benign andmalignant thyroid nodules, owing to proving detailed morphological information and accurately determining the extent of tumor invasion and lymph node metastasis. However, difficulties still existed in the diagnosis of nodules with overlapping signs, especially in multiple nodules (benign coexisted with malignant nodules). Conventional MRI examination had difficulty in the detecting of PTMC without lymph node metastasis, too. Part II Diffusion-Weighted Images Differentiate Benign from Malignant Thyroid NodulesObjective To investigate the feasibility of diffusion weighted imaging (DWI) examination in thyroid; To disscuss the value of apparent diffusion coefficient (ADC) for differentiating benign and malignant thyroid nodules; and to determine the best b-value for DWI examination for differentiation of benign and malignant thyroid nodules.Method forty-six cases (twenty-one benign and twenty-five malignant) with thyroid nodules were include in the study. They were examined with conventional MRI and DWI with b-value of100s/mm2,300s/mm2, and500s/mm2. All mean ADC values of benign nodules and malignant nodules were compared. The receiver operating curve (ROC) method selected the best b-value for the differentiation of benign and malignant nodules. The diagnostic tests were then evaluated (sensitivity, specificity). All patients had detailed surgical procedures recordings, and postoperative biopsy results.Results1. The mean ADC values of benign nodules at b-factor100s/mm2,300s/mm2, and500s/mm2was significantly difficult (F=62.993, p<0.01; t=7.107,8.999,5.710, p<0.01). The mean ADC value decreased as the b-value increased.2. The mean ADC values of benign nodules at b-factor100s/mm2,300s/mm2, and500s/mm2was significantly difficult (F=49.956, p<0.01; t=6.208,9.170,3.808, p<0.01). The mean ADC value decreased along with the b value increasing.3. The mean ADC values for benign and malignant nodules were (3.4326±0.4200) x10-3mm2/s and (2.4693±0.7206) x10-3mm2/s for b-100factor,(2.9873±0.3050) x10-3mm2/s and (1.7879±0.4719) x10-3mm2/s for b-300,(2.7363±0.3698) x10-3mm2/s and (1.444±0.3194) x10-3mm2/s for b-500, respectively.The mean ADC values of malignant nodules were lower than benign. There were significant differences in ADC valuesbetween benign and malignant nodules at all b-factors.(t=5.960,11.074, and13.959, respectively.p<0.01).4. Acutoff value for malignant nodules of2.6052x10x10-3mm2/s at b-300factor yielded sensitivity and specificity of100%and72.7%, respectively.Conclusion:1. There was a significant difference between mean ADC values in benign and malignantnodules. Acutoff value for malignant nodules of2.6052x10x10-3mm2/s at b-300factor yielded sensitivity and specificity of100%and72.7%, respectively.The present study revealed that ADC measurements could potentially quantitativelydifferentiate between benign and malignant thyroid nodules..2. Analysis of the DWI image features and measurement of the ADC value could make up for the deficiencies of conventional MRI examination, especially when there were overlapping signs which were difficult to define by conventional MRI. We propose that diffusion-weighted imaging evaluation should be used for the assessment of thyroidnodules in addition to conventional MRI.3. PTMC is more sensitive in DWI than conventional MRI. Part III Logistic Regression Analysis of Combining Conventional and Functional MRI Features of Thyroid NodulesObjective To create a assessment of thyroid nodules based on conventional andfunctional MRI using Logistic regression that can screen out the specific features for differentiating malignant and benign nodules.Methods fifty-eight noudles (33benign and25malignant)confirmed by pathology were enrolled in the study. Imaging data included conventional MRI sequences (T1WI, T2WI and enhanced images) and functional imaging sequences (DWI images). The original images were processed in a workstation to measure apparent diffusion coefficient (ADC) of nodules. Combining the first and second part of the study results, the following features were evaluated:nodule shape, covering, enhancement of solid component, cystic changes, hemorrhage/calcification, and ADC. The analysis of variance showed statistical difference index into two categories of multivariable Logistic regression analysis, Logistic regression model was established to screen out significant risk factors for malignant nodules, and a ROC curve was made to assess the diagnostic value of the logistic regression model.Results There were three MRI feature as variables (covering, cystic changes, and ADC values) were chosen for the equation by Logistic regression analysis. Corresponding models:Logit(p)=-25.001+2.541x cystic+3.417x covering+22.561×ADC. When P>0.5were supposedfor malignant nodules, P<0.5were supposedfor benign nodules, the diagnosticaccuracywas94.8%. The area under ROC curve was0.955. Conclusion The cystic degeneration, covering, and ADC value play an important role for differentiating benign and malignant thyroid nodules. The Logistic Regression model which was established with routine MRI and DWI, wasvaluable for differentiation of benign and malignant thyroid nodules.