Different Mathematical Models Of Diffusion Weighted Imaging In Differentiating Benign From Malignant Thyroid Nodules

Objective To assess the value of Different mathematical models of diffusion weighted imaging in differentiating benign from malignant thyroid nodules.Methods 85 thyroid nodules from 52 patients underwent the reduce field-of-view multi b-value diffusion weighted imaging. Thyroid nodules were divided into malignant and benign groups based on the pathological findings.Mono-exponential model was used to calculate Standard ADC. Bi-exponential model was used to calculate Slow ADC, Fast ADC and Fraction of fast ADC(f).Stretched-exponential model was used to calculate Distributed diffusion coefficient(DDC) and the heterogeneity of intravoxel diffusion(?). These parameters of both groups were compared and receiver operating characteristic(ROC) curves were drawn to determine the diagnostic thresholds. The correlations between each two of Standard ADC, Slow ADC and DDC in both groups were analyzed. There was a significant difference when P<0.05.Results The Standard ADC, Slow ADC, DDC and ? were higher in benign group than malignant group(P<0.05). Fast ADC and f had no significant difference between two group. DDC for differentiating benign group from malignant group had a largest area under the ROC curve (AUC)(0.985),followed by Standard ADC(0.961), Slow ADC(0.930), a(0.717), Fast ADC(0.555), f(0.538). Using DDC=1.19×10-3mm2/s as the threshold of identifying two group, the sensitivity and specificity were 100.00% and 90.77%.There was more significant correlation between each two of Standard ADC,Slow ADC and DDC in benign group than malignant group.Conclusion Standard ADC, Slow ADC, DDC and a were helpful in identifying benign thyroid nodules from the malignant ones. DDC had the best diagnostic efficiency among the parameters. Compared with mono-exponential model, bi-exponential and stretched-exponential model could represent signal attenuation pattern of thyroid nodules better.