Introduction

The prevalence of papillary thyroid cancer (PTC), as a major subtype of well-differentiated thyroid cancer, has been steadily rising in recent years ¹. Although PTC has a slow growth and low mortality, some aggressive treatments such as extrathyroidal extension (ETE) are associated with poor outcomes ². Due to the non-invasiveness, superior spatial resolution and good soft tissue contrast, magnetic resonance imaging (MRI) is increasingly used for preoperative assessment of PTC. However, the ability of MRI in discriminating ETE is limited because the available evaluation criteria are primarily based on the morphological and anatomical details. Texture analysis is a method for extracting quantitative features from medical images and can be used to detect pathological changes that cannot be perceived by human visional evaluation ³. Previous studies also showed its reliability in differentiating thyroid malignant nodules from benign as well as comparing negative and positive lymph nodal in PTC^4,5. However, to our knowledge, no published study has been performed to investigate if T2-weighted imaging(T2WI) texture features are possible for predicting the extent of ETE in PTC. Therefore, the main goal in this study was to investigate if preoperative T2WI derived histogram and gray-level co-occurrence matrix (GLCM) texture features are feasible in predicting the existence of ETE in PTC patients.

Materials and Methods

A total of 76 patients (mean age 46.11±11.68 years, 63.2% female) with PTC confirmed by pathology in our hospital were enrolled in this study and underwent MRI examinations preoperatively. Each patient also underwent ipsilateral lobectomy or total thyroidectomy. Tissue samplings were collected from lesions during surgery served as a gold standard for histopathological reference and assessment of ETE status. All MRI experiments were performed on 1.5 Tesla MR scanner (GE Signa HD 1.5 T MR scanner; GE Healthcare Systems, Milwaukee, WI, USA) with an eight-channel high-resolution receiver synergy-head/neck phased-array coil. MRI protocols included transverse T1WI, transverse T2WI, diffusion weighted imaging (DWI) and contrast-enhanced MRI. For transverse T2WI scanning, the scan parameters were as follows: repetition time (TR)/echo time (TE), 3500/95ms; matrix size, 128 × 128; field of view (FOV), 140 × 140 mm²; slice thickness, 3 mm; spacing between slices, 1mm. A total scan time was 30 minutes. Texture analysis was performed independently on axial T2WI using 3D slicer (Version V4.10.0) by two senior radiologists blinded to the pathological results. Regions of interest (ROIs) covering the whole tumor were manually delineated along the tumor contour on each section, excluding obvious necrosis and cystic areas (Fig 1). 9 histogram and GLCM texture features were automatically extracted. All statistical analyses were performed in SPSS 17.0 software. Categorical variables were compared between groups using the chi-square test. Comparisons of age, maximum diameter of tumor and texture parameters in different patient groups were performed using the independent-samples t-test or Mann-Whitney U test depending on distribution normality or non-normality, respectively. Spearman correlation analysis was applied to assess the correlation between texture parameters and ETE. Subsequently, multivariate binary logistic regression analysis was performed to identify independent predictors for ETE. Interobserver agreement was evaluated using the intraclass correlation coefficient (ICC). Receiver operating characteristic (ROC) curves were generated to assess the diagnostic performance of the texture parameters in predicting ETE status by calculating the area under the ROC curve (AUC). P< 0.05 was considered statistical significance.

Results

Clinicopathological features of total 76 patients with and without ETE were included in our study (Table 1). The histogram and GLCM parameters between both groups were shown in Table 2. Significantly lower energy and correlation as well as higher entropy and standard deviation were found in PTC patients with ETE than without ETE group (Table 2). Interobserver agreement between both radiologists were excellent for all texture features derived from separately delineated ROIs. The ICC’s range was 0.78-0.89. The logistic regression analysis demonstrated that high entropy was an independent risk factor of ETE (odds ratio, OR = 19.348; 95%CI, 4.578-81.760; p= 0.001; Table 3). Using ROC analysis, the entropy showed moderate to good diagnostic power with a cutoff value > 5.86 (AUC = 0.837, 95%CI, 0.764–0.910) in predicting ETE, yielding sensitivities of 81.5% and specificities of 75.6% (Fig 2).

Discussion and Conclusions

This study performed texture analysis of primary tumors on preoperative transverse T2WI and then investigated the correlation between texture features and the ETE state in PTC. As shown in the results, entropy, energy, standard deviation and correlation extracted from histogram and GLCM differed significantly between patients with ETE and without ETE. Multivariate analysis showed that entropy was an independent risk factor of ETE with an OR of 19.348. Entropy represents the randomness or heterogeneity of the pixel distribution, and thus high entropy distinguishes hidden microcosmic heterogeneity in PTC patients with ETE from those without ETE. In conclusions, texture features extracted from T2WI might be considered potential valuable imaging biomarkers in predicting the ETE state of PTC.

Acknowledgements

This study was supported in part by a grant-in-aid for scientific research from the Technology Plan of Jiangsu (Project No. H2019087), the Technology Plan of Wuxi (Project No. MS201901) and the Science and Technology Development Plan of Wuxi (Project No. N20192027).