Statistic Texture Feature Based Thyroid Nodule Recognition on CT Images
Peng Wenxian
Liu Chenbin
Xia Shunren
Chen Yihong
Liu Rui
Abstract：Objective To evaluate the feasibility of classifying the malignant thyroid nodule from benign one on computed tomography(CT) images based on gray level co-occurrence matrix and gray level gradient co-occurrence matrix texture features.Methods One hundred and thirty four CT images from inpatients underwent thyroid nodule surgery were enrolled in this study.A senior radiologist delineated the thyroid contour manually and segmented the region of interest (ROI).Texture features of GLCM and GLGCM were extracted and scaled to [0,1].Support vector machine was used as classifier.Leave-one-out cross validation (LOOCV) strategy was applied to evaluate the performance.Results Statistic texture features were applied on the classification of thyroid nodule recognition and the results of the proposed method were accuracy 0.76,sensitivity 0.60,specificity 0.86 and area under receiver operating curve (AUC)0.81 respectively.Conclusion The texture features of GLCM and GLGCM can be used as image biomarker in classifying malignant thyroid nodule from benign one on CT images.
Keywords：thyroid nodulesgray level co-occurrence matrixtexture featuresupport vector machine
Publication Date：2017-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：5( 258-262 )