Classification of Ultrasonographic Thyroid Tumor Images to TIRADS Categories via Texture Analysis Methods

Abstract

Thyroid tumors frequently observed disease by using medical imaging methods. Ultrasonography is the most frequently performed method for diagnosis. To determine, tumor is benign or malign, experienced doctors use various techniques. Fine needle aspiration biopsy and follow-up checking are used for determining type of tumor. However, these methods are time consuming and increasing work load of doctors. So, they created a risk stratification system which has called as ACR-TIRADS. Downside of this system is being subjective and for multiple tumors, it will be time consuming due to analyzing multiple features. To ease, doctors work load and help them to obtain more objective classification, on this study we worked thyroid tumors with texture analysis methods and tried to classify them, to their TIRADS classes. As the result of this study, sensitivity found up %82.8, precision %85 and accuracy found up %73.0.