Adenomatous Hyperplasia Of Thyroid Nodule Classification: Texture Feature Analysis Methodoloy Of Ultrasound Images

Abstract

This paper, explores to extract textural feature from ultrasound Adenomatous Hyperplasia thyroid nodule. The extracted texture feature will help the clinician to improve diagnostic accuracy. The Fine Needle Aspiration (FNA) and Histopathology report is the conventional diagnostic methodology. It is an invasive technique and patients are subjected to painful process. Non-invasive, non-contact and low cost imaging tool is essential to increase the clinical diagnostic accuracy. Ultrasound imaging is a potential non-invasive modality to capture Adenomatous Hyperplasia thyroid nodule. Adenomatous Hyperplasia thyroid nodule is the common inflammation in thyroid gland abnormality. The texture features, using Law’s Texture Energy Measures (Law’s TEM), Neighborhood Gray Tone Difference Matrix (NGTDM) and Statistical Feature Matrix (SFM) are extracted from normal and adenomatous hyperplasia of thyroid nodules. The normal and abnormal images are demarcated using T-test analysis. The extracted feature shows significant difference between normal and abnormal images with p value less than 0.001(p < 0.001). Hence the Law’s Texture Energy based feature may be used to identify the pathology in the thyroid ultrasound images.