Majority voting based classification of thyroid carcinoma

AbstractThis paper presents the classification of Papillary carcinoma and Medullary carcinoma cells in Fine Needle Aspiration Biopsy (FNAB) microscopic cytological images of thyroid nodules under varying staining conditions. Initially image segmentation is performed to remove the background staining information in microscopic images using mathematical morphology. Feature extraction is carried out by Discrete Wavelet Transform (DWT) and Gray Level Co-occurrence Matrix (GLCM) and the classification is done using k-Nearest Neighbor (kNN) classifier. The DWT reports the maximum diagnostic accuracy of 97.5% while GLCM reports the diagnostic accuracy of 75.84%. However the diagnostic accuracy of GLCM has been improved as 90% by implementing the majority voting rule