Title:Textural Analysis and Diagonsis Image Fusion Classification of Medical Images Using Various Transforms Authors: Mary Praveena S.; Ila Vennila; Sekar; G.; R Vaishnavi; Nithya; M DOI: Aff: Department of ECE, Sri Ramakrishna Institute of
Technology, Coimbatore, India Author Email: Keywords: wavelet,Ridgelet; Curvelet,radon,histogram Equalization URLs:ABSTRACT-HTML | FULLTEXT-PDF | Abstract:Background: Textural classification of tumor images is done based on histogram equalization and texture analysis techniques of wavelet, Ridgelet and Curvelet transforms. For low quality medical CT images, a general framework based on conventional histogram equalization for image feature enhancement is presented. CT images are taken (brain tumor images) and they are transformed using various transforms such as Wavelet, Curvelet and Ridgelet transforms. The transformed image is given to the feature extraction module, processed to get various features such as Energy, Entropy, Homogeneity, Contrast, Mean, Variance, Standard Deviation, by using these features, the efficiency of various multi resolution image transforms are found. The main objective is to Develop an imaging system for classification of tumour tissues in medical images obtained. The discriminating powers of these transforms are helpful for classification of tumor types such as Glioma, Menin and Metastatic are pondered upon. Analyzing the best multi-resolution transform method and the method which gives the better resolution between similar type of diseased images and dissimilar types of images. The tumor images (glioma, meningioma, metastatic) of the CT scan is taken. Tumor varies for different persons. So two persons per tumor had been taken and three types of transforms (Haar Wavelet, Daubachies Wavelet, Curvelet transform and Ridgelet transform) are applied. From the results Wavelet is sporadic. Ridgelet gives near approximate values.curvelet gives approximate values and is the best multiresolution method.