PROCESSING AND FEATURE EXTRACTION OF BIOMEDICAL IMAGES

Abstract

Processing of Bio-Medical Images, extracting the features and obtaining other important information from them are the latest trends in the disease diagnosis. Digital Mammography is the recent tool in the field of Breast Cancer diagnosis. As the conventional clinical ways of processing and feature extraction is quite cumbersome and error prone for the Radiologists, computer aided processing and extraction of features from them becomes a better option to relay on, in the process of detection of the disease. Two important problems that have been covered in this thesis with regarding to digital mammography are Enhancement of the mammographic images and the second one is detection of subtle and complex microcalcifications, which are an important symptom of breast cancer. A detailed computational procedure has been developed for the quantitative comparison of different enhancement techniques for the digital mammograms. This procedure will gives out the best enhancement technique based on recently proposed quantitative measures. Multiscale decomposition of mammograms based on Multiresolution Analysis using wavelets has been carried out, which helps in detecting the microcalcifications. Wavelet analysis is an extremely powerful data representation method that allows for the separation of images into different frequency bands without affecting the spatial locality. Thus, information concerning localized high-frequency signals such as microcalcifications can be extracted effectively.