Secured system with pattern recognition for biomedical images

Abstract

Images play a vital role in field of life whether it is medical, defense, military, bank, surveillance, traffic monitoring, weather forecast, etc. An image itself explains more than the words. Medical images are very important in doctors as well as patient life. Without the images doctors would not have been able to diagnose the disease and would not be able to give proper treatment. There are various diseases in the world but major deaths are due to cardiovascular disease [world health organization (WHO), 2008]. Thus, the aim of this thesis would be to concentrate on the images related to cardiovascular disease like carotid artery. The medical images are taken from different modalities like MRI, CT, Ultrasound, etc. Developing countries like India where major population is from average class and below average class family who could not afford funds for the treatment of the diseases. Ultrasound images are the best solution for it as they are affordable, harmless, easily available and could image most of the cardiovascular diseases. Thus, the aim of the thesis would be to process the ultrasound images. Ultrasound signals suffer from scattering process [Wagner, (1983)] and produces speckle in the image. The speckle in the ultrasound image degrades the quality of the picture and thus makes the doctors to predict the condition of the patient. The speckle present is of multiplicative as well as additive nature [Frost et al., (1982)]. Therefore, the aim of the thesis would be to filter the speckle present in the images. There are various Linear filters like average [Gonzalez and Woods, (2002)]; Frery, et al., (1997); Narayanan and Wahidabanu, (2009)] weighted average, local statistics mean variance filter [Loizou et al., (2005); Loizou et al., (2002); Christodoulou et al., (2002); Lee, (1981a); Lee, (1980); Lee, (1981b); Kondo et al., (1977); Walkup and Choens, (1974); Kuan et al., (1987); Akl et al., (2012)] and non-linear filters like median [Caloope et al., (2004); Pratt, (1978); Loupas et al., (1989)]