Abstract

We find the usefulness of computers in every field including medical field. Scanning the affected part has become a standard study. Diagnosing a disease at the right time, i.e. early detection, from the study of images enables the physician to take right decision and provide proper treatment to the patient. With the alarming growth of population, it is difficult for every individual patient to get a second opinion from medical expert. In these situations, computer-aided automatic diagnosis system will be much helpful. Diabetic retinopathy is a disorder that arises from increase in blood glucose level. Based on the severity, it has been distinguished into four stages. Diagnosing diabetic retinopathy at an early stage from retinal images and providing proper treatment will save the patient from severe vision loss. The proposed method adopts hybridized GLCM features and wavelet features to classify the fundus images according to the severity of the disease. The method is tested with fundus images collected from Indian Diabetic Retinopathy Dataset. The bias field is an undesirable image foible that formulate during the process of image procurement. Segmentation is the procedure of segregating a digital image into constituent component or substantial segments which help in extracting quality amount of information from the region of interest. There is several bias correction strategies have been recommended till date, all these algorithms helps in reducing bias but none of them perfectly removes bias. When incorporating computer aided diagnosing in treatment planning, the leftover bias cause to inaccurate segmentation which leads to faulty diagnosis of the diseases. This paper scrutinizes the segmentation algorithms over bias corrupted brain MR Images and analyzes which segmentation algorithm efficiently segments the image components even though it is corrupted by bias field. The bench mark brain MR Images with different bias spectrum is employed for the research. Quantitative metrics are adopted to conclude the result. The outcome of this paper tends to provide accuracy in computer aided diagnosing and to elect appropriate segmentation technique while developing bias correction based segmentation algorithm.