ABSTRACT

Medical imaging is the technique used to create images of the human body for clinical purposes of medical science. In bio-medical image processing, stress has been laid on creating several techniques for developing the bio medical images to be more effective and efficient to identify the abnormalities in the medical images. There are mainly two objectives for biomedical image processing. The initial objective is. to create more suitable images for doctors to observe and to identify the diseased regions within the image. The secondary objective is to develop an automatic technique which will be able to diagnose and clearly detect the abnormal region in the medical image. In the initial part ofthis work, edge detection using Sobel and Prewitt operators are done. In that section, effort is taken to establish these two well known operators in the form of developed MATLAB coding which is giving better results than those obtained from toolbox MATLAB operator of Sobel and Prewitt. Image enhancement is the next step of this work because without image enhancement no work in image processing can be further extended. The enhancement is done on X-Ray, CT-Scan and MR images . In each case the result yielded from the bi-level histogram equalization technique is found to be good as compared to those when the same images are enhanced by histogram equalization of MATLAB toolbox. In the next stage of this work, filtering is done on the enhanced images. Median filter is the specialized filter used in this work for filtering. In the same way the MATLAB toolbox median filtering is also applied on the enhanced image. On comparative analysis both qualitatively and quantitatively the devised method of filtering has proved to be better than the toolbox method of filtering. In the next stage of the work is confined to pattern matching/recognition and identification of abnormal regions in an image. In this thesis, the main objective is to detect the abnormalities automatically for which it is necessary to prepare a sophisticated algorithm which can locate, detect and also automatically crop the abnormal regions from the image without any manual intervention. This automated algorithm is prepared on Java platform of Net Beans IDE. This automated processes required very minimal time for completion. After the completion of this process matching of the abnormal region is done from the dataset of various other diseased images of CTScan, MR imaging and X-ray considered in this work.