Intensity based Image Registration on Brain MRI Images
Prathima Devadas, G. Kalaiarasi, M. Selvi · 2020 Second International Conference on Inventive Research in Computing Applications (ICIRCA) · 2020
In recent days, people are prone to many diseases that can be detected effectively through scans. The severity of the problem can be analyzed by taking periodic scans and the treatment also depends on the same. The intensity based image registration was applied to the mammography images in the previous works. In this paper, the evaluation of the intensity based image registration is made on the brain MRI (Magnetic Resonance Imaging) images which are used frequently among people suffering from any brain disorders. The image registration is the process of detecting the distortion that occurs in the image from its original structure. The intensity based registration technique is used to avoid human intervention and make the process convenient for patients. The image registration is performed using various metrics and methods that vary from global transformations to the local transformations. The similarity metrics used for image registration are the sum of the squared differences and mutual information. These metrics are evaluated using the three registration methods namely: Affine, Demons and Hybrid (Affine + B-Spline). The algorithm that results with higher similarity image is taken to be the efficient image registration algorithm for Brain MRI images. MATLAB has been used to analyze this idea and had produced better results.