Study on neurodegeneration at different stages using MR images: computational approach to registration process with optimisation techniques

摘要

The medical imaging can be divided into two global categories: anatomical and functional. Anatomical modalities, such as X-ray, computed tomography (CT), magnetic resonance imaging and ultrasound primarily describe the structural morphology of the object. Few other techniques such as electroencephalography, magneto encephalography, Positron emission tomography and single photon emission computed tomography are classified as functional imaging techniques for mapping brain activities. These images may be used in a complimentary manner to gain full insight into the shape, size and spatial relationships among the anatomical structures and functional details. Physicians often recommend multi-modal medical images to capture complete information of the region of interest under investigation. But due to the change of patient's/sensor's orientation, it is difficult for a physician to integrate all imaging information in a common coordinate frame. An essential step in this integration process is to bring into spatial alignment the images obtained from different modality techniques. This is referred to as Registration. Registration of two images with each other requires geometric transformations such as translation, rotation, scaling and sometimes skewing of one image such that the transformed image aligns with the other one, which is considered a standard/ reference image. In present paper, we have discussed the different steps of registration process for patients having Alzheimer's disease using magnetic resonance images. We have studied the efficiency of registration process with experimental results using different optimisation techniques. Finally, the performance of the proposed optimisation approach is analysed in terms of convergence speed.