Biomedical image analysis based on computational registration methods
João Manuel R. S. Tavares · Portuguese National Funding Agency for Science, Research and Technology (RCAAP Project by FCT) · 2014
Image registration, which has become a paramount research topic, is the process of transforming an image so that the associated entities are properly adjusted to the homologous entities in a second image (Figure 1). Such transformations have not only been applied to static bi-dimensional (2D) and tridimensional (3D) images, but also to 2D and 3D image sequences. For example, in Medical Imaging, computational methods of image registration have been assuming an essential role in supporting enhanced image-based diagnosis by addressing: the automatic identification of regions of interest in images (i.e. image segmentation), the fusion of information acquired by different imaging systems (i.e. image fusion), the more effective follow-up of organs and pathologies, and the definition of the best plans in computer-assisted surgery or in radiotherapy treatments, among other roles [1]. Hence, the computational registration of medical images is an extremely useful tool for clinicians and researchers since, after the accurate registration of the data involved, tasks such as shape reconstruction, comparison of a given clinical case with previous cases are facilitated and can be performed with less subjectivity. Moreover, the identification of regions of interest and information fusion can be handled automatically. Figure 1 – Registration of two plantar pressure images: original images, and the two plantar pressure regions overlapped (in pseudo-colors) before and after the registration process [1, 12]. Other topics of image analysis are usually associated to image registration, including: image matching, i.e., the searching for correspondences between related images [2-6], similarity measurements, optimization, and image interpolation, especially due to the application of the registration transformations in the image discrete domain [1], Figure 2.