Resumen de Improvements in the registration of multimodal medical imaging: application to intensity inhomogeneity and partial volume corrections
Alignment or registration of medical images has a relevant role on clinical diagnostic and treatment decisions as well as in research settings. With the advent of new technologies for multimodal imaging, robust registration of functional and anatomical information is still a challenge, particular in small-animal imaging given the lesser structural content of certain anatomical parts, such as the brain, than in humans. Besides, patient-dependent and acquisition artefacts affecting the images information content further complicate registration, as is the case of intensity inhomogeneities (IIH) showing in MRI and the partial volume effect (PVE) attached to PET imaging. Reference methods exist for accurate image registration but their performance is severely deteriorated in situations involving little images Overlap. While several approaches to IIH and PVE correction exist these methods still do not guarantee or rely on robust registration. This Thesis focuses on overcoming current limitations af registration to enable novel IIH and PVE correction methods.
