Effect on diagnostic image quality of 3D warping registration algorithms for MR angiography in the lower legs

Abstract

Since the birth of functional magnetic resonance imaging (fMRI) in the 90’s and more recently of diffusion tensor imaging (DTI), several clinical applications of these techniques have emerged, however the most widespread clinical trend is their use for pre-surgical planning. The goal of a neurosurgical procedure is to maximize the resection of the lesion (e.g. tumour, vascular malformation, epileptic foci) while minimizing permanent injury, which can be caused either by damaging the cortical areas immediately surrounding the lesion or the white matter tracks at the depths of the lesion. Advanced neuroimaging prior to surgery enables the neurosurgical team to plan the procedure and to envisage its complications, resulting in reduced surgical and recovery times and ultimately reduced morbidity. Furthermore, the coregistered datasets can also be used during neuronavigation although with limited reliability, due to the non-linear brain shift that occurs following the removal of the skull. Despite the clear benefits of pre-surgical mapping, implementing such a service in a busy clinical environment poses some challenges; additional MRI sequences result in increasing scanner occupancy, revision and standardization of acquisition protocols is necessary to achieve reliable brain activations in shorter times, and the complexity of the data analysis requires a multidisciplinary team of experts. During the past year, a pre-surgical planning unit has been implemented in Beaumont Hospital. Motor fMRI and DTI data are acquired with a 1.5T Siemens Magnetom Avanto and transferred into PACS. The raw data is retrieved and transferred to a Linux workstation where is further analysed with SPM8 and Slicer software packages. Functional and tractography results are coregistered and superimposed on a high-resolution anatomical image and a three-dimensional reconstruction of the tumour and white fibre tracks surrounding it is generated. This 3D map can be uploaded in the operating theatre for review prior and during surgery. Future work will address some of the challenges facing the clinical implementation of advanced neuroimaging in neurosurgery, such as the standardization of acquisition protocols for language and memory mapping and the integration of the coregistered maps into PACS and neuronavigation systems.