Model-based reconstruction methods for CT perfusion imaging
M1.4 | Use of prior knowledge for interventional MRI
This sub-project aims at the reconstruction of dynamic time series from fast acquisitions.
Typically, these fast acquisitions are of lower quality (e.g. wrt resolution, contrast, or artefacts) compared to slower scans with higher resolution, the latter being acquired for the purpose of planning. At the same time we know that the object is mainly left unchanged apart from potential non-linear deformations and the presence of an interventional tool (e.g. a needle) with its position being precisely known.
Consequently, a lot is known about the object expecting this prior knowledge to enable the reconstruction of dynamic high resolution and high contrast images.
Therefore, different approaches may be applied including image-based matching and deformation, model-based reconstruction using prior knowledge to support regularisation, or even machine learning methods.
Sub-project kickoff: Jan-01, 2018.