Model-based reconstruction methods for CT perfusion imaging
M1.5 | Volume-of-interest imaging in C-arm CT
A key problem of computed tomography (CT) is the reconstruction of tomographic images from incomplete projection data, commonly termed 'truncation'.
Truncation occurs when the measured region is constrained to not contain the whole patient, but only a spatially limited region-of-interest (ROI) mainly for the purpose of dose reduction. The resulting projection data therefore appear to be abruptly "cut off", representing a high frequency disturbance. Image reconstruction based on truncated projection data therefore gives rise to image artefacts.A typical strategy to counter these artefacts in regular CT is to extrapolate the measured ROI using some smooth function in order to reduce the impact of truncation.
Given truncations being a very common scenario in interventional C-arm CT, the objective of this sub-project is to develop a novel extrapolation method especially suited for volume-of-interest (VOI) imaging in conebeam C-arm CT (CBCT).
This will be realised by (i) incorporating consistency conditions inherent to valid CBCT projections, which have previously been proven to be applicable for related problems such as motion compensation or beam hardening as well as by (ii) including additional a priori information on the intervention itself.