Funding period: Jul 2017 to Sep 2021

Researcher: Daniel Punzet, née Hellge-Theune

Wrap-up

Keywords: CBCT, volume-of interest imaging, truncation, prior knowledge, registration

Background:
Volume-of-interest imaging allows for significant patient dose reduction. However, reconstructed images suffer from severe image artifacts due to the limited data acquisition. Yet, in practice there is typically unused data of the patient available.

Objective:
Utilization of the available prior knowledge to increase image quality of VOl imaging or reduce dose respectively. 

Methods:
Usage of consistency conditions to incorporate prior data properly while maintaining and not overwriting information from VOI imaging acquisitions. This is achieved by registration of priors and the retrieval of further information from the limited data available. 

Results:
Image reconstruction from truncated projections supported by prior volume data offers good image quality while reducing patient dose. Final investigations still need to show how well the method works on clinical devices. 

Conclusions:
Extrapolation methods using solely consistency conditions to improve image quality do not work sufficiently stable, however incorporating available prior data enables good image results.

Originality:
Usage of previously unused information enables patient dose reduction while maintaining sufficient image quality.