Supervised learning for streak artifact reduction in flat detector CT perfusion imaging

Flat detector CT perfusion is a new imaging modality that allows for the quantication of perfusion maps in the interventional suite, saving the time of moving the patient from the scanner room. The acquisition process of FD-CTP bears challenges due to the limitations of C-arm hardware. The reconstruction is subject to large amounts of noise and artifacts. This work investigates a way of reducing streak artifacts through machine learning algorithms. Both simulation and clinical data are used