PROJECT TITLE :

A Deep Learning Reconstruction Framework for Differential Phase-Contrast Computed Tomography With Incomplete Data

ABSTRACT:

Soft-tissue and low-atomic-number samples can be analysed using differential phase contrast computed tomography (DPC-CT). DPC-CT with incomplete projections is a common occurrence because of the technical constraints. When faced with incomplete data, conventional reconstruction methods have problems. Complex parameter selection operations, which are similarly sensitive to noise and take a long time, are common in these types of algorithms. For imperfect DPC-CT data, we present a new deep learning reconstruction approach. The DPC-CT reconstruction approach is tightly coupled with a deep learning neural network in the area of DPC projection sinograms. Complete phase-contrast projection sinogram is not an artefact caused by incomplete data, but an estimated outcome. This framework may be used to reconstruct the final DPC-CT images for a given incomplete projection sinogram after training. This system is tested and proven using synthetic and experimental data sets for sparse-view, limited-view, and missing-view DPC-CT. By using our framework, we are able to get better imaging quality in a shorter amount of time and with a smaller number of parameters. For the DPC-CT discipline, our work promotes the use of the most recent deep learning theory