Project description
Computed Tomography (CT) is a powerful technique used throughout medical imaging, fundamental research and industrial applications. At the moment, the main limiting factor for the resolution one can achieve in CT is the scanner geometry calibration. Calibration is a challenging and expensive procedure, and every error accumulates to result in blurring of images produced by CT.
In this project, you will build a new approach to estimate the geometry parameters accurately in a data-driven manner. You will explore automatic differentiation for framing the geometry calibration as a smooth optimization problem with a suitable image quality metric.
The result will allow for improving the quality of CT images in a general way, enabling better diagnostics, research, and industrial decision-making.
