An Event-Reconstructing Silicon Detector for 1 µm Resolution Spectral Computed Tomography
Time: Fri 2022-04-29 09.15
Location: FA32, Roslagstullsbacken 21, Stockholm
Subject area: Physics
Doctoral student: Christel Sundberg , Medicinsk bildfysik
Opponent: Professor Julia Herzen, Technische Universität München
Supervisor: Professor Mats Danielsson, Medicinsk bildfysik
Computed tomography (CT) is a medical imaging modality in which cross-sectional images of the human body are created using x-rays. Commercial CT scanners utilize energy-integrating detectors to measure the x-ray attenuation. However, photon-counting detectors with energy-discriminating abilities have started to emerge. In a photon-counting spectral detector, photons can be counted individually and the photon energy is registered using energy thresholds. In contrast to energy-integrating detectors, which integrate all photon energies during a measurement interval, this allows for an improved detector performance including an increased signal-to-noise ratio, higher spatial resolution, and improved spectral imaging.
One of the current photon-counting systems that is being evaluated for clinical use is the deep silicon detector developed by the Physics of Medical Imaging group at KTH. This Thesis is based on the deep silicon detector concept and focuses on methods to improve the performance of a silicon photon-counting detector for CT and how these might facilitate event reconstruction. In the first part of the Thesis, three different methods to improve the detector performance are presented. One of the methods describes how information about the charge cloud distribution can be used to improve the spatial resolution. With the proposed method, subpixel resolution can be achieved, corresponding to a spatial resolution equivalent of approximately 1 μm in the most accurate dimension. A silicon detector with double-sided readout electrodes is further proposed which enables estimating the time of the photon interaction with high accuracy. The resulting time resolution of approximately 1 ns can potentially be utilized to identify interactions that originate from the same incident photon. With double-sided readout, it is also possible to dramatically improve the spatial resolution in the direction across the silicon wafer thickness. It is also proposed to utilize an adjustable shaping time in the readout electronics to decrease the electronic noise level. This can be used to improve the detector performance with respect to dose efficiency and power consumption.
In the second part of the Thesis, a method to perform event reconstruction is presented. The method consists of a framework of likelihood functions that are used to estimate the incident photon energy and primary interaction position. Based on this framework, the ability of estimating the photon energy and primary interaction position is evaluated for a case in which the incident photons are assumed to be well-separated in time.
In summary, there is potential in increasing the performance with respect to the spatial, temporal, and energy resolution in silicon photon-counting detectors for CT and the results suggest that event reconstruction might be possible in the future.