Project: Image Reconstruction Methods for Photon Counting Computer Tomography

In order for the new (photon-counting) computed tomography technology to reach its full potential, the newly developed hardware needs to be supplemented with improved data processing algorithms so that measured data is fully utilized and provides the best possible image quality. In this project, we develop the next generation of image reconstruction methods for photon computing computed tomography and evaluate the resulting image quality. In collaboration with the General Electric Research Center in Niskayuna, NY, USA, we have developed a method for correcting for physical effects when taking pictures that can otherwise incluce artifacts. In collaboration with the Department of Mathematics at KTH, we have also developed an image reconstruction method based on deep neural networks, so-called deep learning, which can greatly reduce the noise in the images. In a few years’ time, the combination of photo-counting computed tomography with the next generation of image reconstruction can take the image quality in computed tomography to a whole new level.