Deep learning based spectral CT imaging.

Spectral computed tomography (CT) has attracted much attention in radiation dose reduction, metal artifacts removal, tissue quantification and material discrimination. The x-ray energy spectrum is divided into several bins, each energy-bin-specific projection has a low signal-noise-ratio (SNR) than the current-integrating counterpart, which makes image reconstruction a unique challenge. Traditional wisdom is to use prior knowledge based iterative methods. However, this kind of methods demands a great computational cost. Inspired by deep learning, here we first develop a deep learning based reconstruction method; i.e., U-net with Lpp-norm, Total variation, Residual learning, and Anisotropic adaption (ULTRA). Specifically, we emphasize the various multi-scale feature fusion and multichannel filtering enhancement with a denser connection encoding architecture for residual learning and feature fusion. To address the image deblurring problem associated with the L22- loss, we propose a general Lpp-loss, p>0. Furthermore, the images from different energy bins share similar structures of the same object, the regularization characterizing correlations of different energy bins is incorporated into the Lpp- loss function, which helps unify the deep learning based methods with traditional compressed sensing based methods. Finally, the anisotropically weighted total variation is employed to characterize the sparsity in the spatial-spectral domain to regularize the proposed network In particular, we validate our ULTRA networks on three large-scale spectral CT datasets, and obtain excellent results relative to the competing algorithms. In conclusion, our quantitative and qualitative results in numerical simulation and preclinical experiments demonstrate that our proposed approach is accurate, efficient and robust for high-quality spectral CT image reconstruction.

Beam profile assessment in spectral CT scanners.

In this paper, we present a method that uses a combination of experimental and modeled data to assess properties of x-ray beam measured using a small-animal spectral scanner. The spatial properties of the beam profile are characterized by beam profile shape, the angular offset along the rotational axis, and the photon count difference between experimental and modeled data at the central beam axis. Temporal stability of the beam profile is assessed by measuring intra- and interscan count variations. The beam profile assessment method was evaluated on several spectral CT scanners equipped with Medipix3RX-based detectors. On a well-calibrated spectral CT scanner, we measured an integral count error of 0.5%, intrascan count variation of 0.1%, and an interscan count variation of less than 1%. The angular offset of the beam center ranged from 0.8° to 1.6° for the studied spectral CT scanners. We also demonstrate the capability of this method to identify poor performance of the system through analyzing the deviation of the experimental beam profile from the model. This technique can, therefore, aid in monitoring the system performance to obtain a robust spectral CT; providing the reliable quantitative images. Furthermore, the accurate offset parameters of a spectral scanner provided by this method allow us to incorporate a more realistic form of the photon distribution in the polychromatic-based image reconstruction models. Both improvements of the reliability of the system and accuracy of the volume reconstruction result in a better discrimination and quantification of the imaged materials.