A Reconstruction Toolkit for Multichannel CT

目前，传统的计算机断层扫描（CT）成像仍然处于黑白（1通道）时代，就像1895年伦琴根拍摄的第一张图像一样！传统的x射线成像需要多色x射线源（即具有全光谱的能量），但需要能量不分青红皂白的探测器（记录单个灰度通道）。然而，能量敏感探测器的技术突破开启了“彩色”（多通道）层析成像的新时代。能量选择探测器的每个像素记录由数百或数千个能量通道组成的光谱。目前可用的软件只允许我们依次独立地重建每个（有噪声的）信道，这是一个显著的限制。我们需要释放下一代多通道断层扫描相关重建方法的力量。值得注意的是，注册的能量通道是相互相关的，就像彩色图像的红绿蓝（RGB）通道一样。因此，光谱测量中的噪声和其他不准确性可以通过通道整体处理，从而大大提高成像质量（更高的信噪比和分辨率），并从根本上带来新的机会，例如光谱成像，即直接分解为基本元素。CCP软件旗舰项目的总体目标是扩展现有的单通道图像重建软件（已经由CCPi项目开发），以实现复杂的多通道相关重建方法。一种新型的多通道CT重建工具包（RT-MCT）将被开发，并成为最终用户数据管道Savu的一部分（Savu是由英国Harwell的Diamond Light Source （DLS）开发的基于python的层摄影数据处理模块化平台）。三大主要成像设施是RT-MCT的关键合作者和忠实的初始用户：1)曼彻斯特x射线成像设施（MXIF）是实验室x射线CT成像的领导者，并开发了独特的多通道仪器“the Colour Bay”（使用HEXITEC超光谱探测器的锥形光束几何扫描仪）；2)在ISIS脉冲中子散裂源上新建的国家中子成像和衍射设备（IMAT） （Harwell）。IMAT将利用中子飞行时间（TOF）测量技术对数千个通道进行有效的能量识别，使这种独特的技术成为超光谱技术；3)钻石光源（DLS），位于哈维尔的国家同步加速器设施，有许多成像光束线，包括I18和I14，专门用于x射线荧光，x射线光谱和衍射，所有这些都需要多通道数据集。主要目标是将RT-MCT交付给这些设施，以提供更有效的数据重建和分析。确定了构成RT-MCT的几个工作包，即a)多通道成像的精确数学建模；B)最优重构问题的表述；C)在现有软件框架下高效的算法实现和集成；D)部署到设施和在概念验证案例研究中使用。

Currently, conventional Computed Tomographic (CT) imaging is still in a black and white (1 channel) era, just as it was with the first image Rontgen captured in 1895! Conventional X-ray imaging entails a polychromatic X-ray source (i.e. with a full spectrum of energies) but with energy-indiscriminate detectors (registering a single grey-scale channel). However, technological breakthroughs in energy-sensitive detectors enable a new era of tomographic imaging in 'colours' (multiple channels). Each pixel of the energy-selective detector records a spectrum consisting of hundreds or thousands of energy channels. Currently available software only allows us to reconstruct each (noisy) channel independently in turn, which is a significant limitation. We need to unlock the power of next-generation correlative reconstruction methods for multi-channel tomography. Notably, the registered energy channels are mutually correlated, just like the red-green-blue (RGB) channels of the color image. Therefore, noise and other inaccuracies in spectral measurements can be treated holistically across the channels, leading to massive improvements in imaging quality (higher signal-to-noise ratio and resolution) in addition to fundamentally new opportunities such as spectroscopic imaging, i.e., direct decomposition into fundamental elements. The overall goal of this CCP Software Flagship project is to expand upon existing single-channel image reconstruction software (already developed by the CCPi project) to enable sophisticated multi-channel correlative reconstruction methods. A novel Reconstruction Toolkit for Multichannel CT (RT-MCT) will be developed and become a part of the end-user data pipeline Savu (a modular Python-based platform for tomographic data processing developed at Diamond Light Source (DLS) at Harwell, UK). Three major imaging facilities are key collaborators and committed initial users of the RT-MCT: 1) Manchester X-ray Imaging Facility (MXIF) is a leader of laboratory-based X-ray CT imaging and has developed the unique multi-channel instrument "The Colour Bay" (cone-beam geometry scanner which uses HEXITEC hyper-spectral detectors); 2) A new national Neutron Imaging and Diffraction Facility (IMAT) at the ISIS pulsed neutron spallation source (Harwell). IMAT will take advantage of the neutron time-of-flight (TOF) measurement technique for effective energy discrimination into thousands of channels making this unique technique hyper-spectral; 3) Diamond Light Source (DLS), the national synchrotron facility at Harwell, has a number of imaging beamlines including I18 and I14, dedicated to X-ray fluorescence, X-ray spectroscopy and diffraction, all of which entail multi-channel data sets.The main aim is to deliver the RT-MCT to these facilities to provide much more efficient data reconstruction and analysis. Several work packages are identified which constitute the RT-MCT, namely a) accurate mathematical modelling of multi-channel imaging; b) formulation of optimal reconstruction problems; c) efficient algorithm implementation and integration in existing software framework; d) deployment to facilities and use in proof-of-concept case studies.