Quinn Paul D, Sabaté Landman Malena, Davis Tom, Freitag Melina, Gazzola Silvia, Dolgov Sergey
Scientific Computing, Science and Technology Facilities Council, Rutherford Appleton Laboratory, Harwell Campus, Didcot OX11 0QX, United Kingdom.
Department of Mathematics, Emory University, Atlanta, Georgia 30322, United States.
Chem Biomed Imaging. 2024 Mar 19;2(4):283-292. doi: 10.1021/cbmi.3c00116. eCollection 2024 Apr 22.
The application of X-ray spectro-microscopy to image changes in the chemical state in application areas such as catalysis, environmental science, or biological samples can be limited by factors such as the speed of measurement, the presence of dilute concentrations, radiation damage, and thermal drift during the measurement. We have adapted a reduced-order model approach, known as the discrete empirical interpolation method, which identifies how to optimally subsample the spectroscopic information, accounting for background variations in the signal, to provide an accurate approximation of an equivalent full spectroscopic measurement from the sampled material. This approach uses readily available prior information to guide and significantly reduce the sampling requirements impacting both the total X-ray dose and the acquisition time. The reduced-order model approach can be adapted more broadly to any spectral or spectro-microscopy measurement where a low-rank approximation can be made from prior information on the possible states of a system, and examples of the approach are presented.
X射线光谱显微镜在催化、环境科学或生物样品等应用领域用于成像化学状态变化时，可能会受到测量速度、低浓度存在、辐射损伤以及测量过程中的热漂移等因素的限制。我们采用了一种降阶模型方法，即离散经验插值法，该方法确定如何对光谱信息进行最优采样，同时考虑信号中的背景变化，以便从采样材料中提供等效全光谱测量的准确近似值。这种方法利用现成的先验信息来指导并显著降低采样要求，从而影响总X射线剂量和采集时间。降阶模型方法可以更广泛地应用于任何光谱或光谱显微镜测量，只要可以根据系统可能状态的先验信息进行低秩近似，本文还给出了该方法的示例。
