ROIX-Comp: Optimizing X-ray Computed Tomography Imaging Strategy for Data Reduction and Reconstruction

In high-performance computing (HPC) environments, particularly in synchrotron radiation facilities, vast amounts of X-ray images are generated. Processing large-scale X-ray Computed Tomography (X-CT) datasets presents significant computational and storage challenges due to their high dimensionality and data volume. Traditional approaches often require extensive storage capacity and high transmission bandwidth, limiting real-time processing capabilities and workflow efficiency. To address these constraints, we introduce a region-of-interest (ROI)-driven extraction framework (ROIX-Comp) that intelligently compresses X-CT data by identifying and retaining only essential features. Our work reduces data volume while preserving critical information for downstream processing tasks. At pre-processing stage, we utilize error-bounded quantization to reduce the amount of data to be processed and therefore improve computational efficiencies. At the compression stage, our methodology combines object extraction with multiple state-of-the-art lossless and lossy compressors, resulting in significantly improved compression ratios. We evaluated this framework against seven X-CT datasets and observed a relative compression ratio improvement of 12.34x compared to the standard compression.

Sub-photon accuracy noise reduction of single shot coherent diffraction pattern with atomic model trained autoencoder

Single-shot imaging with femtosecond X-ray lasers is a powerful measurement technique that can achieve both high spatial and temporal resolution. However, its accuracy has been severely limited by the difficulty of applying conventional noise-reduction processing. This study uses deep learning to validate noise reduction techniques, with autoencoders serving as the learning model. Focusing on the diffraction patterns of nanoparticles, we simulated a large dataset treating the nanoparticles as composed of many independent atoms. Three neural network architectures are investigated: neural network, convolutional neural network and U-net, with U-net showing superior performance in noise reduction and subphoton reproduction. We also extended our models to apply to diffraction patterns of particle shapes different from those in the simulated data. We then applied the U-net model to a coherent diffractive imaging study, wherein a nanoparticle in a microfluidic device is exposed to a single X-ray free-electron laser pulse. After noise reduction, the reconstructed nanoparticle image improved significantly even though the nanoparticle shape was different from the training data, highlighting the importance of transfer learning.