Signal Processing for X-ray Computed Tomography Data Compression
Explore advanced compression methods for massive XCT data to reduce storage needs by up to 80% without losing vital scientific detail. Develop predictive models exploiting XCT data redundancy within an open-source framework.
AI-generated overview
Project Description
Project Overview
Non-clinical X-ray Computed Tomography (XCT) generates vast amounts of data, exceeding 10TB annually per scanner. This research tackles the challenge of compressing such large datasets by designing specialized compression methods that exploit the spatial and sequential relationships within the projections and reconstructed volumes. The goal is a substantial reduction in data size while maintaining scientific information fidelity.
What You Will Do
You will join the AC/DC research team working to develop an open-source compression tool for the XCT community. The project involves applying signal processing expertise combined with statistical mathematical frameworks to maximize compression ratios. Techniques include predictor models to estimate XCT slice data, storing the differences which are then efficiently compressed using classical and data-driven models on XCT intensity statistics.
Expected Outcomes
The project aims to produce a tailored XCT compression algorithm achieving 60-80% data reduction without compromising data quality. This will help the scientific community better manage, transmit, and archive large tomographic datasets, reducing environmental and financial burdens.
Why This Matters
As XCT usage grows, efficient data management becomes critical. Current generic compression methods do not fully exploit the redundancy in XCT data. Developing novel, domain-specific compression will enable more sustainable and cost-effective scientific workflows and preserve valuable imaging data for future research.
How to Apply
Eligibility
Supervisor Profile
Dr. Jay Warnett is an Associate Professor at WMG, University of Warwick, specializing in X-ray computed tomography, metrology, and nondestructive evaluation. His research spans high-end imaging and signal processing, particularly for industrial and scientific applications. He has contributed extensively to 3D imaging and tomographic reconstruction techniques, with over 1400 citations and a strong interdisciplinary focus bridging engineering and applied mathematics.