Design of Spatial-Spectral Filters for CT Material Decomposition

Matthew Tivnan, Wenying Wang, Grace J. Gang, J. Webster StaymanarXiv 2020 · DOI · arXiv

Abstract

Spectral CT has shown promise for high-sensitivity quantitative imaging and material decomposition. This work presents a new device called a spatial-spectral filter (SSF) which consists of a tiled array of filter materials positioned near the x-ray source that is used to modulate the spectral shape of the x-ray beam. The filter is moved to obtain projection data that is sparse in each spectral channel. To process this sparse data, we employ a direct model-based material decomposition (MBMD) to reconstruct basis material density images directly from the SSF CT data. To evaluate different possible SSF designs, we define a new Fisher-information-based predictive image quality metric called separability index which characterizes the ability of a spectral CT system to distinguish between the signals from two or more materials. This predictive metric is used to define a system design optimization framework. We have applied this framework to find optimized combinations of filter materials, filter tile widths, and source settings for SSF CT. We conducted simulation-based design optimization study and separability-optimized filter designs are presented for water/iodine imaging and water/iodine/gadolinium/gold imaging for different patient sizes. Finally, we present MBMD results using simulated SSFCT data using the optimized designs to demonstrate the ability to reconstruct basis material density images and to show the benefits of the optimized designs.

Introduction

Conventional Computed Tomography (CT) imaging typically measures X-ray attenuation in Hounsfield Units (HU). While useful, HU values are inconsistent across different scanners and energy settings, limiting their utility for quantitative tissue characterization. Spectral CT addresses this by performing "material decomposition," enabling the quantification of specific substance densities (e.g., iodine concentration in mg/mL). However, existing spectral CT technologies, such as dual-source scanners or photon-counting detectors, often require significant hardware investments.

This work introduces the Spatial-Spectral Filter (SSF), a novel, potentially lower-cost modification for standard energy-integrating detector systems. The SSF consists of a tiled array of metallic foils positioned near the X-ray source to modulate the spectral shape of the beam.

The Spatial-Spectral Filter Concept

The SSF utilizes a tiled array of materials like Tin (Sn), Gadolinium (Gd), and Gold (Au). These materials are chosen for their unique "k-edges"—sudden jumps in X-ray absorption at specific energy levels. As the X-ray source rotates, the filter translates, creating distinct "spectral fingerprints" for different beamlets.

Overview of the SSF hardware design (left) and the resulting spectral modulation for various filter materials (right).

This setup generates "sparse" spectral data, as each region of the patient is only sampled through a subset of the filters. To reconstruct usable images, the researchers employ Model-Based Material Decomposition (MBMD), an iterative algorithm that reconstructs material density images directly from raw projection data.

Optimization via the Separability Index

A critical challenge in SSF design is determining the optimal combination of filter materials, thicknesses, and tile widths. To solve this, the authors developed a new mathematical metric called the Separability Index ($s^2$). Based on the Fisher Information Matrix, this metric quantifies the system's ability to distinguish between signals from different materials (e.g., water vs. iodine) while accounting for noise and cross-talk.

Using a Covariance Matrix Adaptation Evolution Strategy (CMA-ES), the team optimized SSF designs for specific clinical tasks. Interestingly, optimized designs often feature asymmetric tile widths to balance the photon flux reaching the detector across different spectral channels.

Results and Discussion

The study, conducted using digital phantoms and high-resolution simulations, demonstrated the successful simultaneous decomposition of four materials: water, iodine, gadolinium, and gold. This capability could pave the way for "multi-contrast" imaging, where different agents highlight distinct vascular phases in a single scan.

Reconstructed concentration map for a differentiating task in water, demonstrating the system's ability to isolate material signals.

Optimized SSF designs significantly outperformed randomized or equal-spacing designs. For instance, an optimized design achieved a separability index of 0.174, drastically reducing the "anti-correlated noise" that often plagues material decomposition and improving the visibility of small contrast inserts.

Conclusion

The SSF presents a promising path toward high-sensitivity quantitative CT imaging using existing detector technology. While this simulation-based study highlights the potential of optimized spatial-spectral filtering, future work must address real-world factors such as X-ray scatter, detector lag, and mechanical vibrations. Further research into "Adaptive CT"—where filter motion or tube voltage is adjusted in real-time—could further enhance the clinical utility of this technology.

Summary updated

Abstract

Spectral CT has shown promise for high-sensitivity quantitative imaging and material decomposition. This work presents a new device called a spatial-spectral filter (SSF) which consists of a tiled array of filter materials positioned near the x-ray source that is used to modulate the spectral shape of the x-ray beam. The filter is moved to obtain projection data that is sparse in each spectral channel. To process this sparse data, we employ a direct model-based material decomposition (MBMD)to reconstruct basis material density images directly from the SSF CT data. To evaluate different possible SSF designs, we define a new Fisher-information-based predictive image quality metric called separability index which characterizes the ability of a spectral CT system to distinguish between the signals from two or more materials. This predictive metric is used to define a system design optimization framework. We have applied this framework to find optimized combinations of filter materials, filter tile widths, and source settings for SSF CT. We conducted simulation-based design optimization study and separability-optimized filter designs are presented for water/iodine imaging and water/iodine/gadolinium/gold imaging for different patient sizes. Finally, we present MBMD results using simulated SSFCT data using the optimized designs to demonstrate the ability to reconstruct basis material density images and to show the benefits of the optimized designs.
BibTeX
@misc{tivnan2020design,
  author = {Matthew Tivnan and Wenying Wang and Grace J. Gang and J. Webster Stayman},
  title = {{Design of Spatial-Spectral Filters for CT Material Decomposition}},
  year = {2020},
  eprint = {2010.07483},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2010.07483},
}
Figure