A Joint Processing Strategy for Image Quality Improvement in 3D Digital Subtraction Angiography

Xiaoxuan Zhang, Xiao Jiang, Matthew Tivnan, J. Webster Stayman, Grace J. GangarXiv 2023 · DOI · arXiv

Abstract

Three-dimensional digital subtraction angiography (3D-DSA) is a widely adopted technique for clinical evaluation of contrast-enhanced vasculatures. The distribution of a contrast agent such as iodine is often estimated via temporal subtraction. Advancements in spectral imaging technologies such as photon counting detectors offer new opportunities to improve DSA image quality. In this work, we propose a novel joint processing strategy to achieve an iodine image using two-bin spectral measurements from a photon counting detector acquired both the pre- and post-contrast injection. Simulation studies were performed using a digital phantom with iodine-enhanced vessels. The proposed method was compared with temporal subtraction and conventional spectral imaging using just the post-contrast measurements. Imaging performance was evaluated in terms of noise-resolution tradeoffs. Preliminary findings have shown measurably improved image quality given by joint processing, reducing noise by 40% and 70% compared to temporal subtraction and conventional spectral imaging, respectively.

Introduction

Three-dimensional digital subtraction angiography (3D-DSA) serves as the clinical gold standard for visualizing vascular structures, essential for diagnosing conditions like aneurysms. Traditionally, 3D-DSA relies on temporal subtraction, where a "mask" scan taken before contrast injection is subtracted from a "live" scan taken after. While effective, this method is susceptible to patient motion and significant image noise.

Spectral imaging, particularly with the advent of photon-counting detectors (PCDs), offers an alternative through material decomposition. By distinguishing X-ray energies, iodine can be isolated from a single post-contrast scan, potentially mitigating motion artifacts. However, conventional material decomposition often amplifies noise, making it difficult to resolve small vessels. This study proposes a "joint processing" strategy that leverages both temporal and spectral data to overcome these limitations.

Methodology

The proposed framework utilizes four measurements: two energy bins (spectral) acquired at two time points (pre- and post-contrast). The researchers explored two primary algorithmic approaches:

  1. Model-Based One-Step Decomposition (MB-OSD): A direct reconstruction of material densities from raw projection data.
  2. Model-Based Projection-Domain Decomposition (MB-PDD): A two-step process that estimates material line integrals before performing 3D reconstruction.

A key innovation is the inclusion of a masking matrix in the forward model, which effectively "zeros out" iodine in pre-contrast data, allowing for the simultaneous solution of all material components (water, calcium, iodine) across the combined data streams.

Simulation Setup

To evaluate the strategy, a digital simulation phantom was designed to mimic clinical conditions. It included a 1000 mg/mL water background, a 4.0 mm calcium ring to simulate bone, and iodine-enhanced vessels (10 mg/mL) with diameters ranging from 0.3 mm to 4.0 mm.

Digital simulation phantom featuring iodine-enhanced vessels of varying diameters and a calcium ring.

This phantom allowed the researchers to precisely measure noise-resolution tradeoffs and the ability of the system to resolve sub-millimeter structures that are often lost in conventional imaging.

Results and Discussion

The joint processing method (MB-PDD) demonstrated significant improvements in image quality. Quantitatively, it reduced noise by 40% compared to standard temporal subtraction (using energy-integrating detectors) and by 70% compared to conventional spectral imaging.

Noise-resolution tradeoff curves showing the superior performance of joint processing (MB-PDD) compared to conventional methods.

Key findings include:

  • Noise Suppression: By utilizing redundant background information from both temporal and spectral domains, the joint strategy cancels out noise that typically plagues single-domain approaches.
  • Resolution Gains: While traditional energy-integrating detector (EID) subtraction struggled below 0.45 mm, the PCD-based joint processing successfully resolved vessels as small as 0.3 mm.
  • Hardware Advantage: Even on identical PCD hardware, the joint strategy provided a ~12% noise reduction over pure temporal subtraction.

The study also noted that enforcing non-negativity in the MB-OSD approach acted as a powerful noise suppressor, though it introduced a "patchy" image texture that requires further optimization of regularization parameters.

Conclusion and Future Work

The proposed joint processing strategy significantly enhances 3D-DSA image quality by combining the strengths of temporal subtraction and spectral material decomposition. Future work will focus on integrating motion registration algorithms to handle tissue deformation in non-intubated patients and refining regularization parameters to improve image texture. Additionally, more realistic detector models incorporating charge sharing and pulse pileup will be used to validate these findings for clinical translation.

Summary updated

Abstract

Three-dimensional digital subtraction angiography (3D-DSA) is a widely adopted technique for clinical evaluation of contrast-enhanced vasculatures. The distribution of a contrast agent such as iodine is often estimated via temporal subtraction. Advancements in spectral imaging technologies such as photon counting detectors offer new opportunities to improve DSA image quality. In this work, we propose a novel joint processing strategy to achieve an iodine image using two-bin spectral measurements from a photon counting detector acquired both the pre- and post-contrast injection. Simulation studies were performed using a digital phantom with iodine-enhanced vessels. The proposed method was compared with temporal subtraction and conventional spectral imaging using just the post-contrast measurements. Imaging performance was evaluated in terms of noise-resolution tradeoffs. Preliminary findings have shown measurably improved image quality given by joint processing, reducing noise by 40% and 70% compared to temporal subtraction and conventional spectral imaging, respectively.
BibTeX
@misc{zhang2023joint,
  author = {Xiaoxuan Zhang and Xiao Jiang and Matthew Tivnan and J. Webster Stayman and Grace J. Gang},
  title = {{A Joint Processing Strategy for Image Quality Improvement in 3D Digital Subtraction Angiography}},
  year = {2023},
  eprint = {2310.10694},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2310.10694},
}

Loading related content...

Figure