Feasibility of using software-aided selection of virtual monoenergetic level for optimal image quality of acute necrotising pancreatitis based on dual-energy computed tomography: a preliminary study

Yuan Yuan, Kai Liao, Zixing Huang, Liping Deng, Hehan Tang, Yi Wang, Zheng Ye, Xinyue Chen, Bin Song, Zhenlin LiBMC Medical Imaging 2023 · DOI

TL;DR

This study demonstrates that software-guided selection of a virtual monoenergetic level in dual-energy CT scans significantly improves image quality for diagnosing acute necrotizing pancreatitis (ANP). By optimizing this level, researchers achieved higher contrast-to-noise ratios and signal-to-noise ratios, while simultaneously reducing image noise, leading to better visualization of pancreatic necrosis.

What Problem Does This Paper Address?

Acute necrotizing pancreatitis (ANP) is a severe form of pancreatitis that can lead to significant complications and mortality. Accurate and timely diagnosis is crucial for effective management. Contrast-enhanced computed tomography (CECT) is a primary imaging modality for ANP, but image quality can be affected by various factors. Dual-energy CT (DECT) offers the potential to generate virtual monoenergetic images (VMIs) at different energy levels, which can enhance contrast and reduce noise. However, there's no established consensus on the optimal monoenergetic level for visualizing ANP, and operator variability in selecting this level can be a challenge. This paper addresses the need for a standardized and optimized approach to DECT imaging for ANP by investigating the feasibility of software-aided selection of the virtual monoenergetic level for optimal image quality.

Methods

The researchers retrospectively analyzed contrast-enhanced DECT scans from 48 patients diagnosed with ANP. They employed a dual-source DECT system and acquired images in the arterial and portal venous phases. For the portal venous phase, they generated four sets of images:

  1. 100 kVp
  2. Sn140 kVp (140 kVp with a tin filter)
  3. Weighted-average 120 kVp
  4. An "optimal single-energy level for CNR" identified using software.

This software analyzed the iodine contrast-to-noise ratio (CNR) across a spectrum of virtual monoenergetic levels (40-200 keV) to pinpoint the energy level that maximized CNR for pancreatic parenchyma-to-necrosis.

Image quality was assessed using both objective and subjective measures:

  • Objective Measures:
    • Contrast-to-noise ratio (CNR) of pancreatic parenchyma-to-necrosis.
  • Signal-to-noise ratio (SNR) of the pancreas.
  • Image noise.
  • Subjective Measure:
    • A scoring system (1-4) by two experienced radiologists to evaluate the clarity of pancreatic necrosis and diagnostic information.

Statistical analysis involved repeated measures ANOVA and the Student-Newman-Keuls (SNK) test for objective measures, and a non-parametric Friedman rank sum test for subjective scores.

Datasets

The study utilized a dataset of 48 patients diagnosed with acute necrotizing pancreatitis who underwent abdominal contrast-enhanced dual-source DECT examinations at the West China Hospital, Sichuan University, between March 2015 and January 2016. This dataset was used for retrospective analysis. No new benchmarks were introduced; the study evaluated existing DECT data.

Key Insights and Ideas

  • The "Sweet Spot" for ANP Imaging: The core insight is that a specific virtual monoenergetic level, identified by software to maximize CNR, provides superior image quality for ANP compared to standard DECT reconstructions. This suggests that a "one-size-fits-all" approach to monoenergetic levels is suboptimal.
  • Software as a Diagnostic Aid: The study highlights the potential of software-aided selection to standardize DECT image optimization, reducing operator-dependent variability and improving diagnostic confidence.
  • Balancing Contrast and Noise: The optimal monoenergetic level not only boosts the contrast between healthy pancreatic tissue and necrotic areas but also effectively suppresses image noise, a critical factor for clear visualization.
  • Clinical Relevance: Improved image quality directly translates to better visualization of pancreatic necrosis, which is essential for accurate diagnosis, staging, and guiding treatment decisions in ANP.

Key Results

The optimal single-energy level for CNR images demonstrated statistically significant improvements across all measured parameters compared to the 100 kV, Sn140 kV, and weighted-average 120 kV images (P<0.05):

  • CNR of pancreatic parenchyma-to-necrosis: The optimal level achieved the highest CNR (9.99±5.86), significantly outperforming the other groups.
  • SNR of the pancreas: Similarly, the optimal level yielded the highest SNR (11.83±6.30).
  • Image noise: The optimal level resulted in the lowest image noise (9.31±2.96).
  • Subjective diagnosis score: Radiologists assigned the highest scores to the optimal single-energy images (3.88±0.33), indicating superior diagnostic clarity.

The optimal single-energy level for CNR was found to be around 75.04±1.24 keV, ranging from 73 to 78 keV.

Limitations and Future Work

The authors acknowledge several limitations:

  • Small Sample Size: The study included 48 patients, which they suggest is a limitation. They recommend larger trials to increase confidence in the findings.
  • Limited Energy Spectrum Analysis: The analysis was restricted to specific energy levels and did not explore the entire 40-190 keV spectrum. Future work should involve a more comprehensive comparison of single-energy spectrum images across a wider range of keV values.

The authors suggest that future research should focus on expanding the sample size and conducting more extensive analyses of the energy spectrum to further refine optimal imaging protocols for ANP.

Summary updated

BibTeX
@article{yuan2023feasibility,
  author = {Yuan Yuan and Kai Liao and Zixing Huang and Liping Deng and Hehan Tang and Yi Wang and Zheng Ye and Xinyue Chen and Bin Song and Zhenlin Li},
  title = {{Feasibility of using software-aided selection of virtual monoenergetic level for optimal image quality of acute necrotising pancreatitis based on dual-energy computed tomography: a preliminary study}},
  journal = {BMC Medical Imaging},
  year = {2023},
  doi = {10.1186/s12880-023-01032-3},
}
Figure