Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Improving Virtual Contrast Enhancement using Longitudinal Data

Pierre Fayolle, Alexandre Bône, Noëlie Debs, Philippe Robert, Pascal Bourdon, Remy Guillevin, David Helbert

TL;DR

This work tackles the safety concerns of gadolinium-based contrast agents by proposing a longitudinal deep learning framework to virtually enhance low-dose post-contrast MRI into full-dose images. It leverages prior full-dose scans from the same patient, using a conditional GAN to generate a synthetic low-dose input and a 3D V-Net to reconstruct the full-dose image, demonstrating superior fidelity over a non-longitudinal approach. On the ACRIN-DSC-MR-Brain dataset, the longitudinal model shows statistically significant improvements in PSNR and SSIM (and a trend in MSE) at $25\%$ dose, with robustness across simulated dose levels. While promising for reducing GBCA exposure in longitudinal neuro-oncology MRI, the approach relies on synthetic low-dose data and a relatively small, single-dataset test set, warranting validation with real low-dose acquisitions and potential multi-modal extensions in the future.

Abstract

Gadolinium-based contrast agents (GBCAs) are widely used in magnetic resonance imaging (MRI) to enhance lesion detection and characterisation, particularly in the field of neuro-oncology. Nevertheless, concerns regarding gadolinium retention and accumulation in brain and body tissues, most notably for diseases that require close monitoring and frequent GBCA injection, have led to the need for strategies to reduce dosage. In this study, a deep learning framework is proposed for the virtual contrast enhancement of full-dose post-contrast T1-weighted MRI images from corresponding low-dose acquisitions. The contribution of the presented model is its utilisation of longitudinal information, which is achieved by incorporating a prior full-dose MRI examination from the same patient. A comparative evaluation against a non-longitudinal single session model demonstrated that the longitudinal approach significantly improves image quality across multiple reconstruction metrics. Furthermore, experiments with varying simulated contrast doses confirmed the robustness of the proposed method. These results emphasize the potential of integrating prior imaging history into deep learning-based virtual contrast enhancement pipelines to reduce GBCA usage without compromising diagnostic utility, thus paving the way for safer, more sustainable longitudinal monitoring in clinical MRI practice.

Improving Virtual Contrast Enhancement using Longitudinal Data

TL;DR

This work tackles the safety concerns of gadolinium-based contrast agents by proposing a longitudinal deep learning framework to virtually enhance low-dose post-contrast MRI into full-dose images. It leverages prior full-dose scans from the same patient, using a conditional GAN to generate a synthetic low-dose input and a 3D V-Net to reconstruct the full-dose image, demonstrating superior fidelity over a non-longitudinal approach. On the ACRIN-DSC-MR-Brain dataset, the longitudinal model shows statistically significant improvements in PSNR and SSIM (and a trend in MSE) at dose, with robustness across simulated dose levels. While promising for reducing GBCA exposure in longitudinal neuro-oncology MRI, the approach relies on synthetic low-dose data and a relatively small, single-dataset test set, warranting validation with real low-dose acquisitions and potential multi-modal extensions in the future.

Abstract

Gadolinium-based contrast agents (GBCAs) are widely used in magnetic resonance imaging (MRI) to enhance lesion detection and characterisation, particularly in the field of neuro-oncology. Nevertheless, concerns regarding gadolinium retention and accumulation in brain and body tissues, most notably for diseases that require close monitoring and frequent GBCA injection, have led to the need for strategies to reduce dosage. In this study, a deep learning framework is proposed for the virtual contrast enhancement of full-dose post-contrast T1-weighted MRI images from corresponding low-dose acquisitions. The contribution of the presented model is its utilisation of longitudinal information, which is achieved by incorporating a prior full-dose MRI examination from the same patient. A comparative evaluation against a non-longitudinal single session model demonstrated that the longitudinal approach significantly improves image quality across multiple reconstruction metrics. Furthermore, experiments with varying simulated contrast doses confirmed the robustness of the proposed method. These results emphasize the potential of integrating prior imaging history into deep learning-based virtual contrast enhancement pipelines to reduce GBCA usage without compromising diagnostic utility, thus paving the way for safer, more sustainable longitudinal monitoring in clinical MRI practice.

Paper Structure

This paper contains 15 sections, 4 figures, 1 table.

Table of Contents

  1. Introduction
  2. Materials and Methods
  3. Longitudinal Virtual Contrast Enhancement Workflow
  4. Public dataset
  5. Data Preprocessing
  6. Training Strategy
  7. Model Evaluation
  8. Results
  9. Comparative Quantitative Analysis
  10. Qualitative comparison
  11. Dose Variation Evaluation
  12. Discussions
  13. Conclusion
  14. Acknowledgments.
  15. Disclosure of Interests.

Figures (4)

  • Figure 1: Workflow of the Proposed Longitudinal Virtual Contrast Enhancement Model for T1-Weighted Images. Dotted arrow shows the main contribution of the proposed work. cGAN: conditional Generative Adversarial Networks, MSE: Mean Square Error.
  • Figure 2: Boxplot Analysis of Metrics Between the Single Session and Longitudinal Models. MSE: Mean Square Error, PSNR: Peak Signal-to-Noise Ratio, SSIM: Structural Similarity Index.
  • Figure 3: Comparison Between Single Session and Longitudinal Data-Driven Models. PC: Pre-contrast, SD: Standard-dose, LD: Low-dose (25% synthetized dose).
  • Figure 4: Effect of Simulated Dose Levels on the Virtual Contrast Enhancement Metrics. Dotted lines represent linear regression curves fitted across dose levels for each model. Error bars denote standard deviation.