Options
Deep learning for restoring MPI system matrices using simulated training data
Citation Link: https://doi.org/10.15480/882.17185
Publikationstyp
Journal Article
Date Issued
2026-05-13
Sprache
English
TORE-DOI
Journal
Volume
71
Issue
9
Article Number
095029
Citation
Physics in Medicine & Biology 71 (9): 095029 (2026)
Publisher DOI
Scopus ID
Publisher
IOP Publishing
Peer Reviewed
true
Is Supplemented By
Is New Version of
Objective. Magnetic particle imaging reconstructs tracer distributions using a system matrix (SM) obtained through time-consuming, noise-prone calibration measurements. Methods for addressing imperfections in measured system matrices increasingly rely on deep neural networks, yet curated training data remain scarce. This study evaluates whether physics-based simulated system matrices can be used to train deep learning (DL) models for different SM restoration tasks, i.e. denoising, accelerated calibration, upsampling, and inpainting, that generalize to measured data. Approach. A large dataset of system matrices was generated using an equilibrium magnetization model extended with uniaxial anisotropy. The dataset spans particle, scanner, and calibration parameters for 2D and 3D trajectories, and includes background noise injected from empty-frame measurements. For each restoration task, DL models were compared with classical non-learning baseline methods. Quantitative performance was evaluated on simulated data using peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM). For measured data, performance was assessed qualitatively by visual comparison of system matrices and the resulting reconstructions. Main results. The models trained solely on simulated system matrices generalized to measured data across all tasks: for denoising, DnCNN/RDN/SwinIR outperformed discrete cosine transform and soft thresholding baseline by
dB PSNR and up to +0.1 SSIM on simulations and led to perceptually better reconstructions of real data; for 2D upsampling, SMRnet exceeded bicubic by ∼ 20 dB PSNR and ∼ 0.08 SSIM at ×2–×4 but these gains did not transfer qualitatively to real measurements. For 3D accelerated calibration, SMRnet matched tricubic in noiseless cases and was more robust under noise, and for 3D inpainting, biharmonic inpainting was superior when noise-free but degraded with noise, while a PConvUNet maintained quality and yielded less blurry reconstructions. Significance. The demonstrated transferability of DL models trained on simulations to real measurements mitigates the data-scarcity problem, which intensifies with model scale. This enables the development of new methods beyond current measurement capabilities and supports pre-training of large models on simulated data.
dB PSNR and up to +0.1 SSIM on simulations and led to perceptually better reconstructions of real data; for 2D upsampling, SMRnet exceeded bicubic by ∼ 20 dB PSNR and ∼ 0.08 SSIM at ×2–×4 but these gains did not transfer qualitatively to real measurements. For 3D accelerated calibration, SMRnet matched tricubic in noiseless cases and was more robust under noise, and for 3D inpainting, biharmonic inpainting was superior when noise-free but degraded with noise, while a PConvUNet maintained quality and yielded less blurry reconstructions. Significance. The demonstrated transferability of DL models trained on simulations to real measurements mitigates the data-scarcity problem, which intensifies with model scale. This enables the development of new methods beyond current measurement capabilities and supports pre-training of large models on simulated data.
Subjects
magnetic particle imaging
system matrix recovery
machine learning
image restoration
DDC Class
621: Applied Physics
Publication version
publishedVersion
Loading...
Name
Tsanda_2026_Phys._Med._Biol._71_095029.pdf
Type
Main Article
Size
3.22 MB
Format
Adobe PDF