New Developments in QSM II
Digital Poster
Contrast Mechanisms
Wednesday, 13 May 2026
This session covers recent advances in quantitative susceptibility mapping.
|
560-02-001.
synthBFR: Mapping Whole Brain Total Field to Local Tissue Field Using Deep Learning and Realistic Model-Based Synthetic Data
Impact: Background field removal by synthBFR outperformed other deep learning and traditional methods in simulation and resulted in less artifacts near the air-tissue interfaces in QSM, improving diagnostic image quality in vivo.
|
||
|
560-02-002.
Turning Routine MRI into Pseudo Training Data: Scalable Physics-Informed Learning for Quantitative Susceptibility Mapping
Impact: By transforming abundant T1-weighted data into pseudo-susceptibility
maps, this work introduces a scalable strategy addressing data scarcity in
quantitative MRI. Routine clinical images thereby become effective training
priors, broadly enabling physics-informed and generalizable deep learning
models for quantitative MRI reconstruction.
|
||
|
560-02-003.
A Novel Numerical Phantom for Simultaneous QSM and EPT: Design and Evaluation
Impact: We developed a
dual-purpose numerical phantom enabling, for the first time, simultaneous evaluation
of QSM and EPT. This is a valuable resource for optimising both techniques within
a single framework, to facilitate future combined susceptibility and conductivity
mapping.
|
||
|
560-02-004.
Clinical Implementation of On-Console Quantitative Susceptibility Mapping for Brain MRI
Impact: This work demonstrates a fully integrated,
GPU-accelerated QSM pipeline that enables near–real-time susceptibility mapping
directly on the MR console, facilitating routine clinical use for patients with
demyelinating, hemorrhagic, and inflammatory brain disorders.
|
||
|
560-02-005.
Deep Learning QSM using xQSM with Squeeze-and-Excitation Networks in the Head and Neck
Impact: Using squeeze-and-excitation networks improves deep learning
QSM reconstruction in data sets where ground truth data are unavailable for
training, as we have shown in the head and neck.
|
||
|
560-02-006.
Microplastics in MR imaging
Impact: NMP aggregates can be visualised in vivo with susceptibility-weighted MRI. The related image inhomogeneities are suggested to appear as microbleed-mimicing hypointensities. Non-invasive in vivo detection of NMPs may offer new ways to study their potential detrimental effects on health.
|
||
|
560-02-007.
A retrospective study on the effect of exogenous-gas challenge on brain QSM-, mqBOLD-SO2 and χ-separation in rats
Impact: Exogenous-gas challenges induce different
effects on R2*
and QSM measurements, stronger under hyperoxia, suggesting that
exogenous-gas effects need to be considered when applying models
combining multi-echo gradient-echo phase and magnitude.
|
||
|
560-02-008.
Quantifying Susceptibility of U-fibers with High-Resolution QSM at 3T
Impact: This study developed an automated pipeline for juxtacortical white matter segmentation, enabled reproducible quantification of susceptibility of U-fibers for future studies investigating U-fiber iron overload and demyelination and their association with cognitive impairment in neurodegeneration and aging.
|
||
|
560-02-009.
Biophysical contrast source of the Locus Coeruleus in ex-situ human brain at 7T: Correlation between QSM and EPR spectroscopy
Impact: By revealing the iron-related origin of LC biophysical contrast in qMRI, this study has the potential to enable the in vivo assessment of LC integrity and early alterations in neurodegeneration, such as Parkinson's or Alzheimer's disease.
|
||
|
560-02-010.
Conformal Prediction for Rigorous Uncertainty Guarantees in Deep Learning-based QSM
Impact: This
framework provides calibrated uncertainty maps for QSM, allowing clinicians to
distinguish reliable measurements from artifacts. This enhances diagnostic
confidence in neurological conditions, such as assessing the true extent of
hemorrhages or calcifications.
|
||
|
560-02-011.
Echo Sampling Matters: Impact on Quantitative Susceptibility Mapping Accuracy and Regional Variability in Multiple Sclerosis
Impact: This study clarifies how
echo sampling influences QSM reliability across brain regions, providing
practical guidance for optimizing acquisition protocols. These insights support
more consistent QSM quantification across research and clinical settings,
enabling broader application of susceptibility-based biomarkers in neurological
disorders.
|
||
|
560-02-012.
QSM-Derived Limbic Iron Accumulation Predicts Cognitive and Epileptic Outcomes in Encephalitis
Impact: QSM detects limbic iron accumulation reflecting inflammatory injury in encephalitis, correlating with cognitive and electrophysiologic outcomes. These findings position susceptibility mapping as a promising biomarker for prognosis and therapeutic monitoring across autoimmune and infectious encephalitic syndromes.
|
||
|
560-02-013.
Assessing Simple Neural Architectures for Total Field Estimation in Quantitative Susceptibility Mapping
Impact: By achieving precise total field estimation with minimal neural architectures, this study establishes a baseline for Machine Learning (ML) in QSM pre-processing, informs data representation strategies and motivates further exploration of ML approaches for total field estimation.
|
||
|
560-02-014.
Ex vivo QSM of whole human brain hemispheres at 160 µm resolution applied to frontotemporal lobar degeneration
Impact: We demonstrate an
optimized pulse sequence and processing pipeline for ex vivo whole-hemisphere Quantitative Susceptiblity Mapping (QSM). We recapitulate findings of iron-rich
pathology in frontotemporal lobar degeneration and suggest the feasibility of
using alternating bipolar readouts to maximize scan efficiency.
|
||
|
560-02-015.
Realistic and Prior-Guided Background Field Removal in QSM with Variable Body Coverage Simulation
Impact: This study
introduces a prior-guided deep learning method for background field removal in
QSM. By integrating anatomical priors and body-coverage simulations, it
improves reconstruction accuracy and enables more reliable brain susceptibility
mapping for clinical and research applications.
|
||
|
560-02-016.
ViewMotionQSM: A MATLAB Toolbox for Motion Simulation and Evaluation in Quantitative Susceptibility Mapping
Impact: This toolbox provides a reproducible way to
study how motion degrades QSM and investigate the robustness of different
reconstruction methods to motion artifacts. By generating paired motion-free
and motion-corrupted datasets, it facilitates developing and validating motion-correction
strategies without additional acquisition.
|