Skip to content
DraconiticResearch
Search Sources API
226,742 studies31 sourcesUpdated 36 min ago
Peer-reviewedCC BY 4.0Scientific Reports

The ORION framework for outlier-robust parameter estimation in diffusion kurtosis imaging

  1. 1imec - Vision Lab, Department of Physics, University of Antwerp, Antwerp, Belgium
  2. 2icometrix, Leuven, Belgium
  3. 3µNEURO Research Centre of Excellence, University of Antwerp, Antwerp, Belgium

Abstract

Diffusion-weighted images are susceptible to intensity outliers that degrade the accuracy and precision of diffusion tensor and diffusion kurtosis imaging (DKI) parameter estimates. In this work, we introduce ORION, a voxel-wise framework for outlier-robust DKI parameter estimation based on constrained linear least squares estimators. ORION is designed to detect intensity outliers in both diffusion-weighted and non-diffusion-weighted images, ensuring the robustness of DKI metrics against artifactual signal dropouts and signal overshoots. First, the performance of the proposed framework in terms of outlier detection and statistical parameter estimation is compared to that of the previously proposed iterative reweighted linear least squares (IRLLS) outlier-robust approach in a Monte Carlo simulation framework. Next, we compare the DKI metrics extracted from two synthetic outlier-corrupted datasets using ORION, IRLLS, and a conventional weighted linear least squares (WLLS) approach. The synthetic datasets were realistically generated from in vivo MRI data from two participants in the Alzheimer’s Disease Neuroimaging Initiative (ADNI), with one dataset affected by signal dropout and the other by signal overshoot. In comparison to WLLS, both ORION and IRLLS achieved significantly lower errors in estimating white matter DKI metrics in regions affected by intensity outliers, with ORION also outperforming IRLLS on a subset of metrics under signal dropout conditions. Finally, the DKI metrics extracted from the in vivo datasets were found to be in line with the results obtained on the synthetic data. In conclusion, ORION successfully detects and rejects intensity outliers improving upon conventional DKI estimation techniques.

Related

  1. Parameter Estimation for the Intravoxel Incoherent Motion-diffusional Kurtosis Imaging Model Using Synthetic Q-space Learning Toward Breast Tumor Diagnosis
    Kousei Konya, Yuki Ichinoseki, Erina Kato +5
    Magnetic Resonance in Medical SciencesPeer-reviewedCC BYFull textMedical imaging & radiologyJ-STAGE
  2. In Vivo Tensor‐Valued Diffusion Magnetic Resonance Imaging Reveals Isotropic‐Anisotropic Kurtosis Mismatch in a Rat Middle Cerebral Artery Occlusion Stroke Model
    Mingyao Liang, Jianyu Yuan, Chaogang Tang +8
    Journal of the American Heart Association: Cardiovascular and Cerebrovascular DiseasePeer-reviewedCC BY-NC 4.0Full textMedical imaging & radiologyPMC
  3. Predictive performance of diffusion kurtosis imaging combined with magnetic resonance imaging morphological features for MGMT promoter status in glioma.
    Yin Gao, Zhenyu Song, Ping Wang +1
    Translational cancer researchPeer-reviewedCC BY-NC-NDFull textMedical imaging & radiologyEurope PMCPMC
  4. HADD: A hierarchy-aware disentangled diffusion framework with spatio-temporal denoising for 3D human pose estimation
    Jinze Jiang, Zhongheng Jian, Hualing Zheng +3
    PLOS OnePeer-reviewedCC BY 4.0Full textPhysiologyPLOS
  5. Bayesian Parameter Estimation of Cannibalism Dynamics in Larval Atractosteus tropicus Under Different Feeding Conditions
    Juan Carlos González-Aguirre, María Fernanda Jiménez, Gamaliel Blé +3
    MathematicsPeer-reviewedCC BY 4.0Epidemiology & public healthCrossref
BibTeX
@article{anania2026orion,
  title = {{The ORION framework for outlier-robust parameter estimation in diffusion kurtosis imaging}},
  author = {Vincenzo Anania and Ben Jeurissen and Thibo Billiet and Jan Sijbers and Arnold J. den Dekker},
  journal = {Scientific Reports},
  year = {2026},
  month = aug,
  doi = {10.1038/s41598-026-62390-w},
  url = {https://doi.org/10.1038/s41598-026-62390-w},
  note = {CC BY 4.0},
}
RIS
TY  - JOUR
TI  - The ORION framework for outlier-robust parameter estimation in diffusion kurtosis imaging
AU  - Vincenzo Anania
AU  - Ben Jeurissen
AU  - Thibo Billiet
AU  - Jan Sijbers
AU  - Arnold J. den Dekker
PY  - 2026
DA  - 2026/08/13
JO  - Scientific Reports
DO  - 10.1038/s41598-026-62390-w
UR  - https://doi.org/10.1038/s41598-026-62390-w
AB  - Diffusion-weighted images are susceptible to intensity outliers that degrade the accuracy and precision of diffusion tensor and diffusion kurtosis imaging (DKI) parameter estimates. In this work, we introduce ORION, a voxel-wise framework for outlier-robust DKI parameter estimation based on constrained linear least squares estimators. ORION is designed to detect intensity outliers in both diffusion-weighted and non-diffusion-weighted images, ensuring the robustness of DKI metrics against artifactual signal dropouts and signal overshoots. First, the performance of the proposed framework in terms of outlier detection and statistical parameter estimation is compared to that of the previously proposed iterative reweighted linear least squares (IRLLS) outlier-robust approach in a Monte Carlo simulation framework. Next, we compare the DKI metrics extracted from two synthetic outlier-corrupted datasets using ORION, IRLLS, and a conventional weighted linear least squares (WLLS) approach. The synthetic datasets were realistically generated from in vivo MRI data from two participants in the Alzheimer’s Disease Neuroimaging Initiative (ADNI), with one dataset affected by signal dropout and the other by signal overshoot. In comparison to WLLS, both ORION and IRLLS achieved significantly lower errors in estimating white matter DKI metrics in regions affected by intensity outliers, with ORION also outperforming IRLLS on a subset of metrics under signal dropout conditions. Finally, the DKI metrics extracted from the in vivo datasets were found to be in line with the results obtained on the synthetic data. In conclusion, ORION successfully detects and rejects intensity outliers improving upon conventional DKI estimation techniques.
SN  - 2045-2322
ER  -