Se-Hong Oh

Quantitative susceptibility mapping (QSM) measures the intrinsic magnetic susceptibility of tissues. Because susceptibility values are inherently relative rather than absolute, referencing to a region of interest (ROI) is commonly employed to mitigate intersubject and acquisition‐related variability. To validate the feasibility and robustness of a differential ROI reference method in QSM using both digital phantom and in vivo data, particularly for deep gray matter susceptibility analysis. In the digital phantom study, susceptibility values in the caudate nucleus were assessed with and without differential referencing across various simulation conditions (slab widths, brain mask erosion levels, air susceptibility). In the in vivo study, susceptibility differences between 16 patients with Parkinson's disease (PD) and 16 controls were compared across multiple deep gray matter ROIs under variable acquisition parameters. A one‐way ANOVA statistical test was used in the in vivo study with multiple‐comparison correction via the Benjamini–Hochberg false discovery rate (FDR‐BH), with significance set at p < 0.05. In this study, “accuracy” refers specifically to the phantom study where ground‐truth susceptibility values are known. In vivo, where such ground truth is unavailable, “robustness” and “group‐discrimination sensitivity” are used as surrogate indicators of accuracy. Robustness was evaluated by the consistency of susceptibility differences across varying acquisition parameters, whereas group‐discrimination sensitivity was evaluated by the ability to detect disease‐related group differences between PD patients and controls. In the digital phantom study, the proposed method demonstrated stable susceptibility estimates across all conditions, with minimal deviation from true values when referencing anatomically adjacent ROIs. In the in vivo study, referenced susceptibility values (e.g., substantia nigra minus putamen) yielded consistent and significant group differences across acquisition scenarios, whereas unreferenced values were more variable. The differential ROI reference method reduces susceptibility variability associated with acquisition and processing factors. This approach may offer a practical and pathology‐resilient referencing alternative for deep brain analysis in neurodegenerative diseases.

Measuring myelin concentration in the brain has important implications in basic science and clinical practice. In MRI, myelin water imaging (MWI) has been suggested as a surrogate biomarker that provides high sensitivity and specificity for myelin. However, multi‐exponential fitting is ill‐conditioned, and it is sensitive to noise and artifacts, particularly in vivo. To overcome the ill‐conditioned fitting problem, the two‐dimensional ViSTa myelin‐weighted imaging technique was proposed, and it provides a substantially improved myelin‐weighted image. However, it is based on a two‐dimensional single‐slice acquisition scheme, and it is a limitation. In this study, a whole brain‐covered 3D ViSTa sequence, based on a 3D segmented echo planar imaging (EPI) sequence with a pair of slice selective inversion RF pulses, was proposed. To investigate the 3D ViSTa myelin weighted image, the distribution of myelin content in the white matter of the brain was measured using both conventional MWI and ViSTa MWI. The proposed 3D ViSTa method achieves a whole brain‐covered (FOV = 240 × 240 × 128 mm³) myelin water‐weighted image in less than 8 min (1.5 × 1.5 × 4 mm³) and does not require heavy post‐processing. Pseudo‐quantification (apparent MWF) can be provided by normalizing the ViSTa image with a PD‐weighted image. The voxel‐wise correlation between the conventional MWI and the 3D ViSTa yielded a mean correlation coefficient of 0.74 ± 0.03 (mean ± standard deviation of the five subjects), demonstrating a high spatial similarity in myelin‐weighted contrast between the two maps. The proposed 3D ViSTa with pseudo‐quantification may be useful in clinical applications when absolute quantification is not necessary.

Quantitative susceptibility mapping (QSM) provides the spatial distribution of magnetic susceptibility within tissues through sequential steps: phase unwrapping and echo combination, mask generation, background field removal, and dipole inversion. Accurate mask generation is crucial, as masks excluding regions outside the brain and without holes are necessary to minimize errors and streaking artifacts during QSM reconstruction. Variations in susceptibility values can arise from different mask generation methods, highlighting the importance of optimizing mask creation. In this study, we propose QSMmask‐net, a deep neural network‐based method for generating precise QSM masks. QSMmask‐net achieved the highest Dice score compared to other mask generation methods. Mean susceptibility values using QSMmask‐net masks showed the lowest differences from manual masks (ground truth) in simulations and healthy controls (no significant difference, p > 0.05). Linear regression analysis confirmed a strong correlation with manual masks for hemorrhagic lesions (slope = 0.9814 ± 0.007, intercept = 0.0031 ± 0.001, R² = 0.9992, p < 0.05). We have demonstrated that mask generation methods can affect the susceptibility value estimations. QSMmask‐net reduces the labor required for mask generation while providing mask quality comparable to manual methods. The proposed method enables users without specialized expertise to create optimized masks, potentially broadening QSM applicability efficiently.

In magnetic resonance imaging (MRI), variations in scan parameters and scanner specifications can result in differences in image appearance. To minimize these differences, harmonization in MRI has been suggested as a crucial image processing technique. In this study, we developed an MR physics-based harmonization framework, Physics-Constrained Deep Neural Network for Multisite and multiscanner Harmonization (PhyCHarm). PhyCHarm includes two deep neural networks: (1) the Quantitative Maps Generator to generate T ₁ - and M ₀ -maps and (2) the Harmonization Network. We used an open dataset consisting of 3T MP2RAGE images from 50 healthy individuals for the Quantitative Maps Generator and a traveling dataset consisting of 3T T ₁ w images from 9 healthy individuals for the Harmonization Network. PhyCHarm was evaluated using the structural similarity index measure (SSIM), peak signal-to-noise ratio (PSNR), and normalized-root-mean square error (NRMSE) for the Quantitative Maps Generator, and using SSIM, PSNR, and volumetric analysis for the Harmonization network, respectively. PhyCHarm demonstrated increased SSIM and PSNR, the highest Dice score in the FSL FAST segmentation results for gray and white matter compared to U-Net, Pix2Pix, and CALAMITI. PhyCHarm showed a greater reduction in volume differences after harmonization for gray and white matter than U-Net, Pix2Pix, or CALAMITI. As an initial step toward developing advanced harmonization techniques, we investigated the applicability of physics-based constraints within a supervised training strategy. The proposed physics constraints could be integrated with unsupervised methods, paving the way for more sophisticated harmonization qualities.

Lee G, et al. Investig Magn Reson Imaging. 2025 Jun;29(2):100-108. https://doi.org/10.13104/imri.2025.0013

Background and Purpose: Slowly expanding lesions (SELs) are thought to represent a subset of chronic active lesions and have been associated with clinical disability, severity, and disease progression. The purpose of this study was to characterize SELs using advanced magnetic resonance imaging (MRI) measures related to myelin and neurite density on 7 Tesla (T) MRI.Methods: The study design was retrospective, longitudinal, observational cohort with multiple sclerosis (n = 15). Magnetom 7T scanner was used to acquire magnetization‐prepared 2 rapid acquisition gradient echo and advanced MRI including visualization of short transverse relaxation time component (ViSTa) for myelin, quantitative magnetization transfer (qMT) for myelin, and neurite orientation dispersion density imaging (NODDI). SELs were defined as lesions showing ≥12% of growth over 12 months on serial MRI. Comparisons of quantitative measures in SELs and non‐SELs were performed at baseline and over time. Statistical analyses included two‐sample t‐test, analysis of variance, and mixed‐effects linear model for MRI metrics between lesion types.Results: A total of 1075 lesions were evaluated. Two hundred twenty‐four lesions (21%) were SELs, and 216 (96%) of the SELs were black holes. At baseline, compared to non‐SELs, SELs showed significantly lower ViSTa (1.38 vs. 1.53, p < .001) and qMT (2.47 vs. 2.97, p < .001) but not in NODDI measures (p > .27). Longitudinally, only ViSTa showed a greater loss when comparing SEL and non‐SEL (p = .03).Conclusions: SELs have a lower myelin content relative to non‐SELs without a difference in neurite measures. SELs showed a longitudinal decrease in apparent myelin water fraction reflecting greater tissue injury.

In MRI, researchers have long endeavored to effectively visualize myelin distribution in the brain, a pursuit with significant implications for both scientific research and clinical applications. Over time, various methods such as myelin water imaging, magnetization transfer imaging, and relaxometric imaging have been developed, each carrying distinct advantages and limitations. Recently, an innovative technique named as magnetic susceptibility source separation has emerged, introducing a novel surrogate biomarker for myelin in the form of a diamagnetic susceptibility map. This paper comprehensively reviews this cutting-edge method, providing the fundamental concepts of magnetic susceptibility, susceptibility imaging, and the validation of the diamagnetic susceptibility map as a myelin biomarker that indirectly measures myelin content. Additionally, the paper explores essential aspects of data acquisition and processing, offering practical insights for readers. A comparison with established myelin imaging methods is also presented, and both current and prospective clinical and scientific applications are discussed to provide a holistic understanding of the technique. This work aims to serve as a foundational resource for newcomers entering this dynamic and rapidly expanding field.

MRI is an imaging technology that non-invasively obtains high-quality medical images for diagnosis. However, MRI has the major disadvantage of long scan times which cause patient discomfort and image artifacts. As one of the methods for reducing the long scan time of MRI, the parallel MRI method for reconstructing a high-fidelity MR image from under-sampled multi-coil k-space data is widely used. In this study, we propose a method to reconstruct a high-fidelity MR image from under-sampled multi-coil k-space data using deep-learning. The proposed multi-domain Neumann network with sensitivity maps (MDNNSM) is based on the Neumann network and uses a forward model including coil sensitivity maps for parallel MRI reconstruction. The MDNNSM consists of three main structures: the CNN-based sensitivity reconstruction block estimates coil sensitivity maps from multi-coil under-sampled k-space data; the recursive MR image reconstruction block reconstructs the MR image; and the skip connection accumulates each output and produces the final result. Experiments using the fastMRI T1-weighted brain image dataset were conducted at acceleration factors of 2, 4, and 8. Qualitative and quantitative experimental results show that the proposed MDNNSM method reconstructs MR images more accurately than other methods, including the generalized autocalibrating partially parallel acquisitions (GRAPPA) method and the original Neumann network.