Qiuyun Fan

Tensor‐valued diffusion MRI enables disentangling microscopic diffusion anisotropy and isotropic heterogeneity through models such as diffusional variance decomposition (DIVIDE) and diffusion tensor distributions (DTD). However, these models require dense sampling across multiple b ‐values and b‐tensor shapes, making them highly sensitive to noise and acquisition time constraints. In this work, we propose a model‐aware subspace reconstruction framework that integrates DTD‐based DIVIDE modeling with low‐rank signal representations. By exploiting the intrinsic low‐dimensional structure of DIVIDE signals in the diffusion encoding domain, we construct a physics‐consistent subspace that enables joint reconstruction and parameter estimation from highly undersampled data. Simulation and in vivo experiments demonstrate that the proposed approach substantially improves the stability and precision of DIVIDE‐derived microstructural parameters compared with conventional reconstruction methods, while enabling significant acceleration. These results suggest that subspace‐constrained reconstruction provides a principled pathway toward efficient and robust multidimensional diffusion MRI.

Brain development in preterm infants shows marked heterogeneity, often obscured by group‐level analyses. Between the group‐level and the individual difference, subgroup can model the heterogeneity of early developmental trajectories. To characterize this, we analyzed longitudinal functional connectome data from 90 preterm infants (scanned at birth and term‐equivalent age) and 521 full‐term controls from the developing Human Connectome Project. A machine learning model predicted individual brain‐age gap (BAG), quantifying maturational deviation. Clustering of longitudinal BAG trajectories revealed two distinct preterm subgroups with divergent developmental pathways. These subgroups exhibited significantly different functional network architectures and, at 19‐month follow‐up, distinct behavioral outcomes in cognitive, language, and motor domains. Our findings establish that early preterm brain maturation follows identifiable, heterogeneous trajectories, providing a data‐driven framework for early risk stratification.

Models based on magnetic resonance imaging (MRI) and contrast‐enhanced mammography (CEM) radiomics for noninvasive prediction of nonsentinel lymph node (NSLN) metastasis were developed in this study. A total of 270 patients who had undergone axillary lymph node dissection from January 2018 to October 2022 and were diagnosed with 1–2 positive SLNs were included. Radiomic features were extracted from breast cancer voxels in CEM, T1 weighted MRI, T2 weighted MRI, and Diffision Weighted Imaging (DWI) images and kinetic parameter maps of dynamic contrast‐enhanced‐MRI. Correlation analysis, the least absolute shrinkage and selection operator (LASSO), and analysis of variance (ANOVA) were utilized for feature selection. The expectation maximum based matrix completion technique filled in the missing features in a specific modality depending on the features of another available modality. Finally, MRI‐CEM‐based prediction models were developed utilizing a support vector machine. The receiver operating characteristic curve was used to assess the models' performance. The mean AUCs from three‐fold cross‐validation for models based on clinical risk factors, MRI‐CEM radiomic features, and their combination were 0.70, 0.70, and 0.76, respectively. Combining MRI and CEM radiomic features with clinical risk factors improved the predictive ability for NSLN metastasis compared to individual factors alone.

Diffusion‐weighted magnetic resonance imaging (DW‐MRI) has evolved to provide increasingly sophisticated investigations of the human brain's structural connectome in vivo. Restriction spectrum imaging (RSI) is a method that reconstructs the orientation distribution of diffusion within tissues over a range of length scales. In its original formulation, RSI represented the signal as consisting of a spectrum of Gaussian diffusion response functions. Recent technological advances have enabled the use of ultra‐high b‐values on human MRI scanners, providing higher sensitivity to intracellular water diffusion in the living human brain. To capture the complex diffusion time dependence of the signal within restricted water compartments, we expand upon the RSI approach to represent restricted water compartments with non‐Gaussian response functions, in an extended analysis framework called linear multi‐scale modeling (LMM). The LMM approach is designed to resolve length scale and orientation‐specific information with greater specificity to tissue microstructure in the restricted and hindered compartments, while retaining the advantages of the RSI approach in its implementation as a linear inverse problem. Using multi‐shell, multi‐diffusion time DW‐MRI data acquired with a state‐of‐the‐art 3 T MRI scanner equipped with 300 mT/m gradients, we demonstrate the ability of the LMM approach to distinguish different anatomical structures in the human brain and the potential to advance mapping of the human connectome through joint estimation of the fiber orientation distributions and compartment size characteristics.

Diffusion MRI is widely used for the clinical examination of a variety of diseases of the nervous system. However, clinical MRI scanners are mostly capable of magnetic field gradients in the range of 20–80 mT/m and are thus limited in the detection of small tissue structures such as determining axon diameters. The availability of high gradient systems such as the Connectome MRI scanner with gradient strengths up to 300 mT/m enables quantification of the reduction of the apparent diffusion coefficient and thus resolution of a wider range of diffusion coefficients. In addition, biological tissues are heterogenous on many scales and the complexity of tissue microstructure may not be accurately captured by models based on pre-existing assumptions. Thus, it is important to analyze the diffusion distribution without prior assumptions of the underlying diffusion components and their symmetries. In this paper, we outline a framework for analyzing diffusion MRI data with b-values up to 17,800 s/mm² to obtain a Full Diffusion Tensor Distribution (FDTD) with a wide variety of diffusion tensor structures and without prior assumption of the form of the distribution, and test it on a healthy subject. We then apply this method and use a machine learning method based on K-means classification to identify features in FDTD to visualize and characterize tissue heterogeneity in two subjects with diffuse gliomas.

BACKGROUND AND PURPOSE: Our aim was to evaluate an ultrafast 3D-FLAIR sequence using Wave–controlled aliasing in parallel imaging encoding (Wave-FLAIR) compared with standard 3D-FLAIR in the visualization and volumetric estimation of cerebral white matter lesions in a clinical setting. MATERIALS AND METHODS: Forty-two consecutive patients underwent 3T brain MR imaging, including standard 3D-FLAIR (acceleration factor = 2, scan time = 7 minutes 50 seconds) and resolution-matched ultrafast Wave-FLAIR sequences (acceleration factor = 6, scan time = 2 minutes 45 seconds for the 20-channel coil; acceleration factor = 9, scan time = 1 minute 50 seconds for the 32-channel coil) as part of clinical evaluation for demyelinating disease. Automated segmentation of cerebral white matter lesions was performed using the Lesion Segmentation Tool in SPM. Student t tests, intraclass correlation coefficients, relative lesion volume difference, and Dice similarity coefficients were used to compare volumetric measurements among sequences. Two blinded neuroradiologists evaluated the visualization of white matter lesions, artifacts, and overall diagnostic quality using a predefined 5-point scale. RESULTS: Standard and Wave-FLAIR sequences showed excellent agreement of lesion volumes with an intraclass correlation coefficient of 0.99 and mean Dice similarity coefficient of 0.97 (SD, 0.05) (range, 0.84–0.99). Wave-FLAIR was noninferior to standard FLAIR for visualization of lesions and motion. The diagnostic quality for Wave-FLAIR was slightly greater than for standard FLAIR for infratentorial lesions (P < .001), and there were fewer pulsation artifacts on Wave-FLAIR compared with standard FLAIR (P < .001). CONCLUSIONS: Ultrafast Wave-FLAIR provides superior visualization of infratentorial lesions while preserving overall diagnostic quality and yields white matter lesion volumes comparable with those estimated using standard FLAIR. The availability of ultrafast Wave-FLAIR may facilitate the greater use of 3D-FLAIR sequences in the evaluation of patients with suspected demyelinating disease.

High-resolution diffusion tensor imaging (DTI) is beneficial for probing tissue microstructure in fine neuroanatomical structures, but long scan times and limited signal-to-noise ratio pose significant barriers to acquiring DTI at sub-millimeter resolution. To address this challenge, we propose a deep learning-based super-resolution method entitled "SRDTI" to synthesize high-resolution diffusion-weighted images (DWIs) from low-resolution DWIs. SRDTI employs a deep convolutional neural network (CNN), residual learning and multi-contrast imaging, and generates high-quality results with rich textural details and microstructural information, which are more similar to high-resolution ground truth than those from trilinear and cubic spline interpolation.

Accurate and automated reconstruction of the in vivo human cerebral cortical surface from anatomical magnetic resonance (MR) images facilitates the quantitative analysis of cortical structure. Anatomical MR images with sub-millimeter isotropic spatial resolution improve the accuracy of cortical surface and thickness estimation compared to the standard 1-millimeter isotropic resolution. Nonetheless, sub-millimeter resolution acquisitions require averaging multiple repetitions to achieve sufficient signal-to-noise ratio and are therefore long and potentially vulnerable to subject motion. We address this challenge by synthesizing sub-millimeter resolution images from standard 1-millimeter isotropic resolution images using a data-driven supervised machine learning-based super-resolution approach achieved via a deep convolutional neural network. We systematically characterize our approach using a large-scale simulated dataset and demonstrate its efficacy in empirical data. The super-resolution data provide improved cortical surfaces similar to those obtained from native sub-millimeter resolution data. The whole-brain mean absolute discrepancy in cortical surface positioning and thickness estimation is below 100 μm at the single-subject level and below 50 μm at the group level for the simulated data, and below 200 μm at the single-subject level and below 100 μm at the group level for the empirical data, making the accuracy of cortical surfaces derived from super-resolution sufficient for most applications.

BACKGROUND AND PURPOSE: Remyelination represents an area of great therapeutic interest in multiple sclerosis but currently lacks a robust imaging marker. The purpose of this study was to use high-gradient diffusion MRI and macromolecular tissue volume imaging to obtain estimates of axonal volume fraction, myelin volume fraction, and the imaging g-ratio in patients with MS and healthy controls and to explore their relationship to neurologic disability in MS. MATERIALS AND METHODS: Thirty individuals with MS (23 relapsing-remitting MS, 7 progressive MS) and 19 age-matched healthy controls were scanned on a 3T MRI scanner equipped with 300 mT/m maximum gradient strength using a comprehensive multishell diffusion MRI protocol. Macromolecular tissue volume imaging was performed to quantify the myelin volume fraction. Diffusion data were fitted to a 3-compartment model of white matter using a spheric mean approach to yield estimates of axonal volume fraction. The imaging g-ratio was calculated from the ratio of myelin volume fraction and axonal volume fraction. Imaging metrics were compared between groups using 2-sided t tests with a Bonferroni correction. RESULTS: The mean g-ratio was significantly elevated in lesions compared with normal-appearing WM (0.74 vs 0.67, P < .001). Axonal volume fraction (0.17 vs 0.23, P < .001) and myelin volume fraction (0.17 vs 0.25, P < .001) were significantly lower in lesions than normal-appearing WM. Myelin volume fraction was lower in normal-appearing WM compared with that in healthy controls (0.25 vs 0.27, P = .009). Disability, as measured by the Expanded Disability Status Scale, was significantly associated with myelin volume fraction (β = –40.5, P = .001) and axonal volume fraction (β = –41.0, P = .016) in normal-appearing WM. CONCLUSIONS: The imaging g-ratio may serve as a biomarker for the relative degree of axonal and myelin loss in MS.