Biology subjects

Wintermark, M.

Publications and source records attributed to Wintermark, M..

4 recordsLinked to original sources

bioRxiv2026

UniST: A Unified Computational Framework for 3D Spatial Transcriptomics Reconstruction

Spatial transcriptomics (ST) enables the measurement of gene expression in its native spatial context, yet most ST datasets are acquired as two-dimensional (2D) sections. Consequently, the underlying three-dimensional (3D) organization of tissues is only partially observed, and 3D ST data generated from serial sections are typically sparse and heterogeneous, with substantial tissue loss and missing measurements. These limitations pose major analytical challenges for reconstructing coherent 3D tissue architecture, rather than issues of experimental scalability alone. Here, we present UniST, a unified generative artificial intelligence (AI) framework designed to computationally reconstruct dense and continuous 3D ST landscapes from sparse serial sections, without altering the underlying experimental ST technologies. UniST integrates three complementary modules: kernel point convolution with cross-attention layers for point cloud upsampling, optical flow-based interpolation for continuous slice reconstruction, and a graph autoencoder with implicit neural representations for gene expression imputation. Together, these components densify sparse slices, resolve discontinuities, and map spatial coordinates to high-dimensional transcriptomics. Across multiple ST platforms and tissue contexts, UniST accurately restored structural continuity and biologically meaningful expression patterns. In a mouse embryo dataset, UniST reconstructed a dense 3D heart architecture from sparsely sampled slices. In two 3D human cancer tissues, UniST recovered critical spatial features, including tumor-immune boundaries and tertiary lymphoid structures, that were fragmented in the original data. By providing a generalizable computational solution that complements existing ST acquisition protocols, UniST facilitates cost-efficient and scalable reconstruction of 3D ST landscapes, enabling more faithful investigation of tissue organization and disease biology.

bioinformatics↗

bioRxiv2024

Labeling Natural Killer cells with superparamagnetic iron oxide nanoparticles for detection by preclinical and clinical-scale magnetic particle imaging

IntroductionClinical adoption of NK cell immunotherapy is underway for medulloblastoma and osteosarcoma, however there is currently little feedback on cell fate after administration. We propose magnetic particle imaging (MPI) for the detection, localization, and quantification of VivoTrax-labeled NK cells. MethodsHuman-derived NK-92 cells were labeled by co-incubation with VivoTrax for 24 hours then the excess nanoparticles were washed with centrifugation. Cytolytic activity of labeled vs. unlabeled NK-92 cells was assessed after 4 hours of co- incubation with medulloblastoma cells (DAOY) or osteosarcoma cells (LM7 or OS17) using bioluminescent or GFP counts. Labeled NK-92 cells at two different doses (0.5 or 1 x 106) were administered to excised mouse brains (cerebellum), tibias, and lungs then imaged by 3D preclinical MPI (MOMENTUM imager) and localized relative to fiducial markers. NK-92 cells were imaged by clinical-scale MPI under development at Magnetic Insight Inc. ResultsNK-92 cells were labeled with an average of 3.17 pg Fe/cell with no measured effects on cell viability or cytolytic activity against 3 tumor cell lines. MPI signal was directly quantitative with the number of VivoTrax-labeled NK-92 cells, with preclinical limit of detection of 3.1 x 104 cells on MOMENTUM imager. Labeled NK-92 cells could be accurately localized in mouse brains, tibias, and lungs within < 1 mm of stereotactic injection coordinates with preclinical scanner. Feasibility for detection of a clinically relevant dose of 4 x 107 labeled NK-92 cells was demonstrated on clinical-scale MPI. ConclusionMPI can provide sensitive, quantitative, and accurate spatial information on NK cell delivery, showing its potential to resolve a significant unmet clinical need to track NK cell treatments in patients.

biophysics↗

bioRxiv2022

Microstructural alterations in tract development in college football: a longitudinal diffusion MRI study

Background and ObjectivesRepeated concussive and sub-concussive impacts in high-contact sports can affect the brains microstructure, which can be studied using diffusion MRI. Most prior imaging studies, however, employ a cross-sectional design, do not include low-contact players as controls, or use traditional diffusion tensor imaging without investigating novel tractspecific microstructural metrics. MethodsWe examined brain microstructure in 63 high-contact (American football) and 34 low-contact (volleyball) collegiate athletes with up to 4 years of follow-up (315 total scans) using advanced diffusion MRI, a comprehensive set of multi-compartment models, and automated fiber quantification tools. We investigated diffusion metrics along the length of tracts using nested linear mixed-effects models to ascertain the acute and chronic effects of sub-concussive and concussive impacts, as well as associations between diffusion changes with clinical, behavioral, and sports-related measures. ResultsSignificant longitudinal increases in fractional anisotropy and axonal water fraction were detected in volleyball players, but not in football players, along with decreases in radial and mean diffusivity as well as orientation dispersion index (all findings absolute T-statistic > 3.5, p < .0001). This pattern was present in the callosum forceps minor, left superior longitudinal fasciculus, left thalamic radiation, and right cingulum hippocampus. Longitudinal group differences were more prominent and observed in a larger number of tracts in concussed (previously or in-study) football players (p < .0001), while smaller effects were noted in un-concussed players. An analysis of immediate-post concussion scans in football players demonstrated a transient localized increase in axial diffusivity, mean and radial kurtosis in the left uncinate and right cingulum hippocampus (p < .0001). Finally, football players with high position-based sub-concussive risk demonstrated increased orientation dispersion index over time (p < .0001). DiscussionThe observed longitudinal changes in our volleyball cohort likely reflect normal development in this age range, while the relative attenuation of these effects seen in football, and especially concussed athletes, could possibly reveal diminished myelination, altered axonal calibers, or depressed pruning processes leading to a static, non-decreasing axonal dispersion. This prospective longitudinal study demonstrates significantly divergent tract-specific trajectories of brain microstructure, possibly reflecting a concussive and/or repeated sub-concussive impact-related alteration of normal white matter development in football athletes.

neuroscience↗

bioRxiv2019

Diffusion Tensor Tractography of Brainstem Fibers and Its Application in Pain

Evaluation of brainstem pathways with diffusion tensor imaging (DTI) and tractography may provide insights into pathophysiologies associated with dysfunction of key brainstem circuits. However, identification of these tracts has been elusive, with relatively few in vivo human studies to date. In this paper we proposed an automated approach for reconstructing nine brainstem fiber trajectories of pathways that might be involved in pain modulation. We first performed native-space manual tractography of these fibers in a small normative cohort of participants and confirmed the anatomical precision of the results using existing anatomical literature. Second, region-of-interest pairs were manually defined at each extracted fibers termini and nonlinearly warped to a standard anatomical brain template to create an atlas of the region-of-interest pairs. The resulting atlas was then transformed non-linearly into the native space of 17 veteran patients brains for automated brainstem tractography. Lastly, we assessed the relationships between the integrity levels of the obtained fiber bundles and pain severity levels. Fractional anisotropy (FA) measures derived using automated tractography reflected the respective tracts FA levels obtained via manual tractography. A significant inverse relationship between FA and pain levels was detected within the automatically derived dorsal and medial longitudinal fasciculi of the brainstem. This study demonstrates the utility of DTI in exploring brainstem circuitries involved in pain processing. In this context, the described automated approach is a viable alternative to the time-consuming manual tractography. The physiological and functional relevance of the measures derived from automated tractography is evidenced by their relationships with individual pain severities.

neuroscience↗