Biology subjects

Demsar, J.

Publications and source records attributed to Demsar, J..

4 recordsLinked to original sources

bioRxiv2025

QuNex Recipes: Executable, Human-Readable Workflows for Reproducible Neuroimaging Research

Preprocessing and analysis of neuroimaging data are technically demanding, often requiring a combination of multiple software tools, modality-specific pipelines, and extensive parameter tuning to match dataset characteristics. These complexities make it difficult to document workflows in sufficient detail to ensure complete transparency and reproducibility. To address these challenges, we introduce QuNex recipes, a framework for defining and executing complete neuroimaging workflows - encompassing data onboarding, preprocessing, and analysis - in a transparent, machine- and human-readable format. Recipes are implemented as an integrated feature of the Quantitative Neuroimaging Environment & Toolbox (QuNex), a containerized, open-source platform for end-to-end multimodal and multi-species neuroimaging processing. The recipes framework enables seamless integration of QuNex commands with custom scripts and external tools, capturing every processing step and parameter setting. A fully reproducible study can thus be shared and replicated by providing only (a) the QuNex version used, (b) the recipe file, and (c) the data. This approach standardizes workflow specification, enhances transparency, and enables one-command replication of complex neuroimaging analyses. By providing a standardized way to describe and share workflows, recipes facilitate open exchange of best practices and reproducible methods within the neuroimaging community.

neuroscience↗

bioRxiv2025

Cognition-centric brain activity across diverse imaging tasks constrains the representation of mental health in task-evoked brain function

Large-scale neuroimaging studies best powered to investigate mental health biomarkers continue to administer a set of in-scanner tasks that, to date, yield weak brain-mental health associations. Triangulating the source of these weak associations is critical to directing the focus of future work. Using these tasks strong brain-cognition associations as a positive control, we analyzed potential sources of weak brain-mental health associations across five popular fMRI tasks and dozens of task-evoked brain networks. Across tasks and networks, latent mental health factors showed substantially weaker brain associations than cognition despite comparable measurement reliability and greater interindividual variability. Moreover, increasing brain-feature dimensionality improved cognitive but not mental health signal, and task-evoked network profiles were 1.8-2.3 times more homogeneous and 3.5 times lower-dimensional than mental health profiles. Findings implicate a signal evocation bottleneck: common fMRI tasks elicit little interindividual variation in mental health-related brain activity relative to the diversity of mental health itself. Overall, these findings motivate a shift in focus to the development and use of tasks in large-scale studies more explicitly designed to evoke mental health-relevant brain activity. Significance StatementIdentification of robust individual-specific mental health-related brain signals from fMRI remains elusive, limiting the clinical utility of brain imaging. To investigate why, we systematically evaluated brain signals evoked by tasks used by the Human Connectome Project-Young Adult study, which are now commonly included in many large-scale neuroimaging studies. Our findings suggest that, more so than analytical or measurement constraints on detecting these signals, a primary limitation is a failure of these prevalent fMRI tasks to evoke mental health-relevant brain signals in the first place. These findings suggest that progress in psychiatric neuroimaging may require tasks more explicitly designed to evoke mental health-relevant brain activity.

neuroscience↗

bioRxiv2022

QuNex -- An Integrative Platform forReproducible Neuroimaging Analytics

Neuroimaging technology has experienced explosive growth and transformed the study of neural mechanisms across health and disease. However, given the diversity of sophisticated tools for handling neuroimaging data, the field faces challenges in method integration (1-3), particularly across multiple modalities and species. Specifically, researchers often have to rely on siloed approaches which limit reproducibility, with idiosyncratic data organization and limited software interoperability. To address these challenges, we have developed Quantitative Neuroimaging Environment & Toolbox (QuNex), a platform for consistent end-to-end processing and analytics. QuNex provides several novel functionalities for neuroimaging analyses, including a "turnkey" command for the reproducible deployment of custom workflows, from onboarding raw data to generating analytic features. The platform enables inter-operable integration of multi-modal, community-developed neuroimaging software through an extension framework with a software development kit (SDK) for seamless integration of community tools. Critically, it supports high-throughput, parallel processing in high-performance compute environments, either locally or in the cloud. Notably, QuNex has successfully processed over 10,000 scans across neuroimaging consortia (4), including multiple clinical datasets. Moreover, QuNex enables integration of human and non-human workflows via a cohesive translational platform. Collectively, this effort stands to significantly impact neuroimaging method integration across acquisition approaches, pipelines, datasets, computational environments, and species. Building on this platform will enable more rapid, scalable, and reproducible impact of neuroimaging technology across health and disease.

neuroscience↗

bioRxiv2020

Mapping Brain-Behavior Space Relationships Along the Psychosis Spectrum

Difficulties in advancing effective patient-specific therapies for psychiatric disorders highlight a need to develop a stable neurobiologically-grounded mapping between neural and symptom variation. This gap is particularly acute for psychosis-spectrum disorders (PSD). Here, in a sample of 436 cross-diagnostic PSD patients, we derived and replicated a dimensionality-reduced symptom space across hallmark psychopathology symptoms and cognitive deficits. In turn, these symptom axes mapped onto distinct, reproducible brain maps. Critically, we found that multivariate brain-behavior mapping techniques (e.g. canonical correlation analysis) do not produce stable results. Instead, we show that a univariate brain-behavioral space (BBS) can resolve stable individualized prediction. Finally, we show a proof-of-principle framework for relating personalized BBS metrics with molecular targets via serotonin and glutamate receptor manipulations and gene expression maps. Collectively, these results highlight a stable and data-driven BBS mapping across PSD, which offers an actionable path that can be iteratively optimized for personalized clinical biomarker endpoints.

neuroscience↗