Summary:
A multi-institutional team led by Lehigh University has been awarded a four-year, $2.8 million grant from the National Institute of Biomedical Imaging and Bioengineering (NIBIB) to build UniBrain, an integrated artificial intelligence platform that consolidates fragmented neuroimaging workflows. UniBrain combines preprocessing, segmentation, multi-modal integration, and disease prediction into a single deep learning architecture, paired with a conversational agent interface. The goal is to convert raw MRI scans into interpretable diagnostic biomarkers in minutes instead of months, making advanced neuroimaging accessible to clinicians and researchers without extensive computational resources.
Key Facts:
- $2.8M NIH Initiative: The four-year project is funded by NIBIB and led by Lehigh University in collaboration with Massachusetts General Hospital/Harvard Medical School, Emory University, and the University of Pittsburgh.
- Unified Pipeline Architecture: UniBrain replaces the current practice of moving data between multiple specialized software packages by integrating image registration, tissue segmentation, feature extraction, and network-level disease prediction into one cohesive system.
- Agentic AI Interface: A conversational, ChatGPT-like interface will let clinicians and researchers upload raw MRI files, ask diagnostic questions in plain language, and receive automated, interpretable biomarker reports without needing advanced programming skills.
Source: Lehigh University
Extracting clinically actionable biomarkers and neuroscientific insights from brain scans remains one of the most time-consuming challenges in medicine. Today, converting raw neuroimaging data into diagnostic predictions typically requires routing large datasets through a chain of single-purpose tools—motion correction, anatomical registration, tissue segmentation, and separate packages for network modeling. Each stage often requires different software, file format conversions, and manual quality control.
For large clinical cohorts, this fragmented pipeline can take weeks or months, creating a significant barrier for clinicians and neuroscientists who lack advanced coding skills or high-performance computing infrastructure. UniBrain aims to break down these barriers by unifying the entire processing chain into an end-to-end AI framework that automates preprocessing, harmonizes multi-modal imaging, and produces transparent diagnostic outputs.
The project is led by Lifang He, PhD, associate professor of computer science and engineering at Lehigh University, and brings together computer scientists, radiologists, bioengineers, and neuroscientists from Lehigh University, Massachusetts General Hospital/Harvard Medical School (Xiang Li), Emory University (Carl Yang), and the University of Pittsburgh (Liang Zhan).
Consolidating Fragmented Neuroimaging Workflows
Conventional neuroimaging pipelines treat spatial normalization, skull stripping, volumetric segmentation, structural and functional connectome mapping, and pathology classification as separate, sequential tasks. This modular approach multiplies opportunities for error propagation, requires extensive file format translation, and demands manual quality assurance each time data moves between tools.
UniBrain re-engineers that approach by deploying end-to-end deep learning models that perform preprocessing and downstream analysis within a single architecture. The platform brings structural MRI, functional MRI (fMRI), and diffusion tensor imaging (DTI) into a common latent space, enabling integrated feature extraction and joint optimization across modalities. Early evaluations indicate UniBrain’s integrated framework matches or surpasses the accuracy of conventional multi-tool pipelines while cutting processing time dramatically—from weeks to minutes on typical datasets.
“This is a very ambitious project,” said Dr. He. “We’re building AI tools that not only perform analysis automatically but also rethink the entire neuroimaging workflow from start to finish. Solving this problem requires collaboration across many specialties.”
An Agentic, Conversational Interface for Medicine
Beyond the computational backbone, the team is creating an agentic AI interface designed to democratize advanced neuroimaging for frontline clinicians. Modeled on large language model assistants, the conversational system will allow physicians and researchers to upload raw DICOM or NIfTI files and interact with their data using natural language prompts.
Clinicians will be able to ask about specific anatomical anomalies, request automated regional volume comparisons, or monitor changes in functional connectivity over time. UniBrain will autonomously choose optimal preprocessing parameters, run underlying deep learning classifiers, and produce clear, interpretable clinical reports. Those reports will include visual saliency maps and other explainability features that highlight subtle biomarkers that may be difficult to spot by eye.
Security, Privacy, and Clinical Validation
Because neuroimaging data contain sensitive patient information, the platform incorporates privacy-preserving protocols and AI security measures overseen by Lehigh computer scientist Lichao Sun, PhD, to ensure compliance with data governance standards. The project also includes clinical collaborators—Quanzheng Li, PhD; Randy Hirschtick, MD; Daniel Barron, MD, PhD; Boadie Dunlop, MD; and Ying Ding, PhD—who will help validate UniBrain across psychiatric and neurodegenerative cohorts, including Alzheimer’s disease and major depressive disorder.
By connecting raw medical imaging with rapid, interpretable biomarker discovery, UniBrain aims to establish an accessible, standardized foundation for precision neurology and translational neuroscience. If successful, the platform could streamline research workflows, accelerate clinical decision-making, and enable broader adoption of advanced neuroimaging analyses in clinical practice.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The full journal paper will be reviewed in detail upon release.
- Additional context was provided by our staff.
About this neurotech and schizophrenia Research:
- Media Contact: Katie Kackenmeister
- Source: Lehigh University
- Image Credit: Image credited to Neuroscience News