NiChart
Type: Other,
Type: Software,
Keywords: MRI,Neuroimaging,Data Harmonization,Machine Learning,AI,Brain Age,Biomarker,Structural Imaging,Functional Imaging,Diffusion Tensor Imaging
NeuroImaging Chart of AI-based Imaging Biomarkers
NiChart is a comprehensive framework designed to facilitate large-scale, multi-modal brain MRI analysis, functioning as both an open-source software and a cloud platform. By utilizing a library of advanced machine learning models, NiChart transforms complex neuroimaging data into a compact system of informative signatures. NiChart offers robust processing pipelines, diverse ML-based biomarkers, and statistical harmonization tools to minimize inter-scanner variability. The platform provides an interactive, “growth-chart” style interface to view user values computed from MRI images, together with normative reference distributions derived from over 100,000 MRI time points—spanning 79,000 participants and 50 studies collected across 100 distinct scanners over three decades. NiChart’s modular, vertically integrated design is aimed at supporting scalable, reproducible research and facilitating future expansion.
*Allows users to calculate imaging signatures that quantify complex multi-variate imaging patterns of aging and disease-related brain changes.
*Rich and diverse support for machine learning based biomarker models for aging, neurodegeneration, psychiatric disorders, genetics, lifestyle, and cardiovascular risk factors.
*Large multi-study reference set allows users to compare their data with NiChart-based normative distributions
*Image processing tools for deriving a panel of imaging derived phenotype from structural (sMRI), diffusion (dMRI) and functional (fMRI) imaging data
*Performs statistical data harmonization to mitigate scanner variations
NiChart is a comprehensive framework designed to facilitate large-scale, multi-modal brain MRI analysis, functioning as both an open-source software and a cloud platform.
*Study level analysis of a brain MRI cohort
*Comparing imaging signatures of a single MRI scan to the NiChart’s reference values
*Brain imaging related research experiments, publications, and education
*Internet connection for the cloud version.
*For the installation version, a Linux system with 8GB+ RAM and CUDA supported GPU with 6GB+ VRAM is recommended.
Erus, G., Harman, G., Abdulkadir, A., Habes, M., Getka, A., Baik, K., … Davatzikos, C. (under review). NiChart: A neuroimaging brain chart derived from 106,902 scans, with integrated cloud-based tools for MRI processing and machine learning. Nature Neuroscience.
Christos Davatzikos, Director of AI2D Center for AI/Data Science for Integrated Diagnostics
University of Pennsylvania, USA
TEAM / COLLABORATOR(S)
*Alexander Getka, Sr. Software Engineer, University of Pennsylvania
*Dhivya Srinivasan, Sr. Data Analyst, University of Pennsylvania
*Gareth Harman, Sr. Data Analyst, University of Pennsylvania
*Guray Erus, Director of Research, University of Pennsylvania
*Kyunglok Baik, Sr. Data Analyst, University of Pennsylvania
FUNDING SOURCE(S)
*NIH U24NS130411
*NIH RF1AG054409