MRIO: A Magnetic Resonance Imaging Acquisition and Analysis Ontology

Magnetic Resonance Imaging (MRI) is a biomedical imaging technology used to non-invasively visualize internal structures of tissues in three-dimensional space. MRI is widely used in studying the structure and function of the human brain and is a powerful tool for diagnosing neurological diseases in clinical and research settings. However, effective management and analysis of MRI data have always been challenging. To address this challenge, Alexander Bartnik and colleagues developed the MRI Ontology named MRIO for acquiring and analyzing MRI data in their research.

Research Background

MRI technology is widely used in clinical and research settings because it can non-invasively obtain images from within the human body. Clinically, MRI can be used to diagnose neurological diseases, providing treatment guidance by locating and assessing the extent of pathology. In research, MRI data can serve as biomarkers to help develop personalized treatment plans for neurological diseases and increase understanding of brain structure, function, and connectivity. However, the diversity and heterogeneity of MRI data acquisition make data management and analysis complex and laborious.

Research Aims

In both research and clinical practice, accurately and consistently managing and analyzing MRI data requires a “common language.” Although there are existing standards like DICOM and BIDS for organizing and describing imaging data, they still have many shortcomings in the standardization of analysis and derived results. The authors developed the MRIO ontology to provide a logically coherent set of classes and logical axioms for acquiring and analyzing MRI data, aiming to standardize the organization and analysis of the data.

Research Origin

This paper was written by eight researchers: Alexander Bartnik, Lucas M. Serra, Mackenzie Smith, William D. Duncan, Lauren Wishnie, Alan Ruttenberg, Michael G. Dwyer, and Alexander D. Diehl. They are affiliated with the Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, and the College of Dentistry, University of Florida. The paper was published in the journal Neuroinformatics and accepted on April 22, 2024.

Research Methods

MRIO is developed using the Web Ontology Language (OWL) 2 with Protégé and strictly follows the principles of the OBO Foundry for development and maintenance.

Development Management

The development of MRIO used the Ontology Development Kit to ensure dependency management and interoperability. The HermiT OWL 2 reasoner was used to assess the logical consistency of axioms.

Construction

MRIO is built upon the Ontology for Biomedical Investigations (OBI) and Information Artifact Ontology (IAO), using standardized formats to operate on MRI data. Most classes are subclasses of Information Content Entity (IAO:0000030), Data Set (IAO:0000100), and Data Transformation (OBI:0200000).

Validation

Following OBI’s standard contribution process, MRIO internally developed and contributed multiple classes and extended OBI’s Magnetic Resonance Imaging Assay (OBI:0002985) class to accommodate new research paradigms.

Research Process

MRIO standardizes the process by defining entities used for different types of MRI acquisition, such as T1-weighted and T2-weighted imaging. It also incorporates ranges of common MRI acquisition parameters extracted from DICOM headers to describe specific types of MRI acquisitions.

Diverse Acquisition Parameters and Image Dataset Analysis

From the metadata of image datasets to derived data, MRIO defines a set of logical axioms and data properties to comprehensively describe the process and results of image dataset analysis. For instance, for common high-resolution T1-weighted MRI and T2 FLAIR MRI, MRIO provides precise logical definitions and can automatically allocate analyses to facilitate researchers in organizing and reporting MRI data.

Federated Query and Data Management Tools

MRIO offers a systematic approach for managing and querying neuroimaging data. By generating short and precise queries, researchers can easily retrieve and analyze data in the XNAT database without needing an in-depth understanding of SPARQL or SQL languages.

Research Findings

MRIO has now become an important resource supporting a fully automated neuroinformatics platform, effectively promoting the standardization and reproducibility of neuroimaging research.

Significance of the Research

The development and application of MRIO have significant scientific and practical value. By providing a standardized, highly interoperable framework, MRIO not only improves the efficiency of MRI data management but also promotes data sharing and reuse in neuroimaging research.

Scientific Value

• Standardized Acquisition and Analysis: Through standardized processes for data acquisition and analysis, MRIO significantly improves the management and utilization efficiency of MRI data.
• Data Sharing and Reuse: Developed in accordance with the principles of the OBO Foundry, MRIO integrates with a broad range of biomedical ontologies, promoting cross-disciplinary data sharing and reuse.

Practical Applications

• Promotion of Neuroimaging Research: Through automated analysis allocation and smart querying, MRIO provides researchers with a simple and efficient data management tool.
• Cross-Disciplinary Collaboration: MRIO helps combine neuroimaging research with other biomedical fields, driving cross-disciplinary scientific discoveries.

Conclusion

The MRIO ontology provides a standardized framework for neuroimaging research. Through functional and standardized class definitions, it helps researchers efficiently manage and analyze MRI data. Moreover, the development and evolution of MRIO mark a significant progress in the fields of open scientific data and biomedical ontology, laying a solid foundation for future neuroimaging research.