OpenFieldAI - new open-source AI based software for tracking rodents and training open field test models.

Area of Science:

• Neuroscience
• Behavioral Science
• Computational Biology

Background:

• Open field test (OFT) is crucial for rodent behavioral analysis, assessing locomotion, exploration, and anxiety.
• Current analysis methods (manual or commercial software) have limitations like observer bias, high cost, and steep learning curves.
• Automated systems are preferred for objectivity and efficiency in scientific research.

Purpose of the Study:

• To introduce OpenFieldAI, a novel, open-source, Python-based software for automated analysis of rodent behavior in OFT.
• To provide researchers with a user-friendly, cost-effective alternative to existing commercial software.
• To ensure the reliability and accuracy of OpenFieldAI through validation against established methods.

Main Methods:

• OpenFieldAI utilizes the YOLO algorithm for accurate detection and tracking of rodents within the OFT arena.
• The software accepts pre-recorded or live video inputs, calculating key behavioral parameters (speed, distance, time in ROI, entries/exits).
• Features include automatic ROI detection, manual ROI drawing, heat maps, and path tracking for comprehensive neurobehavioral insights.

Main Results:

• OpenFieldAI demonstrated high reliability, with Pearson correlation coefficients above 0.9 when compared to the commercial ANY-maze software for key parameters.
• The software provides essential outputs like centroid graphs, heat maps, and line paths, aiding in detailed scientific interpretation.
• Validation confirmed the software's accuracy in measuring total distance, speed, and time/entries in central and peripheral regions.

Conclusions:

• OpenFieldAI offers a reliable, accurate, and accessible automated solution for rodent behavioral analysis in open field tests.
• Its intuitive GUI and open-source nature make advanced behavioral research more accessible to a wider range of scientists.
• The software provides valuable neurobehavioral data, overcoming the limitations of manual analysis and costly commercial alternatives.