INTRODUCTION: The Open Field Test (OFT) is widely accepted to assess exploratory behavior, locomotor activity, and anxiety traits in rodents. Healthy rodents exhibit strong thigmotaxis. Hence, time spent and entries into central (aversive) versus peripheral zones are commonly used readouts of anxiety traits. While markerless pose estimation tools such as DeepLabCut (DLC) have greatly improved behavioral tracking, there is a lack of accessible, standardized post-DLC pipeline. This limits reproducibility across laboratories and hinders broader application.

AIM(S): To develop a reproducible, low-computational pipeline for spatial analysis of OFT data, using centroid-based metrics derived from DLC outputs, and to evaluate its applicability in disease models.

METHOD(S): We developed a Python-based post-processing pipeline in Jupyter Notebook that calculates the dynamic centroid of tracked body parts. Using interactive rectangular zone selection, we segmented the open field arena into central and peripheral regions and quantified time spent in each. The approach is tolerant to pixel-level noise introduced by shorter DLC training on limited hardware, making it scalable and time-efficient.

RESULTS: The centroid-based approach successfully tracked exploratory patterns and distinguished between central and peripheral zone occupancy, ignoring pixel fluctuations. Time spent in the center versus periphery was automatically computed per animal. The system allowed for rapid adjustment of zone boundaries and could be adapted for other spatial layouts.

CONCLUSIONS: This pipeline provides an accessible and robust framework for analyzing OFT data post-DLC, requiring minimal computational resources and no proprietary software. It enhances reproducibility and allows broader application of DLC-based behavioral assays.

FINANCIAL SUPPORT: Funded by The Polish National Agency for Academic Exchange Strategic Partnerships Grant (BNI/PST/2023/1/00132/U/00001), held by M. Kucharczyk.