(J-392) Automation of Y-maze Behavioral Analysis of Rodents using Python

            Efficiency and repeatability in data analysis is critical in research. Software use is prevalent in neural engineering behavioral studies as they enable interpretation of neurological and motor functions in pre-clinical models. Software programs such as Any maze™ (California, USA) and Ethovision® (Virginia, USA) allow researchers to analyze behavioral and cognitive data on individual rodent models. Utilizing these programs, each video of the rodent behavior is manually set to calculate cognitive and behavioral parameters of interest. These processes require extensive time when analyzing large data sets.   In an effort to do bulk data analyses, our team developed a python-based automated data collection tool to analyze rodent behavioral videos and extract specific parameters and visual characteristics of the behavior.

Our automation program allowed for precise detection and tracking of the mouse within the Y- maze behavioral test. The Y maze is a memory test associated with locomotor activity, and social behavior, which can indicate neurodegeneration and loss of cognitive function (Figure 1). To initiate automation, a Y maze mask was generated, where the maze was divided into four zones. These four zones were further subdivided into smaller sections. Next, a snapshot of the percentage of white color of each section within the established zones was recorded. The animal’s location is detected in a particular section of the maze by the reduction in the amount of white color in the section (mice used were dark colored) (Figure 2). The program can adjust with lighting variations and compensates the initial snapshot values. Unlike commercial products, which captures a small percentage of the rodent’s size, our software, Maze Master Tracker, can track the rodent from nose to tail tip.  

The following seven quantitative parameters are collected from our Y-Maze behavioral memory recognition tests: (a) zone pattern for spontaneous alternation, (b) percentage of spontaneous alternation, (c) time spent within each zone, (d) number of times entered in each zone, (e) distance in pixels, (f) average speed in pixels per second, and (g) total time. In addition, the program captures these parameters with and without considering immobility time.

Heatmaps are an additional feature as it is a data visualization technique that allows analysis of the overall movements during behavioral testing. The software captures 2 forms of heatmaps. Heatmap A depicts a real time video of movement while Heatmap B displays scaled location distribution over time (Figure 3). The Heatmap B allows for analysis of each animal’s individual mobility (Figure 4). For studies that emphasize analysis upon multiple clustering of animals within a group, a combined heatmap can be developed to illustrate a cumulative scaled location distribution of a specific cohort (Figure 5).

            The advantage of automating this process allows for increased accuracy in data analyses and time efficiency in data processing. Researchers can analyze multiple behavioral videos at once, obtain results in a condensed timeframe, and enable accurate detection and tracking of the subject. Data is consistently reproducible, and the software is adaptable. Using an automated program allows for tracking of a larger number of parameters and for reprocessing of previously recorded video libraries. Future improvements of the software could be expanded to track white colored animals and the enhancement could be made to apply the program to a variety of behavioral and cognitive tests in addition to the Y-maze.