Methodological considerations in the use of Noldus EthoVision XT video tracking of children with autism in multi-site studies

Abstract

Animal models of autism spectrum disorders (ASD) contribute to understanding of the role of genetics and the biological mechanisms underlying behavioral phenotypes and inform the development of potential treatments. Translational biomarkers are needed that can both validate these models and facilitate behavioral testing paradigms for ASD in humans. Automated video tracking of movement patterns and positions recorded from overhead cameras is routinely applied in behavioral paradigms designed to elicit core behavioral manifestations of ASD in rodent models. In humans, laboratory-based observations are a common semi-naturalistic context for assessing a variety of behaviors relevant to ASD such as social engagement, play, and attention. We present information learned and suggest guidelines for designing, recording, acquiring, and evaluating video tracking data of human movement patterns based on our experience in a multi-site video tracking study of children with ASD in the context of a parent-child, laboratory-based play interaction.

Keywords: Autism spectrum disorder; Automated behavioral assessment; Parent-child interaction; Play-based assessment; Video tracking.

Fig. 1.

Arena with same-sized regions of interest across two different assessment rooms.

Fig. 2.

Arenas of two different assessment rooms, using site-specific sized regions of interest. The room on the left is smaller than the room on the right.

Fig. 3.

Checkerboard pattern illustrating distortion at floor level (left) and 29-in. level (right). NOTE: Green marks are used as a visual aid to show the points selected for distortion analysis. The blue cross marks the center of the room. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig. 4.

Percent distortion at each point plotted against true distance from center pixel of the checkerboard pattern placed at the floor and 29-inch levels.

Fig. 5.

The standardized path and true measured distance between each step in the path.

Fig. 6.

The marker disc set up on the first step of the standardized path at heights of 1 inch (left), 29 in. (middle) and 41 in. (right).

Fig. 7.

Example of calibration points on the wall (left) and calibration points created using two columnar structures (right).

Fig. 8.

Example of the table ROI measured in the assessment room and delineated with physical indicators at the chosen height (i.e., the red tape at the top of the columns and on the wall). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig. 9.

Example of ROI from the grabbed image to the fully constructed zone in EthoVision XT (left to right). Note that the parallel red floor markings were used to correctly position the Styrofoam columns. The zones were delineated by drawing lines in EthoVision XT, connecting the indicators (i.e., red tape) on top of the Styrofoam columns. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).

Fig. 10.

Example of an original arena with calibration markers (A), frame shift (B) and overlaid images (C) before applying transformations.

Fig. 11.

Steps to fixing a frame shift, including rotation (A), translation (B) and final cropped image (C).