Case Study: Tracking Mouse Footfalls with the Crosshairs Tool
Above: Frames annotated by 20 crowd workers.
Below: Crowd annotations grouped with a clustering algorithm.
This case study explains how to use the crosshairs tool to track mouse footfalls.
Quantitative studies of whole animal morphology and locomotion are difficult to automate due to high variability in image sets caused by changes in background intensity or intrinsic variability between subjects or specimens. Quantius workers are adept at dealing with such variability and also have knowledge of basic concepts in animal morphology such as the difference between toes and fingers, tails and noses.
Here, we leverage these strengths of human intervention to study the footfall pattern of a mouse from a study of left-right alternation in locomotion; returning data that is almost indistinguishable from time-intensive annotations made by a trained expert.
First, we take a movie of a mouse and save it as a set of .png images. Then, we upload the images to Quantius with the following parameters:
- Job name: Mousefootfalls.
- Tool: Crosshairs.
- Slider: Off.
- Instructions: "Click the tip of the tail, the ends of each digit (fingers and toes), and the tip of the nose of the mouse."
- Replicates: 20.
Quantius workers enable natural detection of animal morphologies in difficult phase contrast settings. a Raw data preparation for upload to Quantius. b Job setup parameters in the Quantius interface. c Gait analysis of the mouse using annotations by an expert scientist, and also by a crowd of Quantius workers. Raw Quantius annotations are grouped via a clustering algorithm in Matlab and tracked using TrackMate in FIJI in post-processing. The x position of the tail, digits, and nose of the mouse strongly agree between the expert and Quantius workers, with velocities of the paws accurate to within 5% error.