Browse Trajectories GUI v2 - RodentDataAnalytics/mwm-ml-gen GitHub Wiki

The Browse Trajectories GUI is used as a tool for the manual labelling of the trajectories segments.

Upon loaded, this tool can be used to visualize the trajectories and their segments as well as to indicate various info and statistics such as animal average speed, animal latency, features values, segments length, etc.

With the visualization of each segment the user is given the ability to provide accurate labels and save his labelling work in order to use it latter for the classification process.

Contents

  1. Getting Started
  2. GUI Overview
  3. Labelling the Segments
  4. Common Issues with the Plot Area

Getting Started

In order to start using the GUI a Segmentation Configuration object must be loaded. To load one click on the button Select Configuration File and navigate to the location where the Segmentation Configuration object is saved and select it.

When the object is loaded then the user will be able to use the rest of the features of the GUI.

gui activation

GUI Overview

The GUI can be split into four main parts:

1. Plot Area: Inside this area the arena and the animal trajectories or segments will be plotted. Specifically for the trajectories/segments, the filled bullet point indicates the started position and the white bullet point the ended position of the animal.

2. Navigation Area: Pressing the <= or the => button the next or the previous trajectory will be selected. For quick navigation throughout the trajectories type the trajectory #id (the trajectories are numbered from 1, thus the first trajectory has id=1 and the 10th will have id=10) on the textfield and click the OK button.

3. Information Tables: The information tables shows various information about the selected trajectory (left table) or the selected segment (right table).

Specifically for each trajectory the following information are available:

  • Trajectory #id
  • Animal Group (if there are no animal groups available all animals will belong to group 1)
  • Animal ID
  • Session
  • Day
  • Trial
  • Track (refers to the Ethovision's automatic numbering)
  • Latency (escape latency)
  • Speed
  • Length (path length)

Specifically for each segment the following information are available:

  • Segment #id
  • Segment Offset (the first segment will always have zero offset)
  • Median Radius feature value
  • IQR Radius feature value
  • Focus feature value
  • Centre Displacement feature value
  • Inner Radius Variation feature value
  • Platform Proximity feature value
  • Boundary Eccentricity feature value
  • Longest Loop feature value
  • Latency
  • Speed
  • Length

Note: Refer to List of Features for the complete list of features along with their abbreviations.

4. Trajectory/Segment Selection and Labelling: This panel is used for the selection of the segments of each trajectory and their labelling. For more information about the labelling process refer to the following session.

overview

Labelling the Segments

To label a segment first select it from the Trajectory/Segments list and then select a label from the dropdown menu. Afterwards click on the + button. The label should then appear on the labels field. Multiple labels can be applied for a segment. To remove a label from the segment, select it from the dropdown menu and click the - button.

labelling

The labelled segments can be saved to a CSV file to be used later for the classification process. To save the labelled segments click on the Save button, a CSV file will be created in the selected folder. This CSV file will consists of a column indicating the segments #id following by columns containing the labels of the segments.

labels file

To load a previously saved labels CSV file click on the load button and navigate to the folder where this file is located.

Note: Refer to List of Labels for the complete list of features along with their abbreviations.

Common Issues with the Plot Area

In some cases the Plot Area fail to visualize the arena and the trajectories correctly and this either because the Experiment Properties values were wrong or the trajectory data were uncalibrated. The picture below illustrates (from left to right) a case were the Experimental Properties were wrong, a case were the trajectory data were uncalibrated and a case were the Experimental Properties were correct and the trajectory data calibrated.

labels file

  • Error because of the Experimental Properties: The user needs to make sure that the centre of the arena and the centre of the platform coordinates are provided correctly and that the platform is located on the correct location (refer to Data Input).

  • Error because the trajectory data were uncalibrated: Ethovision is able to calibrate animal trajectories for more information refer to the Ethovision Reference Manual.

Important note: In case the arena or/and the trajectories are not shown correctly then this error must be fixed before proceeding to the Classification process to avoid generating wrong results.