Analog Interval Threshold

Overview

This transform detects continuous time intervals during which an analog signal’s value meets a specified threshold condition (e.g., staying above a value).

Detailed Description

This transformation is used to identify continuous periods, or “intervals,” where a signal’s amplitude meets certain criteria. Unlike event detection, which marks a single point in time, interval detection finds the start and end times of periods that satisfy the condition.

The user can define a threshold and specify whether the signal must be above (Positive), below (Negative), or have its absolute value exceed (Absolute) that threshold.

Additionally, two timing parameters help refine the detection: - A lockout_time prevents the detection of new intervals for a set duration after the start of a previous interval. - A min_duration ensures that only intervals lasting longer than a specified duration are included in the final output.

This transform is useful for isolating epochs of interest in a continuous signal for further analysis. It takes an analog time series as input and produces a digital interval series as output.

Neuroscience Use Cases

This transform is particularly useful for identifying specific “states” or “epochs” in neural or behavioral data.

• Identifying Periods of Movement: From accelerometer or video tracking data, one can find intervals of high activity that correspond to an animal moving, versus periods of rest.
• Analyzing Muscle Activation: In electromyography (EMG), this can be used to determine the duration of muscle contractions by finding intervals where the rectified signal remains above a certain level of activity.

Parameters

This transform has the following parameters:

• threshold_value: The amplitude value the signal must be above/below for the duration of the interval.
• direction: Specifies the condition the signal must meet. Valid options are: "Positive", "Negative", or "Absolute".
• lockout_time: A duration (in the same time units as the data) after the start of a detected interval, during which no new intervals can begin. This is useful for isolating the onset of distinct episodes of activity.
• min_duration: The minimum required length of an interval. Any detected period that is shorter than this value will be discarded.
• missing_data_mode: Defines how to handle non-consecutive time points in the signal.
  • Treat as Zero: Missing time points are treated as if the signal’s value is zero. This can terminate an interval if zero does not meet the threshold condition.
  • Ignore: The algorithm proceeds to the next available data point, effectively ignoring the time gap.

Example Configuration

Here is a complete example of a JSON configuration file that could be used to run this transformation. This example finds all intervals where a signal named “LFP_channel_1” remains above 1.5 for at least 50 milliseconds.

[
  {
    "transformations": {
      "metadata": {
        "name": "Interval Threshold Detection Pipeline",
        "description": "Detects intervals of high activity in an LFP signal.",
        "version": "1.0"
      },
      "steps": [
        {
          "step_id": "1",
          "transform_name": "Threshold Interval Detection",
          "phase": "analysis",
          "input_key": "LFP_channel_1",
          "output_key": "detected_activity_intervals",
          "parameters": {
            "threshold_value": 1.5,
            "direction": "Positive",
            "lockout_time": 0.0,
            "min_duration": 0.050,
            "missing_data_mode": "Treat as Zero"
          }
        }
      ]
    }
  }
]