Filtering Capture Data

There may be cases where an abnormal score or disjointed skeleton signals that there was a data quality issue during the capture. In many of these cases, that doesn't mean that the whole capture is bad and there are several tools that can be used to filter the data and provide valuable clinical insights to your client. This article will outline how to know when filtering is necessary and the recommended filtering process.

When to Filter Data?

While the score is a valuable clinical tool, it can also be helpful to serve as a rater of reliability. The ranges below indicate when it is important to complete a reliability check and filter the data:

• 100% - 80%: Typically no reliability concerns and no filtering is needed. Reliability Check can be completed but is not necessary
• 80% - 40%: There could be minor reliability concerns, but it also could be a clinically relevant score. Recommendation is to complete a Reliability Check but filtering will not always be needed.
• <40%: There are likely reliability concerns, but could also be a clinically relevant score with very poor gait. Recommendation is to complete a Reliability Check and filtering will be needed in most cases.
• <40%*: The addition of the asterisk (*) indicates that there is a reliability error so the Reliability Check and filtering processes should always be completed.

Reliability Check Process:

Based on the score ranges above, it may be recommended to complete a Reliability Check. This process involves using the Charts to evaluate the consistency of the data throughout the duration of the capture following the process below.

1. Open the Charts window by selecting the Bar Chart icon in the bottom toolbar.
2. In the Charts window, select "Foot Angles" in the left column to expand the Foot Angles Chart.
   1. NOTE: We recommend starting with the Foot Angles chart because most often this is where most filtering needs can be identified.
3. Check the Foot Angles Chart to assess whether it is uniform throughout the duration of the capture. The chart does not have to be perfectly uniform because there can be variation in gait, however it should be reasonably uniform where there are no major deviations. For example, the charts below show first an example of a uniform chart followed by a chart which requires filtering:

4. If the Foot Angles Chart does not appear to be reasonably uniform, then filtering is necessary using the process in the Filtering Data section of this Article below.
5. If the Foot Angles Chart is reasonably uniform, click through each of the following charts to further check for any major deviations: Pelvis Angles, Hip Angles, Knee Angles, Ankle Angles, Tibia Angles.
6. If you notice any major deviations you should follow the process outlined in the Filtering Data section, otherwise if all charts are reasonably uniform, then most often there is no filtering needed and the score is clinically reliable.

Filtering Data:

There are two tools in the software that can be used to manually filter the data when necessary. The steps below outline the recommended process to using these tools to ensure the data is clinically reliable.

1. If any major deviations were found during the Reliability Check process, the first step in filtering should be applying an Analysis Limit to omit the deviation from the data. The steps below outline how to apply an Analysis Limit:
   1. Hover your cursor over the position where the good data begins (most often the start of the capture).
   2. Right click, hover over "Events", and select "Analysis Limit". This will create the first limit in the capture.
   3. Next, hover your cursor over the position where the good data ends (right before the major deviation).
   4. Again, right click, hover over "Events", and select "Analysis Limit". This will create the second limit in the capture which defines the portion of data that should be included in the calculations. You should see that the data outside those bounds will be greyed out as shown below:

2. If the score still indicates that there are issues, the Cycles Manager can be used as the second filtering method. Open the Cycles Manager by selecting the "Cycles" icon in the top toolbar.
3. Use the dropdown in the top right to click through the different metric options while checking the chart for uniform maps through the gait cycle. Again, remember that variability in gait is normal and that strides should only be removed if they significantly differ from the other strides, not when there is normal variation or asymmetries. For reference, the charts below show an example of a uniform chart on the left and a chart needing filtering on the right:

4. When abnormal strides are identified, they can be removed by double clicking on the chart on the stride you would like to remove. You may also use the "Auto" button to automatically remove strides that are detected as abnormal.
5. If no abnormal strides are identified then the data is most likely clinically reliable

Summary:

• The Score is insightful for knowing when filtering may be necessary in a capture.
• The Reliability Check process can be used to objectively identify the need for filtering.
• The Analysis Limit and Cycles Manager are useful tools for data filtering when necessary.
  • For a video demonstration of both tools, view the Checking your Capture article
• After using both filtering techniques, if the capture still appears to have issues you can submit a ticket to tech support with the 3mc capture file attached for feedback by following the process in this article: Troubleshooting and Tech Support.