Using the napari-easytrack plugin for napari

The following steps are part of the typical napari-easytrack workflow.

1. Open napari and load your image data.

2. Open the Tracking widget from napari-EasyTrack plugin from the plugins menu.

3. Try out the different default presets for tracking by selecting them from the dropdown menu.

4. Click on Apply Tracking to run the tracking algorithm with the selected preset.

5. If the results are not satisfactory, you can try to improve the segmentation with our Clean Segmentation and Remove Small Objects buttons.

   5.1. The Clean Segmentation button keeps the the largest connected component per label on each frame, and reassigns smaller fragments and small labels to closest neighbour identifier (cell).

   5.2. The Remove Small Objects button removes objects smaller than a specified size threshold.

6. If None of the presets work well for your data, you can customize the tracking parameters by using our Parameter Tuning widget.

7. The Parameter tuning widget required a ground truth dataset to compare the tracking results against.

   7.1. You can load your own ground truth dataset by selecting it on the Ground Truth Layer dropdown menu.

   7.2. Select a Study Name. It is autogenerated, but you can customise it.

   7.3. Select the number of Trials (default is 128, which we think is a good number for most datasets).

   7.4. Select the Timeout (in seconds, default is 60s). For larger datasets or older computers, you might want to increase this value.

   7.4.1. If the Timeout is reached during a trial, a penalty score will be assigned to that trial. The penalty score is defined by the Timeout Penalty parameter (default is 10,000).

   7.5 Sampler is inherited from optuna library. Default is tpe. Other options is random.

   7.6 You can tick on Parallel to enable parallel processing (if your computer has multiple cores).

   7.7 Output Dir is the directory where the temporal tuning rusults will be saved. No need to change this unless you want to save the results in a specific location.

   7.8 You can modify the Voxel Size if your data is anisotropic (default is 1.0 for all dimensions).

   7.9 Click on Start Tuning to begin the parameter tuning process. This process can take around 30 minutes depending on your dataset size and computer performance. Also, the score should be below the Timeout Penalty value.

   7.10 Once the tuning is complete, it can be saved to a JSON file by clicking on the Save Config button.

8. You can use this presets file for future tracking tasks by selecting Custom Preset from the presets dropdown menu.