Skip to content

Track objects and spots over time

Create a trajectory table from repeated detections or segmented objects, then review the links against the source series.

New in 0.16.0a3

Detect Spots per Frame, Build Tracks, and trajectory review are available in this release. They run on the CPU and support one explicit scalar TYX or TZYX series. They do not infer divisions, fusions or biological identity.

Choose what to follow

Starting data Workflow
Bright spots Image Source → select one channel → Detect Spots per Frame → Build Tracks
Repeated intensity patterns Time series plus a fixed spatial template → Detect Spots per Frame (Template match) → Build Tracks
Segmented objects Per-frame labels → Measure Objects → Build Tracks

Select one channel upstream; do not remove T. Each XYZ volume is processed as one frame. If image drift needs correction, explicitly register and apply the transforms before detection or segmentation. Linking does not automatically correct drift.

  1. In Detect Spots per Frame, choose Local peaks or Template match. For template matching, connect a fixed YX/ZYX template with matching spatial sampling. Set the minimum value and separation as described in detection controls.
  2. Calculate and review detections. Maximum detections per frame is a cap, not a count estimate. Increase it if any frame is truncated; Build Tracks refuses capped observations.
  3. Connect the complete table to Build Tracks. Set Maximum displacement / frame, choose pixel or physical distance, and initially leave Maximum missing frames at zero.
  4. Calculate. The two outputs are Tracked observations and Track summary. Use Display Output to inspect/export either table.
  5. Choose Review trajectories… on the node card, or Review trajectories on source… in the inspector. Select a row or current-frame marker to navigate to its T and Z. Sorting the table does not change its identity.

The card shortcut sits beside Open measurements… on Detect Spots per Frame and Build Tracks. It also appears on Measure Objects and Measure Objects + Intensity when their calculated table carries time-series observations. It opens that node's existing read-only source/table review without calculating or changing the selected node. See review availability if the cached result or source is no longer current.

For labels, use Measure Objects or Measure Objects + Intensity on the explicit time series. The original index centroids and label IDs are retained in the observation table. Label IDs are local to a frame; equal IDs at different times are not assumed to be the same object. Imported labels must use a supported label-output route; do not threshold/relabel them merely to preserve their IDs. The supplied object example explicitly segments its synthetic source, so its connected-component IDs may change between frames.

Review gaps and uncertainty

Maximum missing frames = 1 permits linking across one absent frame. The distance allowance is multiplied by elapsed frames; it does not predict motion. Adjacent-frame candidates have priority over older gap candidates.

  • Cyan trails join observed positions. Dashed segments span missing frames; they are not measured or interpolated intermediate positions.
  • Orange trails identify observations with competing feasible links. Inspect them; the flag is not a probability and does not catch every possible error.
  • Yellow highlights the selected track. In 3D, only links with both endpoints on the selected Z plane appear. This is not a projection.

Crossing objects can exchange identities. Increasing the displacement or gap allowance can create plausible but incorrect links. Check representative sequences against independent annotations before drawing biological conclusions.

The preview fits the available window. Trail frames, time/plane selection, table sorting and decimal display affect only review. Changed sources or settings disable the old view; Refresh view loads an already calculated current result and does not start a calculation.

Try the synthetic examples

Open 2D Spots, Missing Frame and Crossing Review to inspect isolated trajectories alongside a crossing pair and an empty frame. Compare the uncertain crossing, rather than assuming the displayed IDs establish identity. Open Anisotropic 3D Object Trajectories for XYZ objects with unequal spatial sampling, a missing observation and a later appearance.

See tracking contracts and output columns for units, linking rules, saved evidence and limitations.