Automated root tracking for long-term confocal imaging
INRAE, Institut Jean-Pierre Bourgin for Plant Sciences
Plant Observatory, Cytology and Imaging Platform · Philippe Andrey, Magalie Uyttewaal · France
Automated root tracking for long-term time-lapse imaging
INRAE's Plant Observatory uses long-term confocal imaging to study dynamic cellular processes during root growth. This project added automated root tracking so the same region could remain in view throughout unattended time-lapse acquisitions.
Capturing these events requires long-term live imaging at high spatial resolution, typically over 12 to 24 hours, on a ZEISS LSM 710 confocal microscope running ZEN Black.
Why growing roots leave the microscope field of view
At high magnification, a growing root tip could move out of the field of view in under an hour. ZEN Black on the LSM 710 has no built-in image-driven tracking.
Keeping the root in frame required manual stage adjustments throughout the session, which is not practical overnight. Datasets came out fragmented, which complicated downstream registration. The instrument was not designed to support unattended long-form tracking.
How image registration controls the microscope stage
SmartLabs built a closed-loop tracking workflow inside ZEN Black, using a VBA macro and a custom interface. After each acquisition the system passes the 3D and, optionally, the multi-channel images to INRAE-BIP (Biological Image Processing), software the customer already had, which computes the image registration and therefore the root-tip displacement. The macro repositions the stage in X, Y, and optionally Z before the next frame. The loop runs unattended for the full session.
At session start the operator defines safe stage boundaries. Every stage movement is checked against them, and if a move would exceed them, the system flags it and responds according to the preference set beforehand. Two acquisition modes run independently: high-frequency 2D frames for tracking, and lower-frequency z-stacks for the scientific data. Each cycle logs the timestamp, the displacement, the move issued, and any warnings. Sessions resume from the last saved state without re-calibration.
ZEN Black automation on a ZEISS LSM 710
ZEN Black exposes automation through a synchronous VBA macro engine with no native subprocess calls, which is more constrained than the Python OAD layer in ZEN Blue. The entire loop has to run inside that environment. Each tracking cycle follows four steps:
- Acquire a frame.
- Export it to a shared folder.
- Invoke the INRAE-BIP registration module and poll for its displacement output.
- Issue the stage move before the next acquisition triggers.
Communication between the macro and the BIP module uses a JSON file handshake: the macro writes a trigger, BIP writes dx/dy/dz, the macro reads it and issues the stage move. BIP stays fully independent and can be updated separately from the ZEN macro.
Bounds and safety
Stage bounds are marked by the operator at session start. Every computed move is checked before it is issued; if it would exceed the bounds, the system flags it and responds per the pre-set preference: stop and wait, or log and clip. A separate jump limit catches registration artefacts before they can become a stage command.
Two cadences in one loop
Two acquisition channels run within the same loop at independent cadences: high-frequency 2D frames for tracking, and lower-frequency z-stacks for scientific data. The macro arbitrates between them each cycle based on elapsed time. Every cycle writes a structured log entry to a per-root session folder. On resume, the macro reads the last written state and continues without re-calibration.
Results of unattended root imaging
| Before | After | |
|---|---|---|
| Session length | Limited to staffed hours | Overnight and multi-day, unattended |
| Dataset continuity | Fragmented by manual correction gaps | Continuous time-lapses, start to end |
| Photobleaching | Multiple illumination events per session | One correction per cycle, precisely bounded |
| Operator during run | Required throughout | Setup only |
Continuous datasets simplify registration and reduce the number of sessions discarded because of positional gaps. The instrument now runs during hours in which it previously could not. The solution required no changes to the LSM 710 or to the ZEN Black installation, and the INRAE-BIP module is used as-is, integrated through a standard interface.
"We already had the image registration side working. What we were missing was the connection back to the microscope. SmartLabs built that part around our existing tools rather than replacing them, which significantly simplified maintenance and developments on our end. We can now run long-term image acquisitions which we could not do before."
Other uses for image-driven stage tracking
The underlying problem is narrow: a specimen moves continuously during imaging, and the microscope has no native way to follow it. That occurs across biological research, and the same gap exists on other platforms and in other fields where samples change position during acquisition. Any organism that grows or migrates faster than the field of view allows at working magnification runs into it.
| Context | The problem we could help with |
|---|---|
| Zebrafish embryos, early development | Rapid tissue reorganization pushes the region of interest out of frame within a single session |
| Cell migration assays | Cells or clusters move across the substrate over hours while the field of view stays fixed |
| Fungal hyphae, pollen tubes | Tip growth is continuous and directional, the same constraint as root tracking |
| Organisms in liquid medium | Slow drift unrelated to biology still fragments datasets over long sessions |
| LSM 710 and 780 on ZEN Black | Same control environment, same absence of native image-driven tracking |
| ZEN Blue instruments | Different scripting environment (Python, OAD), but the closed-loop concept applies wherever stage control and image output are accessible |
| Widefield or spinning disk | Any system whose software exposes stage control and image output to scripting |
| Materials science, in-situ experiments | Samples that deform, grow, or shift position during heating, loading, or environmental change |