What this looks like in practice
Three projects with different problems, different instruments, different contexts.
"Growing roots drift out of the field of view in under an hour at high resolution. Overnight experiments required someone at the microscope all night."
The lab needed long time-lapses of growing root tips, often running for hours or overnight without an operator present. Roots move continuously, and the microscope has no built-in way to follow them.
We built a closed-loop tracking system: after each image, the software estimates how far the root tip has moved by comparing it to the previous image, then repositions the stage to keep it in frame. Sessions can be paused and resumed across workstations. The dataset comes out as a continuous time-lapse, not a collection of manually corrected fragments.
"AFM and LSM were two independent systems. They couldn't capture the same event at the same time. What was needed was simultaneity."
The lab studies virus-cell and immune-cell interactions at the nanoscale, combining AFM topography with fluorescence imaging. The two instruments shared the same frame but had no native synchronisation, so post-hoc alignment could correct drift but not reconstruct what happened simultaneously in both modalities.
We built a synchronisation loop: the LSM sends a TTL trigger to start each AFM line; when the AFM completes a line, it writes a completion file that ZEN detects and uses to trigger the matching optical acquisition. The two systems run in alternating rhythm, coordinated at the line level. AFM lines are stitched automatically and displayed alongside the corresponding fluorescence data.
"Analysing multiple oil filters in a session required separate setup steps for each one. Manual configuration, separate exports, no consistent structure across runs."
The lab analyses particles from used lubricant oil to monitor wear in industrial equipment. Each session involved multiple filters, and the previous process required setting up, running, and exporting each one separately, with results that were inconsistent between sessions.
We built a single guided workflow for the full session. The operator confirms focus once per filter; the system handles tiled acquisition, particle segmentation and classification, and structured output with consistent folder naming and labeled representative images per defect class.
