Skip to content

Time Series

PanSeg imports, processes, and exports time series data. In the current version PanSeg does not take any advantage of the time data, it only processes the data frame-by-frame. Tracking and other processing across time points will be added in a future version.

Supported layouts and formats

The time series layouts are TYX (2D), TZYX (3D), and their multichannel versions TCYX and TCZYX.

Import supports OME-TIFF, h5, and zarr. For OME-TIFF, the input tab prefills the stack layout from the file's OME-XML, and it reads the time spacing from the timing metadata (TimeIncrement or Plane.DeltaT) if the file carries it. OME-ZARR support is planned (#622)

From multiple files

A time series can also come from a set of independent files, one image per timepoint. Select several files in the Open File dialog and they stack into one series in alphanumeric filename order. Each file must be a single image (no time axis of its own), and the stack layout must be spatial; the import adds the T axis itself. Multi-channel files stack per channel, with equal channel counts across files.

The spacing between the files is unknown at import. Set it in the input tab's Details field once you know the acquisition interval; export then writes it like any other time spacing.

Warning

A multifile OME-TIFF series is a different thing: one logical image whose planes are chained across files through the OME-XML (UUID and FileName references), or whose metadata lives in a .companion.ome file. It looks identical to a set of independent images in a file dialog. PanSeg rejects it. The reader would silently fill missing planes with zeros when only part of the set is present, and the plane order comes from the XML, not from the filenames, so filename order would scramble the series anyway.

Export writes the time axis back out. Tiff export of a time series is always OME-TIFF, named with the .ome.tiff extension per the OME-TIFF naming convention, with TimeIncrement when the spacing is known. H5 and zarr export carry the layout and the spacing with it.

Limitations

  • Training is currently not supported
  • Pre- and Postprocessing steps apply to all timepoints.
  • Data must fit into memory, but can be truncated during import (see Import)