ghardo/scientific_data_2026_images

Image data for Hardo, Li, and Bakshi, 2026

Dataset summary

This repository contains the metadata-complete trench image store for the paper "An annotated timelapse imaging dataset on dormancy exit dynamics of Escherichia coli cells in Mother Machine".

The data consists of one xarray-compatible Zarr v2 store:

• 20260307_SB7_exit_snake_V4_1_with_metadata.trenches.zarr

The store occupies approximately 20 GB on disk.

This store contains extracted mother-machine trench movies together with trench geometry, frame timing, stage position metadata, acquisition metadata parsed from the source ND2 file, and acquisition event tables.

What is in this repository

The main image array is:

• data(Trench, T, C, Y, X) with shape (2795, 721, 2, 164, 34)

where:

• Trench indexes extracted trench movies
• T indexes timepoints
• C indexes imaging channels
• Y, X are trench-local pixel coordinates

The two channels are:

• PC for phase contrast
• mCherry

Acquisition details recorded in the store metadata include:

• calibrated pixel size: 0.107869821220548 µm
• acquisition interval: approximately 30 s
• total duration: 721 frames over approximately 6 h
• number of source fields of view: 28
• phase-contrast exposure: 50 ms
• mCherry exposure: 100 ms
• phase-contrast filter wheel setting: 475/20
• mCherry excitation: 594 nm
• mCherry emission filter wheel setting: 632/60

The root store also includes trench-level and frame-level metadata such as:

• fov_index, fov_name, lane_index
• trench crop bounds: x_left, x_right, y_top, y_bottom
• original_p_index
• relative_time_ms
• pfs_offset
• stage_position_um
• orientation, needs_flip

Child groups provide normalized metadata:

• /source_frames for per-FOV frame metadata
• /events for frame-linked and acquisition-linked event tables
• /acquisition for ND2-derived acquisition metadata and lossless raw JSON

Experimental context

The data show E. coli K-12 MG1655 cells imaged in a mother-machine microfluidic device during dormancy exit. Cells were maintained under spent LB plus pluronic for the first 60 min, then switched to fresh LB plus pluronic to induce growth resumption. Imaging was performed on a Nikon Eclipse Ti2 microscope using a Plan Apo 40x / 0.95 NA phase-contrast objective with 1.5x post-objective magnification and a Hamamatsu ORCA-Fusion camera.

This repository contains the extracted trench image data and metadata only. Segmentation masks, trained models, synthetic training data, and curated lineage examples are companion assets in the broader project and manuscript, but they are not part of this image-store repository unless added separately.

Data format and access

This is a directory-style Zarr v2 store. It is best accessed with zarr, xarray, dask, or viewers that can open Zarr microscopy data.

Minimal zarr example

import zarr

root = zarr.open(
    "20260307_SB7_exit_snake_V4_1_with_metadata.trenches.zarr",
    mode="r",
)

images = root["data"]                    # (Trench, T, C, Y, X)
channels = root["C"][:]                  # ['PC', 'mCherry']
times_ms = root["relative_time_ms"]      # (Trench, T)
stage_xyz = root["stage_position_um"]    # (Trench, T, Axis)

xarray example

import xarray as xr

ds = xr.open_zarr(
    "20260307_SB7_exit_snake_V4_1_with_metadata.trenches.zarr",
    consolidated=False,
)

print(ds)
print(ds["data"])

The store was written in an xarray-oriented layout and can also be explored as a datatree with child groups such as /source_frames, /events, and /acquisition.

Repository structure

20260307_SB7_exit_snake_V4_1_with_metadata.trenches.zarr/
├── .zattrs
├── .zgroup
├── .zmetadata
├── data/
├── source_frames/
├── events/
├── acquisition/
├── Trench
├── T
├── C
├── Y
├── X
└── trench-level metadata arrays