This dataset includes images captured during the experiments, processed images obtained through image analysis, the MATLAB code used for image processing, and detailed analysis results in .csv format.

The dataset contains seven experimental folders. Each experimental folder includes two imaging periods: Pre-Biomineralization (PB) and Post-biomineralization (PBM). Within each period, both raw images and segmented images are provided.

In addition, two videos were recorded during each imaging period. The velocity analysis performed using these videos is also available within the corresponding period folders.

Furthermore, all videos from all experiments are also provided together in a separate folder.

Raw / Segmented Image Naming

Files ending with _ch00.tif are images acquired from the Brightfield (BF) imaging. Files ending with _ch01.tif are images acquired from the GFP filter set. Files ending with _ch02.tif are images acquired from the DAPI filter set. Files ending with _overlay.tif are overlay images combining the GFP, BF, and DAPI filters.

Video analysis (Guide for the files provided in the Video/Results folders)

1) CSV Outputs (quantitative results)

1.1) track_speeds_summary.csv

Contains one row per track (i.e., one consistently linked bacterial object over time). Typical columns include:

  TrackID: unique track identifier
  NumFrames: number of frames in which the track exists (track length)
  MeanSpeed_um_s: mean speed of the track (µm/s)
  MedianSpeed_um_s (optional): median of step speeds within the track

This file is used for mean track speed boxplots and comparisons.

1.2) track_speeds_steps.csv

Contains one row per step (i.e., movement between two consecutive detections of a given track). Typical columns include:

  TrackID
  FrameFrom, FrameTo
  dt_s: time step between frames (s)
  dx_um, dy_um: displacement components (µm)
  ds_um: total displacement magnitude (µm)
  Speed_um_s: step speed (µm/s) = ds_um / dt_s

This file is used for velocity histograms (all steps) and for detailed quality assessment.

Notes

Pixel-to-micrometer conversion uses pixelSize = 0.162 µm/pixel. dt_s is the frame interval used during processing (e.g., 2 s). If tracking parameters are changed and the analysis is repeated, these CSV files will also change accordingly.

2) PNG Outputs (images for QA / visualization)

2.1) mask_after_staining_*_overlay.png (binary segmentation masks)

Black background / white objects. Each file corresponds to one time-series frame. White pixels represent segmented bacterial objects used for tracking. These masks are the direct input for centroid extraction and object linking.

2.2) frame###_trackIDs.png (per-frame track ID overlays)

Examples: frame001_trackIDs.png, frame002_trackIDs.png, ... Shows the original frame (or overlay preview) with TrackID labels drawn on top. Used for manual verification that IDs correctly follow the same objects over time and to identify possible mis-linking (ID swaps, jumps, false merges/splits).

2.3) plot_velocity_hist_all_YYYYMMDD_HHMMSS.png

Histogram of all step speeds (from track_speeds_steps.csv). Represents the distribution of instantaneous (step-wise) speeds.

2.4) plot_mean_speed_box_YYYYMMDD_HHMMSS.png

Boxplot of per-track mean speeds (from track_speeds_summary.csv). Each data point corresponds to one track’s mean speed. Used as the main summary metric for comparing mobility / immobilization across conditions.

3) Quick interpretation guide

Many values near 0 in the step-speed histogram, or a low median in the mean-speed boxplot, indicate a large fraction of low-mobility / near-immobile tracks. A wide interquartile range (IQR) and many outliers indicate strong heterogeneity, i.e., a mixture of slow- and fast-moving tracks. Suspicious distributions should always be cross-checked using the frame###_trackIDs.png files to confirm that IDs do not jump between objects.

4) Processing notes (traceability)

This output set was generated by a velocity tracking script that:

  segments bacteria in each frame,
  extracts centroids (geometric centers) for each object,
  links detections across frames into tracks based on spatial proximity and object-area similarity,
  computes step and track speeds, and exports the outputs described above.

If tracking appeared incorrect with default settings, maxLinkDist_um was adjusted (e.g., to 200 µm) to better match the observed motion between frames and to reduce artificial track fragmentation or ID switching.

Image segmentation code: Each experimental folder contains an image analysis code file named Image_processing.m. Detailed statistical data are also provided there in files named Statistical_Data_EXPERIMENTNAME.xlsx.

Video naming

The videos are named according to the following information: Video ID, experimental condition, replicate number, imaging stage, and recording number.

Video ID: unique identifier assigned to each video Experiment / Condition: microfluidic system and surface condition used in the experiment Replicate: experimental replicate number Stage:

  PB = Pre-biomineralization
  PBM = Post-biomineralization

Recording: recording number within the same replicate and stage

The video naming scheme is summarized below:

Video S1–S4: PDMS (aerobic), replicate 1

  S1: PB, recording 1
  S2: PB, recording 2
  S3: PBM, recording 1
  S4: PBM, recording 2

Video S5–S10: PDMS (O2-limited), replicates 1 and 2

  S5: replicate 1, PB, recording 1
  S6: replicate 1, PB, recording 2
  S7: replicate 1, PBM, recording 1
  S8: replicate 1, PBM, recording 2
  S9: replicate 2, PBM, recording 1
  S10: replicate 2, PBM, recording 2

Video S11–S14: Glass flow cell (untreated), replicate 1

  S11: PB, recording 1
  S12: PB, recording 2
  S13: PBM, recording 1
  S14: PBM, recording 2

Video S15–S18: TPS-treated glass flow cell, replicate 1

  S15: PB, recording 1
  S16: PB, recording 2
  S17: PBM, recording 1
  S18: PBM, recording 2

Video S19–S22: APMDES-treated glass flow cell, replicate 1

  S19: PB, recording 1
  S20: PB, recording 2
  S21: PBM, recording 1
  S22: PBM, recording 2