Organoids / 3D

01 · Pipeline Engineering

Turn an experimental method into an executable system.

Microscopy analysis often begins as a working method understood by one scientist: instrument settings, processing decisions, and scripts developed around a particular experiment. Fovea Lab provides a structured environment for turning that method into a pipeline that can be run, examined, and developed over time.

A scientist can begin in the visual workspace, ask the agent to help develop a stage, or open the underlying code when direct control is needed. Each way of working modifies the same pipeline rather than creating a separate version of the analysis. The scientific purpose of the experiment remains connected to the computation, allowing the workflow to mature from early exploration into a system that another researcher can understand and use.

02 · Evidence Integrity

Make confidence part of the computation.

A final measurement does not reveal whether the source data were complete, whether a model behaved unexpectedly, or whether an important parameter changed between runs. Fovea Lab evaluates the analysis as it runs and preserves the evidence needed to understand the result afterward.

Checks are placed at the stages where failures can occur. Scientists can return from a result to the pipeline that produced it, inspect whether the run remained within accepted conditions, and trace unexpected behavior to the point where it entered the analysis. Verification becomes part of the workflow itself rather than a report reconstructed after the experiment is finished.

03 · Agentic Development

Let AI improve the workflow without taking control away from the scientist.

Fovea Lab gives agents access to the structured experimental context that governs the pipeline. Instead of producing isolated code without understanding the surrounding system, an agent can inspect a specific stage, propose a bounded revision, and evaluate its effect against criteria established by the scientist.

A proposed change remains separate from the approved workflow until it has been reviewed. Its effect can be compared with the current version, while the decision to adopt it remains with an authorized user. AI can accelerate pipeline development while the platform preserves accountability and a stable record of what was actually run.

Pipeline architectures, module by module.

Pipeline backbone — reshapes per modality. Select a module for its sub-steps, tools, and deliverable.

Organoids / 3D

High confidence

Ingest

Read multi-well confocal or light-sheet z-stack acquisitions for organoid, spheroid, or biofilm assays into a harmonized, chunked, analysis-ready representation with per-well and per-object indexing. The key challenges are proprietary vendor formats (CZI, LIF, ND2, TIFF series), large z-depth (50-200 planes), and correct physical calibration (voxel size XYZ, z-step).

Operations

Enumerate acquisition filesWalk plate/well directory tree or OMERO dataset; build a file manifest mapping plate/well/field/z/channel to file paths.
Read raw z-stacks via Bio-Formats / bioioLoad each z-stack preserving all planes, channels, and physical metadata (pixel size XY, z-step, objective, NA) using a format-aware reader.
Validate axis order and calibrationAssert TCZYX axis order; verify z-step is non-zero and consistent across wells; flag wells where z-spacing differs from the acquisition protocol.
Rechunk and write OME-ZarrWrite each well's z-stack as an OME-Zarr store (chunk size tuned for 3D access patterns, e.g. 1x1x32x512x512); build multiscale pyramid for large fields.
Attach OMERO or plate metadataoptionalLink well-level annotations (condition, passage, matrix, timepoint) from the plate map; write OME-NGFF well/plate metadata.

Inputs

Raw 3D z-stack acquisitions from confocal or light-sheet microscopeCZI · LIF · ND2 · OME-TIFF · TIFF series · LIF · VSI
Plate map / assay manifestCSV · JSON · XLSX

Outputs

Per-well harmonized 3D z-stacks with physical calibrationOME-Zarr · OME-TIFF
Ingest manifest with file-to-well mapping and validation flagsJSON · Parquet

QC checks

Incomplete z-stack detection — Flag wells where the number of z planes acquired is below the expected acquisition protocol value (e.g. < 70 of 80 expected planes).
Z-step consistency — Check that z-step is uniform across all wells; flag outliers >5% deviation from the plate median (indicates stage drift or protocol error).
Voxel calibration presence — Assert that physical pixel size (XY and Z) is non-null and non-zero in the OME-XML/NGFF metadata; fail the well if missing.
Missing well completeness — Verify that all expected wells have corresponding image files; report missing wells as failed.

Failure modes it guards against

Incorrect z-step written by vendor software — Cross-check OME-XML z-step against acquisition protocol; allow manual override in the ingest manifest.
Axis order ambiguity (CZYX vs TCZYX) — Enforce explicit axis labeling via bioio/ome-zarr-py; add a unit test that reads back a known z-stack and asserts shape/axis.
Partial z-stack (imaging stopped early) — Flag wells with n_z_planes < threshold; exclude or pad during preprocessing.
Metadata stripped by file copy — Use Bio-Formats or bioio readers that preserve embedded OME-XML; never use plain TIFF export from the microscope without metadata.

Tools (9)

Bio-FormatsPRIMARYOSS · GPL-2.0-or-later
Read >150 proprietary microscopy formats (CZI, LIF, ND2, VSI, TIFF series); preserves all vendor metadata including z-step and objective.
bioioPRIMARYOSS · BSD-3-Clause
Pluggable Python image IO framework for microscopy; successor to AICSImageIO with format-specific plugins (bioio-czi, bioio-ome-tiff, bioio-ome-zarr).
ome-zarr-pyPRIMARYOSS · BSD-2-Clause
Convert, validate, and write OME-Zarr/OME-NGFF datasets including plate/well layout and multiscale pyramids.
AICSImageIOOSS · BSD-3-Clause
Python image reader/writer for microscopy formats and OME-Zarr/TIFF workflows; predecessor to bioio, still widely used.
zarrLibrary · MIT
Chunked, compressed N-dimensional array storage; backing store for OME-Zarr.
tifffileLibrary · BSD-3-Clause
Read/write TIFF, BigTIFF, OME-TIFF and pyramidal image variants; handles large z-stacks with memory mapping.
OMEROOSS · AGPL-3.0
Image data management, metadata, rendering, and access control for bioimaging facilities; enables programmatic ingest via OMERO.py.
FractalOSS · BSD-3-Clause
Distributed bioimage workflow system centered on OME-Zarr plates; automates ingest + rechunking for HCS data.
ImarisCommercial · commercial
Commercial 3D/4D image visualization and analysis platform; can ingest CZI/ND2 and export to HDF5/IMS format for large volumes.

Metadata captured

Field	Type	Example	Standard
pipeline_run_id	string	organoids-ingest-2026-06-05-run01	—
software_versions	object	{'bioio': '3.3.0', 'ome-zarr-py': '0.10.2', 'python': '3.11', 'container': 'sha256:abc123'}	REMBI
input_dataset_uri	uri	s3://client-study/plate01/raw/	OME-NGFF
output_dataset_uri	uri	s3://client-study/plate01/zarr/plate01.zarr	OME-NGFF
plate_id	string	Plate_2026-05-12_001	OME-NGFF
well_id	string	C07	OME-NGFF
voxel_size_xyz_um	array (micrometer)	0.325, 0.325, 2.0	OME-XML
z_step_um	float (micrometer)	2.0	OME-XML
n_z_planes	int (count)	80	OME-XML
objective_magnification	float (x)	10.0	OME-XML
objective_na	float	0.45	OME-XML
channel_names	array	DAPI, GFP, Brightfield	OME-XML
acquisition_instrument	string	Zeiss LSM 900 confocal	REMBI
raw_file_format	enum	CZI	—

You get

Harmonized OME-Zarr plate store · OME-Zarr
Ingest validation report · JSON / HTML

References

Organoids / 3D: a pipeline architecture for trustworthy imaging.

Turn an experimental method into an executable system.

Make confidence part of the computation.

Let AI improve the workflow without taking control away from the scientist.

Pipeline architectures, module by module.

Ingest