A platform for building trustworthy imaging pipelines.
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.
Cell Painting
High confidenceIngest
Read raw 5–6 channel HCS plate images from proprietary microscope formats and convert them into harmonized, chunked, plate-aware OME-Zarr stores. This stage preserves the full plate/well/site/ channel hierarchy so every downstream per-cell measurement remains traceable to its acquisition context, and validates completeness of the well-site matrix before any analysis begins.
Operations
- Vendor file discoveryScan plate directories and companion metadata files for expected wells, sites and channels; detect missing fields.
- Format detection and reader selectionIdentify vendor format (OME-TIFF, CZI, ND2, FLEX, Opera Phenix/Harmony, LIF, etc.) and select the appropriate Bio-Formats or Python reader.
- Channel-map harmonizationMap vendor-specific channel names/indices to the canonical Cell Painting biological channel labels (DNA, ER, RNA, AGP, Mito, optional Brightfield).
- Chunked OME-Zarr conversionWrite analysis-ready OME-Zarr with multiscales, per-channel metadata, plate/well/site axes and zattrs compliant with OME-NGFF v0.4+.
- Checksum and integrity validationVerify file readability, pixel-dimension consistency, missing-well accounting and SHA-256 checksums; fail closed on incomplete data.
- GPU-accelerated decode (optional)optionalUse cucim or CLIJ2 for GPU-accelerated TIFF/Zarr I/O when throughput bottleneck is on disk-to-memory transfer at scale.
Inputs
- raw 5–6 channel HCS plate images (multiwell, multisite, multichannel)OME-TIFF · CZI · ND2 · FLEX · LIF · TIFF · PerkinElmer Opera Phenix exports
- plate acquisition folders, companion XML and optional plate mapCSV · XLSX · JSON · XML
Outputs
- harmonized plate image store with multiscales and channel metadataOME-Zarr · OME-TIFF
- ingest manifest and file inventory with per-well statusYAML · JSON · Parquet
QC checks
- Well/site completeness — Verify the ingested well-site matrix matches the expected plate layout; report missing and extra sites.
- Channel count per site — Confirm every site has exactly the expected number of channels (5 for standard Cell Painting, 6 with brightfield); fail on channel dropout.
- Image dimension consistency — Verify all sites have identical pixel dimensions (X×Y) and bit depth; flag size mismatches that would break tiling.
- Pixel intensity saturation check at ingest — Compute fraction of saturated pixels (at detector max) per channel per site; high saturation at ingest flags overexposure before correction.
- Checksum integrity — SHA-256 hash verification of raw input files to confirm lossless transfer and bit-identical reads.
Failure modes it guards against
- Missing wells or sites (plate dropout) — Fail closed with per-well status table; require minimum fill fraction before proceeding.
- Channel name mismatch — Require explicit channel-map validation step; cross-check against JUMP/Cimini 2023 canonical channel order.
- Partial file reads (truncated images) — Validate pixel-dimension and checksum on every file; quarantine and report failures.
- Vendor metadata loss — Extract OME-XML companion or embedded metadata at ingest; fall back to plate-level config YAML if vendor metadata absent.
Tools (13)
Read >150 proprietary and OME microscopy formats; preserve vendor metadata and OME-XML companion.
Python image reader/writer for microscopy formats and OME-Zarr/TIFF workflows.
Convert, validate and write OME-Zarr/OME-NGFF datasets; generate multiscales and plate attrs.
- bioioOSS · BSD-3-Clause
Successor ecosystem to AICSImageIO; modular plugin-based bioimage I/O for OME-Zarr and microscopy files.
- bfioOSS · MIT
Blocked, tiled bioimage file I/O wrapping Bio-Formats for scalable, chunk-parallel workflows.
- zarrLibrary · MIT
Chunked N-dimensional array storage backend for OME-Zarr; supports local, S3, GCS, Azure.
- tifffileLibrary · BSD-3-Clause
Read/write TIFF, BigTIFF, OME-TIFF and pyramidal variants; zero-copy mmap for large stacks.
- FractalOSS · BSD-3-Clause
Distributed bioimage workflow orchestration centered on OME-Zarr plates; manages Cell Painting ingest tasks at scale.
- OMEROOSS · AGPL-3.0
Image data management server; centralizes storage, metadata, access control and rendering for bioimaging.
- cucimOSS · Apache-2.0
GPU-accelerated image I/O and processing (TIFF, Zarr); NVIDIA RAPIDS ecosystem.
- Revvity HarmonyCommercial · commercial
Vendor-native analysis platform for Opera Phenix/Phenix Plus; reads proprietary plate exports directly.
- MetaXpressCommercial · commercial
Molecular Devices HCS image acquisition and analysis; native reader for ImageXpress plate formats.
- Fiji (ImageJ)OSS · GPL-2.0-or-later
Macro-based batch scripting for format conversion, metadata extraction and preliminary QC before OME-Zarr conversion.
Metadata captured
| Field | Type | Example | Standard |
|---|---|---|---|
| plate_id | string | BR00117035 | OME-NGFF |
| well_id | string | B03 | OME-NGFF |
| site_id | string | s4 | OME-NGFF |
| channel_map | object | {'DNA': 'C1', 'ER': 'C2', 'RNA': 'C3', 'AGP': 'C4', 'Mito': 'C5'} | OME-XML |
| pixel_size_xy | float (micrometer/pixel) | 0.65 | OME-XML |
| microscope_model | string | Opera Phenix Plus | REMBI |
| objective_na | float | 0.75 | OME-XML |
| bit_depth | int (bit) | 16 | OME-XML |
| acquisition_software_version | string | Harmony 5.3 | REMBI |
| exposure_times_ms | object (millisecond) | {'DNA': 100, 'ER': 150, 'RNA': 200, 'AGP': 300, 'Mito': 200} | OME-XML |
| binning | int | 1 | OME-XML |
| wells_expected_vs_found | object | {'expected': 384, 'found': 382, 'missing': ['P24', 'A01']} | — |
| ingest_checksum_sha256 | string | a3f2...c9 | — |
| zarr_chunk_shape | array (pixels) | 1, 1, 1, 1024, 1024 | OME-NGFF |
| software_versions | object | {'bioformats': '7.x', 'ome-zarr-py': '0.9.x', 'python': '3.11', 'container': 'sha256:...'} | REMBI |
You get
- Plate OME-Zarr · OME-Zarr
- Ingest manifest · Parquet/YAML
References
- OME-NGFF: a next-generation file format for expanding bioimaging data-access strategies paper
- OME-NGFF specification spec
- REMBI: Recommended Metadata for Biological Images paper
- Cell Painting, a high-content image-based assay for morphological profiling (Bray et al. 2016) paper
- Image-based Profiling Handbook docs
- JUMP Cell Painting dataset — Three million images (Nature Methods 2024) dataset