ReadImageMetadata

ReadImageMetadata reads an image and returns lightweight metadata for ingestion checks. It reports shape, dtype, ndim, and a pragmatic axes guess derived from dimensionality, generated channel names, and available pixel-size metadata.

Inputs

  • input_image: image file to inspect.

Outputs

  • shape, dtype, and ndim: array metadata read from the file.

  • axes: pragmatic axis guess using common BioImageFlow defaults: YX, ZYX, CZYX, and TCZYX.

  • channel_names: generated names such as channel_0 when a C axis is inferred.

  • pixel_sizes: X, Y, and Z physical pixel sizes when TIFF metadata exposes them; otherwise None.

Dependencies and Core Libraries

imageio, tifffile, NumPy, and Python XML parsing for OME-TIFF metadata.

Assumptions

This is a lightweight reader. It does not normalize metadata across every microscopy container, and axis names are guesses unless the file format exposes stronger metadata.

Use it before layout validation or conversion when a workflow needs to branch or report basic image properties.

Minimal Example

from bioimageflow_core import Arguments
import bioimageflow_io_tools

metadata = bioimageflow_io_tools.ReadImageMetadata().process_row(Arguments(input_image="source.tif"))
assert metadata.axes == "CZYX"

Expected Results

Synthetic 4D fixtures report the expected shape, dtype, CZYX axes, generated channel names, and empty pixel sizes. Real OME-TIFF images should additionally populate pixel sizes when physical-size tags are present.

Failure Modes

Missing files, unsupported formats, unreadable paths, or unsupported dimensionality for the axes guess raise the underlying reader or validation error.