Loading...
 
Skip to main content
Scientific Volume Imaging B.V.

Scientific Volume Imaging

Deconvolution - Visualization - Analysis


Confidence Levels

A microscopy image file contains image data, that is, the intensity per voxel, but it often also contains so-called metadata: data describing the imaging conditions.

The metadata of a file may store parameters such as the sampling sizes, the emission and excitation wavelengths, in case of a confocal microscope the pinhole radius, and many others.

The Huygens software is capable of reading metadata from many types of microscopic files, and can therefore use the parameters for deconvolution.

Unfortunately, these parameters are not always fully reliable. For this reason Huygens uses a confidence scale to indicate the reliability of a certain parameter.

The Huygens confidence scale is as follows:

  • Verified: the parameter has been purposely verified by the user, confirming its correctness.
  • Reported: the parameter has been reported by the microscope manufacturer.
  • Estimated: the parameter might have some uncertainty such as a missing physical unit.
  • Default: the parameter was missing in the metadata and has been assigned a default value.

The level assigned to a parameter, default, estimated or reported depends much on the file type (h5, ics, lsm, etc). As an example, h5 and ics files can store all the necessary parameters. Other file types such as ome or zvi do store some but not all parameters, whereas in file types such as tiff and tiff-based file types usually not much, if any, metadata is saved. In these cases the microscopic parameters can be specified by templates.

When prompted to choose how to override a microscopic template, the user is asked whether or not to read parameters from the metadata of the image. If so, the user can select which level of confidence is acceptable. If the confidence level of a parameter in the file is below the selected level, the parameter is substituted from the template.

As a rule of thumb, the higher levels (reported, verified) are less risky, whereas the lower levels (default, estimated) are only recommended if the user is really sure that the metadata is correct.