Breakout session on Standards for Quantifying Cell Dynamics
held at Workshop number 5, January 2016
Question: What are some of the challenges to standardizing quantitative cell biology?
- Need for Standardized samples
A major topic of discussion was the need for standardized samples that researchers could use to compare imaging methods and processing algorithms. Suggestions included:
- standardized microscopy calibration samples
- organism-specific standard samples, for example a standardized set of Drosophila embryos fixed and stained all in the same way.
- Pre-made fluorescence slides of samples, all prepared at once, that could be distributed around the world, allowing any results to be compared between groups.
- Develop a set of standard images to be used as benchmarks for analysis tools, so that one can compare results using the same set of data.
- Researchers who have already developed home-made internal standards should be able to publish them.
- Can we develop a prize for development of useful standard samples or image sets?
- AFM community has developed such standards for their work, we can learn from their approach.
2. Who will lead efforts on standardization?
- Private Institutes could take a lead role
- If some sort of accession number can be assigned to standards, then journals might move to require that information.
- For setting up a database of standards that could be made available, one possible example is the ImageJ plugin page, which allows users to find the plugins that they need while giving credit to those that developed them.
3. Can we develop a standard data format for representing cell shape and dynamics?
- must be able to handle 2D and 3D data, curvature, density, texture measures
- many different raw data types such as brightfield, fluorescence, electron microscopy
- need to represent not just the images but also the experimental procedures used for obtaining them, in a standardized machine readable format.
- in addition to raw data, do we also store and represent various processed forms and derived datasets? or is it better to simply store the necessary details to repeat any processing, using some standardized descriptors of image processing steps?
- while a variety of formats already exist for microscopy raw data such as Open Microscopy, there is less development of formats for data derived from images, such as motion and shape features.