I describe below something I have been putting together for my own
needs. The purpose of this email is twofold
(i) am I reinventing the wheel (if so where is that wheel?)
(ii) would this be of interest to others.
Regards
Jeremy Craven
I want a good way for keeping track of (for instance) which roi (zip)
file goes with which image if I have a large (or even small) set of
images, and also how to just keep track of filenames in a series or
collection and move easily from one spectrum to the next.
I come from an NMR background where it tends to be quite standard to
have a set of (often 2D) spectra and/or "peak picks" for each spectrum
in a series. Sometimes that series is a logical series (such as against
time, or a variable NMR parameter, or a variable sample parameter [such
as ligand concentration, or temperature]). This is essentially the same
as multipoint roi picking in ImageJ/Fiji.
When developing collaborations with people generating equivalent imaging
datasets I have not found that fiji/imagej has much to offer beyond the
roi zip files, and also using rois associated with slices of an ND
dataset. It is quite cumbersome to keep the name of roi files tightly
linked to the corresponding image.
I have therefore developed my own simple system based round an alias
text file that links spectra to ROI files and a simple but extensible
GUI (basically a set of configurable buttons) that allows this to be
linked to operation such as picking and/or editing (and autonaming)
ROIs, or making classifications. As one steps forwards and backwards in
the series the roi file is automatically saved and the new one reloaded.
The alias file is simple but flexible and so far this has very naturally
lent itself to dealing with yeast cell budding, zebra fish neuromast
analysis, and analysing hyphal growth.
This also allows effectively simple automatic documentation of the
association of (often complex) filenames from microscopes with a simple
logical variable (such as strain and image number). Again from an NMR
perspective I have found this invaluable for taking grunt work away from
the user, and for helping keep projects in good order and keep
supervisors clear of where data is.
The system I have is deliberately simple text file oriented (with the
GUI only allowing shortcut buttons) so that it does not end up with
undecipherable "project files" containing endless metadata, and is aimed
at being heavily (python) rescripted. It is also based on a mindset that
there are a lot of image processing jobs where human intervention is
essential, but where we want the drudgery of that to be removed as much
as possible so that the human does what they are best at, i.e. subtle
image analysis. It also aims to add as little as possible to the
familiar Fiji/ImageJ environment.
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
