FLIM in OMERO

[dzmacdonald]

In collaboration with the BioInformatics group in Dundee we will add FLIM functionality to OMERO.

related to ​insight#1228

Current Workflows

Currently users using SPCImage have to go through a number of steps to generate their analysis results.

1. Draw an ROI around region of interest (cell, part of cell, cell cycle)
2. Perform analysis in SPCImage
   1. Select cut off for k1
   2. output results to excel where the k1, k2 values are transformed in to E'fret conditon, (1-k1/k2*100)
   3. Generate heat map from E'fret, each map must have min, max set by hand
3. Repeat for each region of interest
4. Combine results to get averages for different treatments, cell cycle.

Typically a user will generate 100 fret images per day, each days aquisiton can take up to 2 days to analyse.

OMERO Workflows

1. In OMERO the user should be able to draw the ROI in different regions on the image
2. select the fret, no fret images
3. Run analysis
   1. Generate the Heatmaps and upload/attach in OMERO
   2. Generate the summary results
4. Allow the visualisation of the chi value to see how good the fits are.
   1. Allow thresholding on chi value to remove from heatmap

This will save users many hours of copying and pasting from SPCImage to Excel, and then Excel to other graphing software.

[jmoore]

Notes from discussion Mar. 11:

FLIM

 - tasks to get done
  -- NDIM model
  -- get ready for scripts in anger
  -- get ready for new camera
  -- encorporating Pieta's script
   --- it's broken (mathematical, mixed variables)
   --- will take work to get it working
   --- basic version is running against OMERO
  -- optimizations for values like SPImage
  -- dealing with ROIs
   --- drawing them, storing them on the server
   --- some version of trunk (matches scripts)
   --- Donald will work on ROIs today
  -- visualization mechanism (heatmaps)
  -- rolling back the fake FLIM
  -- submissions
   --- submitting 200 images to server
   --- not running it on the server (nor in parallel)

 - time scales
  -- tried to talk to him 2 weeks ago (conference)
  -- he's doing it now. (starts today)

 - discuss NDIM Monday
  -- won't be implementing anything right away
  -- Donald: hacking something up
  -- Jean-Marie: if data comes as now, we're fine

 - tickets
  -- roi
   --- difference which rois for measurement and for viz
   --- namespaces

 - mongo
  -- 3 days away from insight writing to mongo
  -- omero pojos ok
  -- was modifying insight gateway
  -- Chris: means changing everything?
  -- ROI is the starting point (drawing)
  -- Jason: don't like the namespace "measurement"/"visualization" is something user needs to understand
   --- from users POV, need "infinite flexibility" of tagging
   --- Jean-Marie: is going to be extremely painful
  -- Chris: starting with a single keyword (before alternative storage)
  -- NB: tags don't have a namespace right now
  -- combo box drop down with the various workflows
  -- #namespace manager

 - preview
  -- john: get another nightshade
  -- get a grid engine
 
 - decisions
  -- move alternative storage out of sprint
  -- donald works on flim
  -- discuss the plan for next mongo
  -- next sprint: what else suffers?

 - end of this iteration
  -- good chance to have roi, flim working
  -- good chance of some visualization
  -- some heatmap, uploaded as an image or jpeg annotation
  -- may not be fast (on cluster)
  -- how much work to get scripting working?

 - iteration 5
  -- by end, flim done? no, but handing something to david by next week
  -- on the cluster (or optimized)
  -- (conservative and add a extra week, i.e. middle of week)

 - docs for clients
  -- Examples of different things in the scripts

 - things that have to be maintained
  -- matlab / cellprofiler

 - shooting for a middle ground
  -- have to decide what we're going for (4.2)
  -- is there any technological middle ground?
  --