Simple image processing for high content screening assays
General information:
Currently I am aware of just one group trying to define an image standard for many microscope image data types. (OME Open microscopy environment). Storing image data is one thing (see data storage), analysing the images another.
I like developing routines in a graphical way, therefore being a Labview fan for a long time. The advantage of a graphical LabView programming/scripting environment is obvious (National Instruments). One simply works with the logical flow scheme of a program (not the code itself). The program displays its function in a flowchart, saving programming and code understanding time. (Very efficient if you have several students/engineers working on the same project). Once developed modules can be “recycled” simply by drag&drop from one “flowchart” to the other.
 
Rapid prototyping in a scientific environment?
In a scientific environment it is often difficult to predict, if a newly developed analysis routine meets the biological demands. Fast evaluation of different procedures is mandatory. I think a rapid prototyping environment complements any development platform to get a feeling for the real software needs and the initial “play phase”. Often many assays can be analysed with very simple routines (some people don’t like to hear that :-).
Trick is to find the right combination of rapid prototyping and sophisticated programming.

The following figures show you simple image processing routines used in screening applications:

a) Cell nucleus stained with DAPI or HOECHST

b) Simple automatic (metric) threshold

c) dilation of the detected nuclei

d) subtraction of the original nucleus
-> for very simple (but efficient) ER or cytosol measurements

Example: Measurement of intracellular transport rate

Below are some assay/screening examples we performed, using basic image processing routines:
For our assays/implementations we use mostly LabView vision libraries (rapid prototyping), but any software routines (imageJ, self made) are just fine.

2003: Golgi integrity assay image  (see “Starkuviene,V ;Liebel,U. et al.”  )
-> 8 novel interactors identified 

blue:Hoechst (cell nucleus); red:unknown/novel gfp-tagged human protein; green:golgi marker (6c6); image acquired fully automatic with the “cube” microscope (LUI microscope control software) and cell detecting autofocus (objective 40x NA 0.95)

image processing:
(+) golgi integrity measurement (fragment analysis)
(+) transfection detection (nucleus/cytosol)
(+) cell nucleus detection
(+) image processing was done with custom LabView Vision scripts

2000-2003
Screen of novel human proteins.
Do they have any effect on cell proliferation or cell morphology?
-> 14 novel poteins identified (paper: “Arlt et al.”)
 

stainings:blue:Hoechst (cell nucleus); red:Anti BrdU AB; green:unknown/novel gfp-tagged human protein; image acquired fully automatic with “cube” microscope and cell detecting autofocus (20x NA 0.7)

image processing:
(+) BrdU incorporation measurement
(+) transfection detection (nucleus/cytosol)
(+) cell nucleus detection
(+) image processing was done with custom LabView Vision scripts

2001-2003
Screen of novel human Er/Golgi proteins.
Do they have any effect on the intracellular transport machinery?
-> 21 novel proteins identified (paper “Starkuviene et al.”)
 

left: stainings:blue:Hoechst (cell nucleus);red:Anti vsv-g AB;green:unknown/novel gfp-tagged human protein;4th colour (not shown); VSV-G marker; image acquired fully automatic with “cube” microscope and cell detecting autofocus (20x NA 0.7)

right: image processing results of a “hit” (x-axis: transfection “intensity”, y-axis: transport ratio)
(+) transport ratio measurement
(+) transport marker measurement
(+) transfection detection (nucleus/cytosol)
(+) cell nucleus detection
(+) image processing was done with custom LabView Vision scripts