Seeking input on volumetric and per-plane collection of measurements

features
3d

#1

Hello everybody,

We’ve started implementing volumetric image processing and analysis in CellProfiler. There are many decisions to make and we’d like to gather some input to help us with these decisions. Currently we’re working on loading multi-page TIFs as volumes (instead of the traditional stack of x-y plane images).

As we are developing this ability, we asked the question: In doing volumetric analysis in CellProfiler, would you expect measurements to be collected per-volume or per-slice? Or both?

Our preference is that volumetric pipelines should have volumetric measurements, and that no x-y plane measurements should be collected. An additional pipeline will be necessary If you wish to collect x-y plane measurements. We have several reasons for this:

First, though volumes are captured as a stack of discrete two-dimensional images that doesn’t mean they should be processed and analyzed discretely. Why impose the limitations of volume capturing on processing and analysis? Do measurements on discrete x-y planes of a volume make sense?

Second, it makes pipelines easier to develop and understand. We believe it would keep the interfaces of the measurement modules simple. Additionally, you can easily scan each pipeline and know exactly which measurements are collected volumetrically or per-plane.

Third, it’s easiest to implement. Which is a selfish answer from us, but it also benefits you because you’ll see volumetric processing be supported in stable releases of CellProfiler (e.g., those you can grab from the downloads page) sooner! And who doesn’t want volumetric processing, like, yesterday?

We’d like to know if collecting only volumetric measurements is acceptable for your volumetric profiling needs. How often would you be required to capture two-dimensional measurements on volumes?

We’d really appreciate your feedback!

:slight_smile:


#2

I am very excited to try out the 3D processing and analysis! Per-volume measurements would be most useful. Two-dimensional measurements (e.g. diameter, length, area) would be desired, but this again would be per-volume rather than per-slice. I haven’t thought of any scenarios for requiring per-slice measurements yet.


#3

Hello,
I you far from release ? I am really interested in this 3D volumetric process,
for my purpose I would like to be able to select the cell among a tissue, so maybe need to draw the edge of cells on the 2D plan and then calculate the volume of those cells
good luck, we are all supporting you ^^


#4

Thanks a lot! Check out the beta release here: http://cellprofiler.org/releases/

You may need to right-click (on the Mac) and choose Open to over-ride warnings to get it going. I believe a new beta release will be out Monday!