Maximum image size allowed in CP2?

Dear all,

I have a couple of questions in running cellprofiler 2.0.

  1. what is the maximum image size that is allowed in CP2? I compiled a pipeline to process a large bright field image tile, ~20MB, and I got an error message: “error while processing IdentifyPrimaryObjects”. After playing with the pipelines, I figured that if I resize it to ~4MB, it went through all the pipelines without a problem. My computer has Intel Quard Core processor, 8G RAM, currently running Win 7 64bit. I also wonder if it has something to do with my computer hardware or if it is the CP2.0 itself.

  2. When I perform image based feature extraction, I notice that all images, either 8bit or 16bit images, are automatically converted to 0.0000-1.0000 when first read in. This is fine if it is 8bit image (0-255). However, when it comes to 16bit image (0-65535), I’m loosing some precision there, 1/65535= 1.53x10^(-5). I wonder if there is a way to keep the raw image settings when first read in CP2 and let user decide if he wants to convert the image to 0.0000-1.0000 range.

Thanks,

Er


Er Liu, Ph.D
Postdoc Scientist
La Jolla Bioengineering Institute
3535 General Atomics Court, Suite 210
San Diego, CA 92121
Ph: (858) 456-7500 x124
Fx: (858) 456-7540
Email: eliu@ljbi.org
ljbi.org/

Generally speaking, CP will have memory problems with images larger than 2000 x 2000 pixels or so. These issues also depend on the number of intermediate (and similarly) images generated by other modules.

The pixel values are stored as floating point numbers so we don’t anticipate a loss of precision when considering the values written to a spreadsheet or database. Keep in mind, though, that the object identification modules assume that the input image is scaled from 0 to 1.

Regards,
-Mark