Open OME-TIFF in CellProfiler?
Posted: Tue Jan 17, 2017 2:19 pmhi,
i have created a single large OME-TIFF that describes sparse time lapse data (thanks to Roger Leigh's response to me on this forum some months ago). Sparse means that there are some time points that do not have images for all channels (the reason: phototoxicity in long-running imaging of live cells). in these cases i instruct the reader (via UUID and IFD reference in the XML header) to either use a blank pane or the previous available image in that channel.
problem: to my knowledge, CellProfiler doesn't know how to read this information (at least the LoadImages and Metadata modules don't), even though i can import and view the stack successfully in Fiji and Omero. this is true not only for my script-created, sparse OME-TIFF but also for the sample file here:
http://downloads.openmicroscopy.org/ima ... es.ome.tif
When CP loads that file, it derives 21 time points and 1 channel rather than 7 time points and 3 channels. Feedback so far from CP is that some pre-processing will be necessary that essentially creates a new header with the correct dimensional counts (t = 7, c = 3, etc), e.g. by modifying the file's properties in FIJI and saving. however this means replacing the OME-XML (which also has these counts) with this simpler header.
does anyone have experience loading 4D data from an OME-TIFF into cellprofiler for processing? i hope i'm overlooking something obvious. this working was one of my giant assumptions.
so far i am trying to come up with a way to retain the tidiness of keeping images and metadata together (in Omero) while still allowing CP to load this correctly. i can create per-channel tifs and use a companion.ome file, but still, each channel's tif has a header that CP apparently won't understand. since i have only one Z, this is a 1 dimensional stack at this point, so that's fine. but in the case of my sparse data, the benefit of references to a "placeholder" image when one isn't present disappears - most likely i'll need to fill all the holes with copies of that image (could i compress only these "blank" images?).
thanks in advance
david
i have created a single large OME-TIFF that describes sparse time lapse data (thanks to Roger Leigh's response to me on this forum some months ago). Sparse means that there are some time points that do not have images for all channels (the reason: phototoxicity in long-running imaging of live cells). in these cases i instruct the reader (via UUID and IFD reference in the XML header) to either use a blank pane or the previous available image in that channel.
problem: to my knowledge, CellProfiler doesn't know how to read this information (at least the LoadImages and Metadata modules don't), even though i can import and view the stack successfully in Fiji and Omero. this is true not only for my script-created, sparse OME-TIFF but also for the sample file here:
http://downloads.openmicroscopy.org/ima ... es.ome.tif
When CP loads that file, it derives 21 time points and 1 channel rather than 7 time points and 3 channels. Feedback so far from CP is that some pre-processing will be necessary that essentially creates a new header with the correct dimensional counts (t = 7, c = 3, etc), e.g. by modifying the file's properties in FIJI and saving. however this means replacing the OME-XML (which also has these counts) with this simpler header.
does anyone have experience loading 4D data from an OME-TIFF into cellprofiler for processing? i hope i'm overlooking something obvious. this working was one of my giant assumptions.
so far i am trying to come up with a way to retain the tidiness of keeping images and metadata together (in Omero) while still allowing CP to load this correctly. i can create per-channel tifs and use a companion.ome file, but still, each channel's tif has a header that CP apparently won't understand. since i have only one Z, this is a 1 dimensional stack at this point, so that's fine. but in the case of my sparse data, the benefit of references to a "placeholder" image when one isn't present disappears - most likely i'll need to fill all the holes with copies of that image (could i compress only these "blank" images?).
thanks in advance
david