These are chat archives for epnev/ca_source_extraction
Matlab implementation of a source extraction and spike inference algorithm for large scale calcium imaging data analysis, based on a constrained matrix factorization approach.
Hi @ivosonntag the Sources2D demo is somewhat outdated. I'd recommend you to look either at
demo_memmap.m or the complete
run_pipeline.m that also use memory mapping to deal with large datasets. By memory mapping you can load smaller patches of the dataset in parallel and process them locally, thus significantly reducing the memory requirements. Take also a look at the wiki to see more details about these approaches. I plan to update the Sources2D to reflect the current state of the code since it is easier to work with a class, but it hasn't happened yet. However, Sources2D still has an upsample function that you could use before deconvolution/DF_F extraction.
The value P.b refers to the baseline for an individual component and P.cs_ftrs is for the case where the data is the concatenation of multiple individual trials with gaps between them (nothing to worry about). P.b gets a value when you call update_temporal_components.
run_CNMF_patches. If your changes have changed this file, then better to send it to me privately, otherwise feel free to put in a PR.