A completely automated MATLAB-based trainable image cell segmentation, fluorescence quantification and analysis suite, particularly well suited for high-throughput time lapse fluorescence microscopy of in vivo bacterial cells.
I'm using SuperSegger to quantify fluorescence fluctuations between daughter cells. Since I care about the true pixel value of the images, I have to flatten the fluorescence image since the illumination of the field is not uniform. To do this, I usually take a handful of images of a homogeneously fluorescent slide as well as images of the camera shot noise. I then flatten the image via
where I_flat is the corrected image, I is the raw image, and I_field is the image of the fluorescent slide.
My current workflow is to preprocess the images (i.e. renaming), flatten the fluorescence channels I care about, and resave them as .tif files. This can have some pitfalls as it requires saving an image which, through the fattening, could have signed pixel values which then get forced to unsigned upon saving.
It would be great if one could add fluorescence flattening in as a CONST parameters. It would be able to take either a) a covariance matrix from fitting the illumination to a two-dimensional Gaussian or b) an averaged image of illumination profile and camera shot noise.
I would fork the repo and make the change myself, but I'm still not very familiar with the code base (plus, I am far more fluent in Python).
P.S. Your software is a lifesaver for me. Lineage tracking has suddenly become a piece of cake.
I'm using SuperSegger to quantify fluorescence fluctuations between daughter cells. Since I care about the true pixel value of the images, I have to flatten the fluorescence image since the illumination of the field is not uniform. To do this, I usually take a handful of images of a homogeneously fluorescent slide as well as images of the camera shot noise. I then flatten the image via
where I_flat is the corrected image, I is the raw image, and I_field is the image of the fluorescent slide.
My current workflow is to preprocess the images (i.e. renaming), flatten the fluorescence channels I care about, and resave them as
.tiffiles. This can have some pitfalls as it requires saving an image which, through the fattening, could have signed pixel values which then get forced to unsigned upon saving.It would be great if one could add fluorescence flattening in as a
CONSTparameters. It would be able to take either a) a covariance matrix from fitting the illumination to a two-dimensional Gaussian or b) an averaged image of illumination profile and camera shot noise.I would fork the repo and make the change myself, but I'm still not very familiar with the code base (plus, I am far more fluent in Python).
P.S. Your software is a lifesaver for me. Lineage tracking has suddenly become a piece of cake.