Best method for image segmentation for time-series images (50+)?

I apologize if this is somewhat repetitive. I am very, very new to image segmentation and do not have much background on machine learning. I posted a similar question here (Cell segmentation methods for time-series images?). However I am now analyzing time-series data that is not in video format, but rather individual images at different times.

I am measuring fluorescence for cells using time lapse images. For each sample, I have 50-100 cells. I currently manually select the ROIs (individual cells) using HCImage and measure the change in fluorescence. I generally have 50+ images that are 3 seconds apart for one sample. I select the majority of the ROIs in the time 0 seconds and time 3 seconds images and then scroll through the stack to select any ROI I didn’t notice before. The cells/ROIs do not move. However, the fluorescence of the cell/ROI does change during different times. I have attached images at time 0 seconds, time 3 seconds, time 6 seconds, and time 9 seconds as an example.

My question- Is there any program that would automatically select ROIs so I do not have to spend time manually examining the images at different times to select the ROI? I was recommended ImageJ, but I am not sure if I am able examine multiple images (50+ in one sample) in a reasonable amount of time. My cells are also clustered together in certain areas of the image (please see attached images). In addition, is it possible to train a sample and use the same parameters/instructions to select ROIs in other samples?

I am new to image segmentation and have no prior experience. Thank you for your help.
*Images removed at request of author.

I would suggest that you start by reading the literature on calcium imaging, where you will come across lots of different algorithm to address the type of data you are generating. You will have to figure out which one will work best/is most appropriate for your data. A quick search for ImageJ, calcium imaging and plugin immediately provides a series of options, e.g

There are also lots of tools outside of ImageJ such as NETCAL (MatLab;

Some of these also include segmentation algorithm, but that may be the challenging part for your data (depending on what ROIs/cells you are actually interested in). If you are only interested in the really bright cells (1 in image ‘time 3’ and 2 in image ‘time 6’), this shouldn’t be too challenging. You could start with a maximum intensity projection of your time series to identify the brightest spots, use that to segment the image and then analyse each ROI over time. However, it looks from the sample images like the pixels in the bright cells may be saturated, which wouldn’t be good if you are interested in the fluorescence intensity.
Finally, please try to avoid repeatedly posting the same question in the hope of getting a quicker/different response. If you need/want to clarify some aspect on your original post, edit the original post rather than re-posting the question.
Hope this helps,