ROI detection/registration area is smaller than the full field-of-view

[sarahapark]

Hello!
I am having trouble with the ROI detection in one of my data sets where the ROI results cover a smaller area compared to the entire field-of-view. My other data sets were processed without any trouble. This is what the results looks like for the bad ROI detection.

I think the problem lies with the registration. Here is a screenshot of a frame using "View registered binary":

The ops settings were default settings except for save_mat = 1, reg_tif = 1, soma_crop = 0, diameter = 16, 16.
I also tried the following changes to the ops one variable at a time, in addition to the above mentioned changes, but none yielded better registration results that cover the entire field-of-view:
nimg_init = 200 or 600 (default 300)
batch_size = 250 (default 500)
block_size = 64, 64 (default 128x 128)
denoise = 1 (default 0)
two_step_registration = 1 and keep_movie_raw = 1 (default 0 and 0)
two_step_registration = 1 and keep_movie_raw = 1 and denoise = 1 (default 0 and 0 and 0)
maxregshift = 0.2 or 0.05 (default 0.1)
maxregshiftNR = 2 or 10 (default 5)
smooth_sigma_time = 1 (default 0)

Does anyone have any experience with a similar issue and can suggest what I could try to improve the registration/ROI detection?
Thank you!
Sarah