TrackObjects for a Migration Assay

Hello,
I am currently trying to consistently track cells in a zvi movie file that contains approximately 350 images. This is for a migration assay to determine how far cells move toward a stimulus, as well as how linear their trajectory is. Ultimately, I’d like to be able to calculate the velocity of as many cells as possible in this movie file. The cell cytoplasm has been stained, which I realize can, unfortunately, complicate things, as well as the fact that my cells are of a high density. I have been using TrackObjects, as well as other modules to improve the image. I have gotten it to work decently, but I would still like to see an improvement in how well the objects are tracked. The linearity of an object can only be accurately tracked (I think) if the identification is consistent. My problem is that the identification of the cells doesn’t stay consistent for the entire 350 images for many of the cells. In one image, the cell may be labeled as 8, and then in the next, it might get merged with another object or may simply disappear. I was wondering if someone might be able to provide some advice on how to improve my pipeline, if possible.

LoadImages
Crop (I remove the back portion of the cells. I’m mainly interested in the first few rows of cells)
EnhanceOrSuppress features (To improve edges)
Apply Threshold (To remove background)
Identify Primary Objects
Track Objects
Save Images
Export to Spreadsheet
Conserve memory (to hopefully make everything go a little faster)

Any help at all would be appreciated. I have attached the actual pipeline. I can email a small sample sequence if necessary but I can’t upload it here since the forum won’t accept zvi file uploads.

Thanks for any and all help,
-Alex
Time Lapse Velocity Pipeline.cp (9.07 KB)

I tried uploading the sample sequence as a zvi and avi movie file but neither could be uploaded to the post. I have attached the sample image sequence with each image in the sequence as an individual file. They are numbered in the correct order.

Thanks again for any and all help!
-Alex




















This is the second half of the sample image sequence.

-Alex




















Hi Alex,

Sorry I missed your post before – have you made progress since you last posted? I loaded your pipeline and images, but as you suspected, the segmentation is hard, even by (my untrained) eye. So regardless of the enhancement and segmentation tweaking I could suggest, yes, your TrackObjects will be thrown off by almost certain mis-segmentation.

Can you use a nuclear marker, which would be nice and easily segmentable, or maybe CellTracker might give more signal near the cell membrane which we could use to help segment?

Best,
David