Home Image-Pro General Discussions

Transferring tracked object outlines between "movies"

Hi all!  New to ImagePro Premier (9.3.3 B6468) and still trying to figure out a lot of the features.  I have a pretty basic project that has some not so basic (I think) problems.  A set of cells change brightness over time, and move a bit (1 channel IF image, easiest case!).  Unfortunately, the background is high, and the cells can sometimes be very difficult to distinguish from the background in the normal unprocessed image.

Now, I have already converted my set of frames to 12bit since the filters do not seem to work on 16bit.  In the next few steps I was able to roughly get what I wanted in terms of filtered adjustments, such that I could get decent segmentation throughout the movie.  Of course, the intensity measurements for each of the segmented objects are now meaningless due to all of my adjustments to make them detectable vs the background (the unsharp mask filter was limited to 7x7, but Edge Plus could be made big enough, though it left me with a mask).

In short, I am stumped as far as how to use the tracks and segmentation I have made in the modified movie to get the accurate intensities from the unmodified movie!  I was hoping it would be as simple as saving and loading them in, but while I can do that for the tracks, I am stuck trying to import the segmentation objects.  End result should be a set of intensities over time per object from the original movie.

Appreciate any and all pointers, I may be going about this in entirely the wrong way!  And if any of the description did not make sense I can flesh it out a little more :)


Best Answers

  • Accepted Answer

    First I want to mention that the filters should work on 16-bit image, so the conversion should not be necessary (you may check other parameters, such as display range, if you don't have expected result).

    Also, you can use large Low-Pass filters using either Large 2D filters or 3D Filters (set Z=1), which don't have size limits.

    You can transfer outlines created by Count/Size to Intensity tracks using Features Manager (add Measurements for Features Manager, activate Intensity Tracking and set these outlines as "Intensity tracks").


  • Accepted Answer
    Yes, when you set the outlines from Features Manager as Intensity Tracks, they don't move, they stay at the same position and give you intensity measurements for all frames. That's the difference between Intensity Tracking and Motion Tracking (where outlines can move).

    There is no easy way to move tracking outlines from all frames on one sequence to another. The solutions can include merging of 2 sequences to one RGB sequence (one sequence as Red and another as Green), then doing segmentation on Red and measuring "Intensity Green". Another solution can involve using Arithmetical operations between masks (Mask all frames) and sequences. (Process-Math-Calc)

    If all cells drift in the same direction, you may try to Align the sequence before measurements (e.g. using "Align by Feature"), so all cells will be static on the aligned sequence and you can use Intensity Tracks for measurements.

    The outlines display on the tracking objects is controlled by the "Show object outlines" option in the tracking options dialog, View group:


  • edited May 2018 Accepted Answer
    Yes, we have this tool. Use "Normalize" function (Process tab, Combine group, Sub-sampling button dropdown). 
    Just draw ROI on an empty background area and click Normalize, all frames in the sequence will be normalized to keep the intensity within the ROI area the same.



  • Thanks for the quick response!

    To clarify on the 16 bit filters not "working," for my 16 bit .czi sequential image sets the filters are grayed out.  Looking at things a bit more closely, it looks like the entire movie is converted into TIFF files when going to 12bit, which is probably what makes the filters function.  My mistake.

    The low pass filters and other filters unfortunately do not improve the manual/bright segmentation significantly, unfortunately, but the Zen software does not have a size limit on it's unsharp mask, so I have been using that for the moment.  It looks like there is a significant background shift over the course of the time series due to the effects of autofluorescence in the media which I am trying to account for.  With the cells moving around, it may be tricky to try to deal with the fluctuations.

    I will try using the Features Manager as soon as I get a chance!

  • Sorry, I am not sure I quite followed that once I started trying to do it.

    I can generate ROIs for my objects in Count/Size or Measure/Motion Tracking (Motion Tracking works beautifully for this on the processed movie), but when I use Count/Size, only the first frame of my data is used to generate the ROI, and so the Intensity track will not follow the cells as they move.  On the other hand, I can take these static ROIs and transfer them to the original movie using the Features Manager in the Select tab.

    When I generate the ROIs through the Motion Tracking, they correctly follow the cells (and track intensity), but I am still somewhat stumped as far as getting the Motion Tracking to transfer to another movie.

    Side question:  Sometimes when I do Motion Tracking I can see the ROIs as I play through my movie (it is interesting to watch the outline change shape as the cell changes), but other times they do not I am not sure how to get those ROIs to show up along with the tracks.  Is there a way to toggle on and off the ROI's each track is using for it's measurements?  Opening the program on a new computer restores them, but I do not know how I turned them off!
  • Thanks, I think the merging of the two files into Red/Green channels would be great!  And that was exactly the checkbox I needed.

    Ah, I had a lot typed here, but it looks like I just needed to convert the files to .tiffs before I could get access to the Composite function.  That looks like a great first step, and I got an image set with R=255 and G=255 for the two channels (.mcs file).  Once I click on "Objects" within Measure for tracking though, it seems to swap everything back to mono.  I am unsure how to just threshold on red or green with the limited options available in Manual.

    This does help though, since at least the combined intensities are easier to segment than the original, but I suspect I am missing something easy to get this to work the way you have described.  Also, once it converts to mono for Object tracking, I can no longer get the Intensity:Green or Intensity:Red from the Data Types to register (their columns are empty).

    In case anyone else does this, the file first needs to be formed into a composite using both movies, and then Merged as well!  Everything is working perfectly now.

    Another extra question: while this seems like a decent way to handle images with three channels (pseudo RGB), is there a way to process images with 4+ fluorescent channels?  Not that I need it now, just thinking about future projects :)

    Thanks for all the help!

  • Oops, hopefully one last quick question.  As I mentioned, the background varies quite a bit over the course of the experiment due to autofluorescence recovery (back to the original mono movie again).  I can adjust for this after the fact by creating an Intensity Track object as you mentioned earlier in an area where no cells show up. 

    Is there a way to use that or something like it to smooth out the entire movie?  I attempted to use the Subtract Dark Background and "From points" with the points selected in a safe area, but neither seemed to affect the entire movie very much.  The calculator can subtract a single value or an image, but I am not sure how to reference the mean intensity from an ROI as it changes from frame to frame.
  • I am not sure what you mean by "smooth". You can try filters, try 3D filters, e.g. Low-Pass or Gaussian.

    When you use Calc between sequence and an image, the operation is executed between every frame of the sequence and image. Try using New button (instead of Apply), so you can compare sequences side by side, before and after.


  • Sorry, smooth as in the background intensity starts low, peaks high within a few frames, then gradually lessens over the course of the time series.  If I select the entire image area, I can see the same exact mean intensity curve as if I select a small area of background, which given the low proportion of background in the entire image tells me that the autofluorescence variation is affecting both the cells and the media.  I would like to subtract out a fixed value in each frame (Calculator), but what that fixed value is will change significantly for each frame of the 91 frame movie. 

    I am looking for an automated way to generate a value that represents the amount of background (say the mean intensity of an ROI where no cells pass through at any time point) in each frame and subtract that value from the entire frame.
  • That is perfect!  Thank you so much for all of your help!

Sign In or Register to comment.