camachodejay / polymer3d Goto Github PK

Need to fuse the old code with the new code from Rafa while allowing the user to choose the method he prefers to use according to how each one performs on one example taken from the data set and analyze with both method.

According to the mail you sent to the Japanese professor the maximum fps we can get is 50. Last time, as I was not aware I tried 100 (10ms exp). It seems to run okay but I did no measure the acquisition time to check).

Now that I try to open it I realize that the camera are not perfectly synchronize (e.g. in the TIF, the 10 first frame or so are on one camera, then they alternate, then they are only on the other camera for the 10 last or so).

This seems to cause issue to the program that load the data. Although it might not be necessary, (it works fine with 20ms exp time), it might be interesting to make the code more robust so it can work with data where camera are not only alternation. (I am not sure yet what the problem actually is).

In transmission data, "bright spot" are dark and background is bright. Thus we need to somehow tell the program (movie.info probably) that it is transmission and if that is the case it needs to reverse the intensity scale so localization and other function built for bright spot are still working fine.

When trying to plot the tracking trace as the movie of the particle movie plays, they did not overlap perfectly.

I already checked the phasor and post processing, it is fine. The reason might be some mistake in the overlay/cropping/translating. And in anycase there will be issue due to the SRCorrection which is shifting toward a plane of reference.

This needs to be investigated more

The 1d fit for the determination of the z position (fitting of the phasor magnitude dependence on z) is relatively slow using least-squares fitting, we should try to improve this step as it is currently the rate-limiting step in the analysis workflow

New method for ROI sub pixel localization, based on: doi: 10.1063/1.5005899

The name used for different fonction tend to use the file extension (e.g. ome ==> getomeinfo) while they are already place in a specific folder : MovieLoad.ome.getomeinfo. The repetition makes things unecessary complicated.

Naming convention is somewhat inconsistent ==> Lower camelCase will be used

MovieLoad will be changed to loadMovie

The code was currently coded assuming that all the channels were fluorescence and needed to be processed. However, we recently found useful to have transmission (or another fluorescence colour) in one camera and fluorescence in the other. The code needs to be changed so it can handle that type of data.

After discussing with Rafa, the best would be to preprocess the data so the rest of the code only know about the part it needs to analyze.

Branch to fix issues related to the segmentation and the pore-size calculation.

We need the functionality to load the calibration file and obtain the channels indices, this way when loading a new movie we use the ROI index information to extract and arrange the channels.

Need to analyze the ellipticity of the pores around the cell

1. segmentation of the cell
2. segmentation of the pores
3. plotting distance dependency from the cell walls

Gradient fit can currently, particularly in the case of noisy data (might also be due to too narrow Gaussian) gives value that does not make sense (e.g. position of 10000 in a 13x13 grid).

Action : The program should be adapted to give NaN when value calculated is outside of the probing radius of the gradient (default = 4) is obtained so the user knows that something went wrong and that he should not trust that value.

New Issue/New branch for all the work concerning the segmentation algorithm and the analysis of the poreSize distribution that is needed for Johannes project. Segmentation and calculation will now be performed using the three-dimensional information.

A. Not convenient to type the name of the file in the main => use getfile?
B. Should support the loading of single or multi camera (only support multi)
C. Should handle both z-stack and time-lapse (currently two different function are needed from which one is not in the repository)
D. getomeinfo should store more information in movie info such as: number of Camera, max number of frame per camera,...

Here we will work on the SP localization and tracking on the multiplane setup.

Just noticed that the saving of the calibration data occur by step of 100 frames but actually prior to that the data is first fully loaded which is not what we want as we were doing 100 frames step to avoid memory overloads.

The zCalibration and test_calibration that use a z-stack and get the z position from the ellipticity currently only work with gradient Fit. Modification will be performed to allow the user to choose which algorithm to use.

Question whether to load or not existing candidate, existing fitting,... are annoying when analyzing multiple file in chain.

In iss11 we turned most of the code into object oriented approach (all classes developped can be found in +Core). This means that a lot of the previous code (e.g. zCalibration) are not needed anymore as they have been re-thought and re-built better. In this branch we will clean all the unecessary code to make it more user-friendly