Tuesday, February 28, 2006

Heiko, Virginie, Darren

(1) Heiko paper's movie making. Restrictions on movies are very tight. I down-sized 250Mb movies to 1.5Mb, but it still seems to be TOO BIG! What a journal. (I don't name it)

(2) Virginie & Darren project.
Very interesting, vector field works very well.

Wednesday, February 22, 2006

Plans on CMCI seminars and other things...

CMCI plans to organize intra-EMBL seminar series centered on Imaging issues. Major goal is to share information on imaging techniques and to get people be known to each other. I talked with many people who might be interested in this seminar and it seems that many people are really interested in this seminar. Since number of people is large, I started to think it might be better to do a kick-off of this seminar with a symposium, with al interested people taking in two days about their works. I talked with Jan today about this and he agrees with doing such a symposium. One of the merit is that we can then also involve the attendance of outstation scientists who might not be able to come to Heidelberg weekly. I will start setting schedule for this kick-off symposium.

- other things today-

ALMF Microscopy course is going on today, many people strolling around inside ALMF. Stefan asked me in which order one should do image subtraction and background subtraction: I wasn't sure about this. All I could show, was that in case of 32-bit results, noise actually increases its dynamic range by image subtraction so I would do backgoround subtraction first.

1. with Heiko: Cholesterol paper revising

(1) FRAP Curve fitting evaluation part, gammaQ things.
(2) Updating Frap Graphs .

2. with Darren
Darren liked the vector field analysis I did for Virginie. I gave him a pdf document explaining the optical flow estimation. We deided to collaborate.

3. Discussion with Francois
... about currently writing paper on substrate patterning.

vector field paper revising

I am (still) working on a paper about analysis of intracellualr vesicle cargo protein dynamics, or the transport of protein between different organelles, and its in the almost final form but for a year I havn't touched it (lazy guy!) . The method used is the gradient-based optical flow estimation. The paper describes about details of vesicle transport dynamics. I think that the protein sorting within cell can be described as stochastic processes, and this paper will be a primer for this purpose (I hope...).

Keyward links:
Mathworld: stochastic processes
Wikipedia: stochastic processes
Probability theory and stochastic processes

Tuesday, February 21, 2006

Virginie: Fish things plan

Virginie: Vec field analysis for the MT orientations. (local differences)
+ New sequences in PPK71
--> tried with the sequence 2 and clear differenc in the histogram.

Categorrizing Signals: Nucleus, Spot signals

Task: seperate three types of localization automatically:
--> same type of task appears in many applications.

A: small spots a bit remote from the nucleus
B: Large aggregate of replication sites, surrounding the nucleus.
C: diffuse cytoplasmic signal

In general, thresholding strategy will be the key for this type of problem .


1. Threshold DAPI at a sufficient level. (segment each cell according to nucleus)
--> do nucleus signals are always with similar intensity?
-- if not, threshold level must be set for each nucleus.
-- according to the size of nucleus?
-- dynamic thresholding --> determine the best threshold level.

2.a. Dialate Thresholded DAPI signal, to make a mask for each cell.
--> define iteration of dialate operation.
2.b. Get the centroid of the Dapi signal.
--> particle analysis

3. within the mask (=ROI), threshold to select high intensity signal.
4. count particles within the mask, get average area and intensity.

5. plotting parameters
- distance from the nucleus
- average area
- average intensity
- number of counts in graphs. (3D graph?)

--> can populations be seperated? (cluster analysis)

Monday, February 20, 2006

Heiko Collaboration, Gaspar, Fatima

1. Heiko

Making figures for a paper, which will be submitted in a short time (deadline is this weekend!).

a. Simualtion results.
b. Tracks of vesicle movements overlaid to the original frames.

for the second task, I modified Igor procedure


that the tracks will be color coded according to their timing.

2. Gaspar

Phototaxis things. Tracks to do circular statistics, probably will be done by 2D cross-correlation.

3. Fatima

While I was telling her how to analyze her signal, I realize that a manual for Gaussian-fit tracker must be written.

Friday, February 17, 2006

Fixing a macro.

I fixed an ImageJ macro "4D_PE_Processor" . I wrote this sometime ago, probably in 2004 summer to process 4D sequences of Medaka fish embryo. Since those files were really huge (Gbytes!) data set, batch procesing script was something necessary. It batch processes projections, resizing and contrast enhance ment for the tiff-file series generated by perkin-Elmer UltraView software. There was a e-mail from a friend in Weizmann Institute in Israel that the macro does not read his image sequence. The problem was rather simple: file naming. Since the frame number and z-slice number information could be known only in the file name, if the file name has something wrong, then the macro does not recognize the file properly.

A problem about Perkin-Elmer Tiff files are that it has two types of file naming. In old software, file naming is

prefix_****_***.tif (* are digits)

and the new one is like

prefix_T****_(channel wavelength)_Z***.tif.

the first four digits indicate time points and the latter three digits indicate Z slice position. I made the macro so that it reads time point by the order of underscore ("_"). Then if one uses underscore in the prefix, the detection of timepoints and z-slices will be totally messed up. As a work round, I changed the macro to recognize "_0" rather than "_" alone. It needs more improvement since the macro does not work if timepoint is more than 999 or if z-slices are greater than 99.