icstutorial.html

cell migration resource imaging icsmatlab

    0.0107
</pre><p>The beam radius, <tt>w</tt>, should be close to the set value, 0.4. The diffusion coefficient <tt>D</tt>, should be close to the set value 0.01.
      </p>
      <h2>SPATIO-TEMPORAL ICS (STICS)<a name="22"></a></h2>
      <p>STICS measures the velocity of flowing particles.  Let's generates an image series with 5 particles/um^2 immobile, 5 particles/um^2
         flowing with direction "down-left" i.e., x component 0.0707 um/s, y component -0.0707 um/s.  The pixel size will be 0.06 um,
         and the time resolution 0.1 seconds.
      </p>
      <p>This gives a total speed of sqrt(2(0.0707^2)) = 0.1 um/s.</p><pre class="codeinput">imageSeriesFlow = simul8tr(256,256,100,[5 5],<span class="string">'none'</span>,[0 0],[1 1],0.06,0.1,<span class="string">'g'</span>,0.4,0,12,[0 0],[0.0707 0],[-0.0707 0],[0 0],0,0);
</pre><p>The function <tt>velocity</tt> performs the STICS analysis:
      </p><pre class="codeinput">[Vx,Vy] = velocity(imageSeriesFlow,0.1,0.06,<span class="string">'y'</span>,20,<span class="string">'n'</span>)
</pre><pre class="codeoutput">Current plot held
</pre><img vspace="5" hspace="5" src="ICSTutorial_06.png"> <p>where the analysis was performed up to lag 20, <tt>'y'</tt> denotes immobile filtering, with time resolution 0.1s and pixel size 0.06 um. <tt>'n'</tt> denotes no whitenoise (there would be with real data).  When prompted, select end of the "linear" region of the plot.  Because
         this simulation has high S/N and temporal coherence, the plot should be linear over the range displayed.  Just click somewhere
         on the right hand side (i.e., high tau) of the plot.  <tt>Vx</tt> and <tt>Vy</tt> will be calculated, and should agree with the set velocity components of 0.071 and -0.071.  The magnitude of the velocity
         measured by STICS is given by <tt>V</tt>:
      </p><pre class="codeinput">V = sqrt(Vx^2+Vy^2)
</pre><pre class="codeoutput">V =
    0.1052
</pre><p><tt>V</tt> should be close to the set speed, 0.1um/s Let's compare this result with TICS.  As with diffusion, we calculate the TACF,
         given 0.1 second temporal sampling:
      </p><pre class="codeinput">GtFlow = tics(imageSeriesFlow,0.1);
</pre><p>For samples undergoing 2D flow, the TACF has the following analytical form:</p>
      <p><img vspace="5" hspace="5" src="ICSTutorial_eq530260.png"> </p>
      <p>We can fit the first 60 lags of the simulation TACF to this model:</p><pre class="codeinput">flowCoeff = flowfit(GtFlow(1:60,1),GtFlow(1:60,2));
</pre><img vspace="5" hspace="5" src="ICSTutorial_08.png"> <p>The flow speed can be calculated from this fit:</p>
      <p><img vspace="5" hspace="5" src="ICSTutorial_eq89431.png"> </p>
      <p>Or, in MATLAB:</p><pre class="codeinput">v = w/flowCoeff(2)
</pre><pre class="codeoutput">v =
    0.0977
</pre><p><tt>v</tt> should be close to <tt>V</tt>, the speed recovered from the STICS analysis. Finally, we can calculate the % immobile:
      </p><pre class="codeinput">percentImmobile = flowCoeff(3)/(flowCoeff(1)+flowCoeff(3))*100
</pre><pre class="codeoutput">percentImmobile =
   46.5158
</pre><p><tt>percentImmobile</tt> should be close to 50, the value set in the simulation.
      </p>
      <h2>IMAGE SERIES IMPORTING AND MANIPULATION<a name="34"></a></h2>
      <p><b>Importing Image Series</b></p>
      <p>A 16-bit Fluoview TIFF image series can be imported using <tt>rd_img16</tt>:
      </p>
      <p><tt>image_data=rd_img16(filename)</tt>;
      </p>
      <p>where <tt>filename</tt> a string which contains the filename to load.  (e.g., <tt>'C:\MyData\ImageSeries.tif'</tt>.)  Similary, <tt>rd_imgser</tt> can load 8-bit RAW files:
      </p>
      <p><tt>image_data=rd_imgser(filename,sizex,sizey,numimg)</tt>;
      </p>
      <p>where <tt>sizex</tt> and <tt>sizey</tt> are the dimensions of an image, and <tt>numimg</tt> is the number of images in the series.
      </p>
      <p>If the file is a Fluoview TIFF and the image series was collected on an Olympus microscope, <tt>FluoInfo</tt> can extract data collection parameters from the file header:
      </p>
      <p><tt>[XDim,YDim,TDim, PMTVoltage, PMTOffset, PMTGain, LaserPower] = FluoInfo(filename);</tt></p>
      <p>where <tt>XDim</tt> and <tt>YDim</tt> are the pixel sizes, in um, in the X and Y directions, and <tt>TDim</tt> is the time between images, in seconds.  <tt>PMTVoltage</tt>, <tt>PMTOffset</tt>, and <tt>PMTGain</tt>, give the PMT settings, and <tt>LaserPower</tt> is the laser intensity, in percent.
      </p>
      <p><b>Correcting for Background Noise</b></p>
      <p>Background intensity counts can be subtracted from an image series using <tt>wnCorr</tt>:
      </p>
      <p><tt>imageSeriesCorrected = wnCorr(imageSeries);</tt></p>
      <p>You will be prompted to select a background region.  Its mean will be subtracted from the image series.</p>
      <p><b>Cropping a Region for Analysis</b></p>
      <p>An image series can be cropped to select a region of interest using <tt>serimcrop</tt>:
      </p>
      <p><tt>croppedImageSeries = serimcrop(imageSeries);</tt></p>
      <p>You will be prompted to interactively select the region of interest.</p>
      <h2>REFERENCES<a name="35"></a></h2>
      <p><b>Image correlation Spectroscopy (ICS)</b></p>
      <p>[1] Petersen, N., P. Hoddelius, P. Wiseman, O. Seger, and K. Magnusson. 1993. Quantitation of membrane receptor distributions
         by image correlation spectroscopy: concept and application. Biophys. J. 65:1135-46. <a href="http://www.biophysj.org/cgi/content/abstract/65/3/1135">http://www.biophysj.org/cgi/content/abstract/65/3/1135</a></p>
      <p>[2] Wiseman, P., and N. Petersen. 1999. Image correlation spectroscopy. II. optimization for ultrasensitive detection of preexisting
         platelet-derived growth factor-beta receptor oligomers on intact cells. Biophys. J. 76:963-77. <a href="http://www.biophysj.org/cgi/content/abstract/76/2/963">http://www.biophysj.org/cgi/content/abstract/76/2/963</a></p>
      <p>[3] Costantino, S., J. W. Comeau, D. L. Kolin, and P. W. Wiseman. 2005. Accuracy and dynamic range of spatial image correlation
         and cross-correlation spectroscopy. Biophys. J. 89:1251-1260. <a href="http://www.biophysj.org/cgi/content/abstract/89/2/1251">http://www.biophysj.org/cgi/content/abstract/89/2/1251</a></p>
      <p><b>Temporal Image Correlation Spectroscopy (TICS)</b></p>
      <p>[4] Wiseman, P., J. Squier, M. Ellisman, and K. Wilson. 2000. Two-photon image correlation spectroscopy and image cross-correlation
         spectroscopy. J. Microsc. 200:14-25. <a href="http://www.blackwell-synergy.com/doi/abs/10.1046/j.1365-2818.2000.00736.x">http://www.blackwell-synergy.com/doi/abs/10.1046/j.1365-2818.2000.00736.x</a> and correction: <a href="http://www.blackwell-synergy.com/doi/full/10.1046/j.0022-2720.2001.001013.x">http://www.blackwell-synergy.com/doi/full/10.1046/j.0022-2720.2001.001013.x</a></p>
      <p>[5] Kolin, D. L., S. Costantino, and P. W. Wiseman. 2006. Sampling Effects, Noise, and Photobleaching in Temporal Image Correlation
         Spectroscopy. Biophys. J. 90:628-639. <a href="http://www.biophysj.org/cgi/content/abstract/90/2/628">http://www.biophysj.org/cgi/content/abstract/90/2/628</a></p>
      <p><b>Spatio-temporal Image Correlation Spectroscopy (STICS)</b></p>
      <p>[6] Hebert, B., S. Costantino, and P. W. Wiseman. 2005. Spatiotemporal image correlation spectroscopy (STICS) theory, verification,
         and application to protein velocity mapping in living CHO cells. Biophys. J. 88:3601-3614. <a href="http://www.biophysj.org/cgi/content/abstract/88/5/3601">http://www.biophysj.org/cgi/content/abstract/88/5/3601</a></p>
      <p><b>MATLAB Documentation</b></p>
      <p>Getting started in MATLAB: <a href="http://www.mathworks.com/access/helpdesk/help/techdoc/learn_matlab/learn_matlab.html">http://www.mathworks.com/access/helpdesk/help/techdoc/learn_matlab/learn_matlab.html</a></p>
      <p class="footer"><br>
         Published with MATLAB&reg; 7.0.1<br></p>
      <!--
##### SOURCE BEGIN #####
%% MATLAB ICS Tutorial
% February 27, 2006
%
% By David Kolin

%% INTRODUCTION
% These notes are meant to serve as a very brief introduction to the accompanying MATLAB
% image correlation spectroscopy code.  They are not intended to be a
% comprehensive lesson on the underlying theory of ICS, or an introduction to MATLAB.  For those details, see
% the references given at the end of this webpage.
%
% The MATLAB code in this tutorial is in a |monospaced font|.  It can
% simply be copied from this page to the command window.  All of the
% ICS .m files in the unzipped directory should be added to the MATLAB path: File REPLACE_WITH_DASH_DASH> Set
% Path... REPLACE_WITH_DASH_DASH> Add with Subfolders...  You'll need the Optimization and
% Image Processing Toolboxes to use all of the included .m files.
%
% Most of the included ICS .m files have a (small!) help section, which can be called by |'help
% filename'|.  All comments and suggestions are very much appreciated, and can be sent to
% [e-mail].

%% SIMULATOR
% A simulation of fluorescently labeled particles imaged on a confocal microscope
% can be generated using simul8tr.m, which has the following syntax:
%
% |simulation = simul8tr(sizeX,sizeY,sizeT,density,bleachType,bleachDecay,qYield,pixelsize,
%                        timesize,PSFType,PSFSize,PSFZ,noBits,diffCoeff,flowX,flowY,flowZ,
%                        countingNoise,backgroundNoise);|
%
% where:
%
% |sizeX| and |sizeY| are the dimensions of the simulation, in pixels.
%
% |density| is the particle density in particles per um^2.
%
% |bleachType| determines if the fluorophores bleach, and is either
% |'none'| for no bleaching or |'mono'| for a monoexponential decay in
% average intensity.
%
% |bleachDecay| determines how quickly particles bleach. It is the rate constant
% for the monoexponential bleaching, in the same units at the time step
% (usually between 0 and 0.05, given 1 Hz imaging rate).
%
% |qYield| is the quantum yield of the fluorophores (usually 1).
%
% |pixelsize| is the size of a pixel, in um (usually ~0.1).
%
% |timesize| is the time between frames, in seconds (usually ~1).
%
% |PSFType| gives the shape of the point spread function of the imaging system.
% It is either |'g'| for a 2D Gaussian, or |'a'| for an airy disk.
%
% |PSFSize| is the e^-2 radius, in um, for a Gaussian PSF, or the distance
% to the first zero of the airy disk, in um.
%
% |PSFZ| is the size, in um, for the Z dimension of the PSF.  For
% 2D simulations, set this to 0.  The PSF in Z is always a Gaussian,
% regardless of |PSFType|.
%
% |noBits| is the number of bits used in the image normalization, imitating
% a A/D converter (usually 12).
%
% |diffCoeff| is the diffusion coefficient, in um^2/s.
%
% |flowX|, |flowY|, and |flowZ| are the flow speeds in each of the
% directions, in um/s.
%
% |countingNoise| is the noise associated with the PMT amplification
% electronics (see the August 2005 Costantino BJ paper for details).  Usually between 1 and 20.
%
% |backgroundNoise| is the noise associated with spurious background counts
% (see the same paper).  Usually between 0 and 0.3.
%
% As an example, let's create a simulated image series with the following characteristics:
% 256 x 256 pixels with 100 images, 10 particles per um^2, 1 s per image,
% 0.1 um/pixel, particles with a quantum yield of 1, a Gaussian convolving function
% with an e^-2 radius of 0.4 um, with particles diffusing at 0.01 um^2/s, and no noise

imageSeriesDiff = simul8tr(256,256,100,10,'none',0,1,0.1,1,'g',0.4,0,12,0.01,0,0,0,0,0);
%%
% The first image in the series should look something like this:

imagesc(imageSeriesDiff(:,:,1))
axis image
colormap(gray)

%% SPATIAL ICS (ICS)
% We can determine number densities and aggregation states using ICS.  First, we calculate a
% 2D spatial autocorrelation function (SACF) for each image in the simulation which we created earlier:

ICS2DCorr = corrfunc(imageSeriesDiff);
%%
% The SACF of an image should be a 2D Gaussian.  We can view the SACF for first image:

s=surf(ICS2DCorr(:,:,1));
axis tight
colormap(jet)
xlabel('\eta','FontSize',12)
ylabel('\xi','FontSize',12)
zlabel('r(\xi,\eta)','FontSize',12)
set(s,'LineStyle','none')
title('Spatial Autocorrelation Function for First Image')
%%
% Next, we crop the stack of SACFs around the central peak, since the
% fitting algorithms work better when the noise at higher spatial lags is
% removed.

ICS2DCorrCrop = autocrop(ICS2DCorr,12);
%%
% As an example, let's look at the SACF of the first image to see how much was cropped:

s=surf(ICS2DCorrCrop(:,:,1));
axis tight
colormap(jet)
xlabel('\eta','FontSize',12)
ylabel('\xi','FontSize',12)
zlabel('r(\xi,\eta)','FontSize',12)