calc_correlation.m File Reference

Calculate correlation coefficients for img, script to run from Correlation panel of im_browser. More...

Functions
function [ corr_coef , model , objMask , smooth_corr , granularity , roi_size ] =	calc_correlation (img, img_mask, mode, type, slope_length, darkthres, dt, auto_thres, inv_sw, a_contrast, relative_to_frame, objMask)
	Calculate correlation coefficients for img, script to run from Correlation panel of im_browser. More...

Detailed Description

Calculate correlation coefficients for img, script to run from Correlation panel of im_browser.

Function Documentation

function [ corr_coef , model , objMask , smooth_corr , granularity , roi_size ] = calc_correlation	(		img,
			img_mask,
			mode,
			type,
			slope_length,
			darkthres,
			dt,
			auto_thres,
			inv_sw,
			a_contrast,
			relative_to_frame,
			objMask
	)

Calculate correlation coefficients for img, script to run from Correlation panel of im_browser.

Parameters

img	- stack of images (h,w,color_channel,frame)
img_mask	- take only part of the original image into analysis. Mask(h,w) defines the area for analysis
mode	- mode, Absolute, Relative (current-previous), Relative cummulative
type	- type of correlation: Pearson, Overlap, Manders, Intersection...
slope_length	- number of frames to be used for smoothing calculation
darkthres	- object separation threshold, use 0 to analyse the whole image
dt	- time between frames
auto_thres	- [optional] automatic thresholding by Otsu's method switch: 0 - off, 1 - on
inv_sw	- [optional] instead of correlation, return displacement, usually 1-minus-correlation
a_contrast	- [optional] if not NaN, is used for adjust contrast individually for each roi box in each frame
relative_to_frame	- [optional] frame number for absolute mode
objMask	- [optional] provide object mask for analysis

Return values

corr_coef	- vector with amplitudes of correlation coefficients
model	- difference map, current_frame -minus- previous
objMask	- object mask, generated with Otsu's method or with specified threshold value (darkthres)
smooth_corr	- smoothed correlation coefficient with the window size=slope_length
granularity	- granularity of the objMask
roi_size	- size of the roi object in pixels

Types of coefficients:

1. Pearson: Pearson colocalization coefficient, intensity based as

   figure;
   ax=gca;
   set(ax,'Visible','off');
    text(0.2,0.5,'$r_p = \frac{\sum((R_i-\bar{R})(G_i-\bar{G}))}{\sqrt{\sum{(R_i-\bar{R})^2\sum{(G_i-\bar{G})^2}}}}$','interpreter','latex','fontsize',14);

2. Overlap, overlap, intensity based coefficient as

   figure;ax=gca;
   set(ax,'Visible','off');text(0.2,0.5,'$r_o = \frac{\sum{R_i \times
   G_i}}{\sqrt{\sum{R_i^2}\times\sum{G_i^2}}}$','interpreter','latex','fontsize',14);

3. Manders, intensity based coefficient: M1 (current vs relative) and M2 (relative vs current) as

   figure;ax=gca;set(ax,'Visible','off');text(0.2,0.5,'$M_1 = \frac{\sum{G_{i,coloc}}}{\sum{G_i}}$','interpreter','latex','fontsize',14)
   figure;ax=gca;set(ax,'Visible','off');text(0.2,0.5,'$M_2 = \frac{\sum{R_{i,coloc}}}{\sum{R_i}}$','interpreter','latex','fontsize',14)

4. Intersection coefficient, object based as

   figure;ax=gca;set(ax,'Visible','off');
    text(0.2,0.5,'$Rth_i =\Big\{ \frac{\mbox{0, if} R_i \le R_{bg}}{\mbox{1, if} R_i > R_{bg}} $','interpreter','latex','fontsize',14)
    text(0.2,0.3,'$I =\frac{\sum{Rth_i \cap Gth_i}}{\sum{Rth_i} + \sum{Gth_i} - \sum{Rth_i \cap Gth_i}} $','interpreter','latex','fontsize',14);

   ;
5. Mask displacement, object based coefficient. Calculates displacement of the thresholded image as

   figure;ax=gca;set(ax,'Visible','off');
   text(0.02,0.5,'$Rth_i =\Big\{ \frac{\mbox{0, if} R_i \le R_{bg}}{\mbox{1, if} R_i > R_{bg}} $','interpreter','latex','fontsize',12)
   text(0.02,0.3,'$D_m =\frac{\sum{Rth_i \ominus Gth_i}}{\mbox{total number of pixels}}, \mbox{where }\ominus \mbox{is symmetric difference} $','interpreter','latex','fontsize',12);

6. Image displacement, displacements of masks that are generated from the difference of compared images as
   • take difference between current and relative images
   • threshold the difference image with the provided parameter to generate two masks of displacement
   • take a union of both masks
   • calculate the coefficient as sum of mask's union / by total number of pixels in the area

     See Also

     ib_calcCorrelation, correlationAnalysis_script

Referenced by ib_calcCorrelation().

Here is the caller graph for this function: