dimgstats.cpp File Reference

Calculates statistical measures within a set of images or between sets of images. More...

#include "Bstring.h"
#include "rwimg.h"
#include "img_combine.h"
#include "math_util.h"
#include "utilities.h"
#include "options.h"
#include "timer.h"
#include "Vector3.h"

Functions
int	img_fom_sqrt (Bimage *p)
Bimage *	img_students_t_test (int n, Bimage *p, vector< double > &weight)
	Applies the Student's t-test to one set of images. More...
Bimage *	img_students_t_test_equal (int set1, int set2, Bimage *p1, Bimage *p2, vector< double > &weight)
	Applies the Student's t-test to two sets of images, assumes the two distributions have approximately the same variance. More...
Bimage *	img_students_t_test_unequal (int set1, int set2, Bimage *p1, Bimage *p2, vector< double > &weight)
	Applies the Student's t-test to two sets of images, assumes the two distributions have unequal variance. More...
Bimage *	img_f_test (int set1, int set2, Bimage *p1, Bimage *p2, vector< double > &weight)
	Applies the F-test to two sets of images. More...
int	main (int argc, char **argv)

Variables
int	verbose
const char *	use []

Detailed Description

Calculates statistical measures within a set of images or between sets of images.

Author

David Belnap and Bernard Heymann

Date

Created: 20051213

Modified: 20190208 (BH)

Function Documentation

img_f_test()

Bimage * img_f_test	(	int	set1,
		int	set2,
		Bimage *	p1,
		Bimage *	p2,
		vector< double > &	weight
	)

Applies the F-test to two sets of images.

Author

David Belnap

Parameters

set1	number of files in set 1.
set2	number of files in set 2.
*p1	average image with variance (FOM), set 1
*p2	average image with variance (FOM), set 2
weight	list of weights.

Returns

Bimage* F-test image and significance image (as FOM).

The F value is computed for each pixel:
        F(i) = var1(i) / var2(i)

Variance (as FOM of an average image) is input.  The significance 
level is computed and stored as FOM of F-test image.  Output is
in floating point.

Reference: Press W.H. et al (1992) Numerical Recipes in C.

img_fom_sqrt()

int img_fom_sqrt

(

Bimage *

p

)

img_students_t_test()

Bimage * img_students_t_test	(	int	n,
		Bimage *	p,
		vector< double > &	weight
	)

Applies the Student's t-test to one set of images.

Parameters

n	number of files in the set.
*p	average and variance of the set.
weight	list of weights.

Returns

Bimage* significant level image (floating point).

First, average (avg) and variance (var) images are calculated:
        var = [1/(N-1)] sum(x - avgx)^2
Second, the t value is computed and returned as an image:
        t = avg / sqrt[ var/N ]
Finally, the significance is calculated and returned as the FOM of the image:
        sig = betai(dof/2, 0.5, dof/(dof + t*t))
where dof (degrees-of-freedom) is calculated as:
        dof = ws*(1 - 1.0L/N)
and ws is the weight sum.
All images are converted to floating point.

Reference: Press W.H. et al (1992) Numerical Recipes in C. Milligan and Flicker (1987) J. Cell Biol. 105:29-39.

img_students_t_test_equal()

Bimage * img_students_t_test_equal	(	int	set1,
		int	set2,
		Bimage *	p1,
		Bimage *	p2,
		vector< double > &	weight
	)

Applies the Student's t-test to two sets of images, assumes the two distributions have approximately the same variance.

Author

David Belnap and Bernard Heymann

Parameters

set1	number of files in set 1.
set2	number of files in set 2.
*p1	average image with variance (FOM), set 1
*p2	average image with variance (FOM), set 2
weight	list of weights.

Returns

Bimage* t-test image and significance image (as FOM).

Input average images for sets 1 and 2, with variance as FOM portion
of the images.  The numerator portion of the variance (Vn) is 
computed from the variance.  The "standard error of the difference 
of the means" is computed:
SD = sqrt[ ( (Vn1 + Vn2) / degrees_of_freedom ) * (1/set1 + 1/set2) ]

and used to compute the t value:
            t = ( avg1 - avg2 ) / SD

Finally, the significance level (probability) is computed.

Reference: Press W.H. et al (1992) Numerical Recipes in C. Milligan and Flicker (1987) J. Cell Biol. 105:29-39.

img_students_t_test_unequal()

Bimage * img_students_t_test_unequal	(	int	set1,
		int	set2,
		Bimage *	p1,
		Bimage *	p2,
		vector< double > &	weight
	)

Applies the Student's t-test to two sets of images, assumes the two distributions have unequal variance.

Author

David Belnap

Parameters

set1	number of files in set 1.
set2	number of files in set 2.
*p1	average image with variance (FOM), set 1
*p2	average image with variance (FOM), set 2
weight	list of weights.

Returns

Bimage* t-test image and significance image (as FOM).

Input average images for sets 1 and 2, with variance as FOM portion
of the images.  The t value is computed:
        t = ( avg1 - avg2 ) / sqrt[ var1/ws1 + var2/ws2 ]

(ws = weighted sum, default = N).  The significance level 
(probability) is computed.  Output is in floating point.

Reference: Press W.H. et al (1992) Numerical Recipes in C. Milligan and Flicker (1987) J. Cell Biol. 105:29-39.

main()

int main	(	int	argc,
		char **	argv
	)

Variable Documentation

use

const char* use[]

verbose

int verbose

extern