CScaledCorrelationBC Class Reference

A class that computes the scaled correlation for a vector of time stamps and a continuous signal (Binary - Continuous). More...

#include <ScaledCorrelation-BC.h>

List of all members.

Public Member Functions
	CScaledCorrelationBC (int piScaleWindow, int piCorrelationWindow, int piTrialLength, int &piIsError)
	Constructor.
	~CScaledCorrelationBC ()
	Destructor.
void	ComputeScaledCorrelation (int *piaTimeStampsA, int piNrTimeStampsInA, float *pfaSamplesB, int piNrSamplesInB, int pbUseFisherZTransform)
	Compute the scaled correlation of one binary signal with one continuos signal.
float *	GetScaledCrossCorrelogram ()
	Returns the buffer with the computed scaled correlogram; returns NULL if there was an error; please check the buffer for NULL before using it!!
float *	GetPBsCoefficientSums ()
	Returns the sum of valid Point Biserial coefficients of correlation for each bin of the correlogram.
int *	GetPBsCoefficientCounts ()
	Returns how many Point Biserial coefficients of correlation have been averaged for each bin of the scaled correlogram.
int *	GetDistributionOfCorrelationCoefficients (int &piNumberOfBins, float &pfBinSize)
	Returns the distribution of coefficients of correlation.
int	ModifyScaleWindow (int piNewScale)
	Changes the scale segment size for the scaled correlation.
int	ModifyCorrelationWindow (int piNewCorrelationWindow)
	Set the size of the correlation window; the parameter specifies the correlation window; for example for a window of -100..+100 pass 100 as a parameter.
int	ModifyTrialLength (int piNewTrialLength)
	Set the length of the trial.
int	ModifyAllParameters (int piNewScale, int piNewCorrelationWindow, int piNewTrialLength)
	Sets all parameters at once.
int	GetScaleWindow ()
	Get the size of the current scale segment.
int	GetCorrelationWindow ()
	Get the size of the correlation window.
int	GetTrialLength ()
	Get the trial length in original sampling units.
int	GetDistributionOfCorrelationCoefficientsBinNr ()
	Get the number of bins of the distribution of coefficients of correlation.
float	GetDistributionOfCorrelationCoefficientsBinSize ()
	Get the size of a bin of the distribution of coefficients of correlation.

---

Detailed Description

A class that computes the scaled correlation for a vector of time stamps and a continuous signal (Binary - Continuous).

For details on the Scaled Correlation algorithm see paper on "Scaled Correlation Analysis" by Danko Nikolic, Raul C. Muresan, Weijia Feng, and Wolf Singer.
For pairs of spikes and LFP signals in neuroscience, this measure is also known as "spike-triggered LFP average".
Timestamps corresponding to events of one binary signal and one continuous signal must be provided. The time stamps of the binary signal must match the samples of the continuous signal, i.e. the first sample of the continuous signal corresponds to time stamp 0.

---

Constructor & Destructor Documentation

CScaledCorrelationBC::CScaledCorrelationBC	(	int	piScaleWindow,
		int	piCorrelationWindow,
		int	piTrialLength,
		int &	piIsError
	)

Constructor.

Parameters:

piScaleWindow	- The scale on which you compute the scaled correlation; use the same units as the sampling unit of your time stamps
piCorrelationWindow	- The size of the cross-correlation window (eg. 80 for a cross-correlation with lags of -80..+80); use again the same units as the sampling of your time stamps
piTrialLength	- The size of the trial in original sampling units
piIsError	- A return parameter, set to 0 on success. It is set to -1 by the constructor if there is a mismatch of window sizes and the GetScaledCorrelogram() function will return NULL;

Warning:

The sampling frequency of the continuous and binary signals has to be the same! Also the trial lengths have to match!

CScaledCorrelationBC::~CScaledCorrelationBC

(

)

Destructor.

---

Member Function Documentation

void CScaledCorrelationBC::ComputeScaledCorrelation	(	int *	piaTimeStampsA,
		int	piNrTimeStampsInA,
		float *	pfaSamplesB,
		int	piNrSamplesInB,
		int	pbUseFisherZTransform
	)

Compute the scaled correlation of one binary signal with one continuos signal.

Parameters:

piaTimeStampsA	- The vector holding the time stamps of the binary variable
piNrTimeStampsInA	- Number of time stamps in the first vector
pfaSamplesB	- The vector holding samples of the continuous signal
piNrSamplesInB	- Number of samples in the second vector
pbUseFisherZTransform	- Set to 1 to use the Fisher Z transform to average the coefficients of correlation; set to 0 for normal computation;

Warning:

The sampling frequency of the continuous and binary signals has to be the same! Also the trial lengths have to match!
The time stamps of the binary signal must match the samples of the continuous signal, i.e. the first sample of the continuous signal corresponds to time stamp 0.

int CScaledCorrelationBC::GetCorrelationWindow

(

)

Get the size of the correlation window.

Returns:

Returns the size of the correlation window

int * CScaledCorrelationBC::GetDistributionOfCorrelationCoefficients	(	int &	piNumberOfBins,
		float &	pfBinSize
	)

Returns the distribution of coefficients of correlation.

Returns:

A pointer to the internal buffer that holds the distribution of correlation coefficients

The number of bins in the distribution (piNumberOfBins); this is equal to the size of the buffer

The size of one bin of the distribution (pfBinSize); the size of the bin is in sampling units of the original signals

int CScaledCorrelationBC::GetDistributionOfCorrelationCoefficientsBinNr

(

)

Get the number of bins of the distribution of coefficients of correlation.

Returns:

Returns the number of bins used to compute the distribution of correlation coefficients; the value equals the size of the internal buffer that stores the distribution of correlation coefficients (see also function: GetDistributionOfCorrelationCoefficients)

float CScaledCorrelationBC::GetDistributionOfCorrelationCoefficientsBinSize

(

)

Get the size of a bin of the distribution of coefficients of correlation.

Returns:

Returns the size of the bins used to compute the distribution of correlation coefficients (see also function: GetDistributionOfCorrelationCoefficients)

int * CScaledCorrelationBC::GetPBsCoefficientCounts

(

)

Returns how many Point Biserial coefficients of correlation have been averaged for each bin of the scaled correlogram.

Returns:

A pointer to the internal buffer containing the counts of Point Biserial correlation coefficients averaged for each bin. The size of the buffer is (2*iCorrelationWindow+1), and the count of valid Point Biserial coefficients, that were averaged to compute the scaled correlogram at lag 0, is at position iCorrelationWindow

float * CScaledCorrelationBC::GetPBsCoefficientSums

(

)

Returns the sum of valid Point Biserial coefficients of correlation for each bin of the correlogram.

The size of the buffer is (2*iCorrelationWindow+1), and the sum of coefficients for lag 0 is at position iCorrelationWindow. Use this function to get the sum of all valid coefficients that were used to compute the average correlation coefficient for each bin of the ScaledCrossCorrelogram: ScaledCrossCorrelogram[i] = PBsCoefficientSums[i] / PBsCoefficientCounts[i].

Returns:

A pointer to a buffer holding the sum of correlation coefficients for each bin (without dividing them to the counts; the scaled correlogram = the sum of coefficients / the count of coefficients).

Warning:

If no valid coefficient was found for a given bin, the buffer containes a value of NAN at that position. Please check for this in the code!

float * CScaledCorrelationBC::GetScaledCrossCorrelogram

(

)

Returns the buffer with the computed scaled correlogram; returns NULL if there was an error; please check the buffer for NULL before using it!!

The size of the scaled correlogram buffer is (2*iCorrelationWindow+1), and element with lag 0 is at position iCorrelationWindow;

Returns:

Returns a pointer to the internal scaled correlation buffer. Do not write into the buffer!
The buffer has a NotANumber (NAN = -100000000.0) value for bins where correlation could not be defined because of lack of variance in the original signals.

int CScaledCorrelationBC::GetScaleWindow

(

)

Get the size of the current scale segment.

Returns:

Returns the size of the scale segment window

int CScaledCorrelationBC::GetTrialLength

(

)

Get the trial length in original sampling units.

Returns:

Returns the length of the trial in original sampling units

int CScaledCorrelationBC::ModifyAllParameters	(	int	piNewScale,
		int	piNewCorrelationWindow,
		int	piNewTrialLength
	)

Sets all parameters at once.

Parameters:

piNewScale	- The new size of the scale segment
piNewCorrelationWindow	- The new size of the correlation window
piNewTrialLength	- The new length of the trial

Returns:

Returns 0 on success, -1 if the new scale window is too small; -2 if the new correlation window is too small; -3 if the new scale window is larger than the trial size; -4 if the trial length is smaller than the correlation window

int CScaledCorrelationBC::ModifyCorrelationWindow

(

int

piNewCorrelationWindow

)

Set the size of the correlation window; the parameter specifies the correlation window; for example for a window of -100..+100 pass 100 as a parameter.

Parameters:

piNewCorrelationWindow

- The new size of the correlation window

Returns:

Returns 0 on success, -1 if the window is too small; -2 if the window is larger than the length of the trial

int CScaledCorrelationBC::ModifyScaleWindow

(

int

piNewScale

)

Changes the scale segment size for the scaled correlation.

Parameters:

piNewScale

- The new size of the scale segment

Returns:

Returns 0 on success, -1 if the scale window is too small; -2 if the new scale window is larger than the trial size

int CScaledCorrelationBC::ModifyTrialLength

(

int

piNewTrialLength

)

Set the length of the trial.

Parameters:

piNewTrialLength

- The new length of the trial

Returns:

Returns 0 on success and -1 if the trial length is smaller than the scale window; -2 if the trial length is smaller than the correlation window

---

The documentation for this class was generated from the following files:

• ScaledCorrelation-BC.h
• ScaledCorrelation-BC.cpp