Installation and Use - mwgeurts/gamma GitHub Wiki

The following sections provide instructions on installing, configuring, and executing CalcGamma().

Installation
Compatibility
Requirements
Syntax
Example
Input Arguments
Output Arguments

Installation

To install CalcGamma(), copy the CalcGamma.m MATLAB file from this repository into your MATLAB path. If installing as a submodule into another git repository, execute git submodule add https://github.com/mwgeurts/gamma.

Compatibility

See Unit Testing for system configuration details on which this function was validated.

Requirements

The Parallel Computing toolbox and a CUDA-compatible GPU are required to run GPU based interpolation. CPU interpolation is automatically supported if not present. Parallel execution of 1-D interpolation using GPU was introduced in MATLAB R2013a, 2-D interpolation in R2013b, and 3-D interpolation in R2014a.

To test whether the Parallel Computing toolbox is installed and your GPU device is compatible, execute gpuDevice(1) in MATLAB. If successful, the response should be a CUDADevice object.

Syntax

To compute Gamma, execute CalcGamma() with one of the following syntaxes:

gamma = CalcGamma(reference, target, percent, dta);
gamma = CalcGamma(..., 'OptionName', OptionValue);

Example

reference.start = [-10 -10]; % mm
reference.width = [0.1 0.1]; % mm
reference.data = rand(200);

target.start = [-10 -10]; % mm
target.width = [0.1 0.1]; % mm
target.data = rand(200);

percent = 3;
dta = 0.5; % mm
local = 0; % Perform global gamma
   
gamma = CalcGamma(reference, target, percent, dta, 'local', local);

Input Arguments

Variable	Type	Description
reference	structure	Reference data containing the fields `start`, `width`, and `data`. The field `start` is an array containing the coordinates along each dimension of the first voxel, `width` is an array containing the width of each voxel along each dimension, and `data` is an n-dimensional array
target	structure	Target data containing the fields `start`, `width`, and `data`. The field `start` is an array containing the coordinates along each dimension of the first voxel, `width` is an array containing the width of each voxel along each dimension, and `data` is an n-dimensional array
percent	double	Gamma absolute criterion percentage
dta	double	Gamma Distance To Agreement (DTA) criterion, in the same units as the reference and target width structure fields

The following options can be passed to this argument as name/value pairs:

Variable	Type	Description
local	logical	Indicates whether to perform a local (1) or global (0) Gamma computation. If not present, the function will assume a global Gamma computation.
refval	double	Reference value for the global absolute criterion. Is used with the percentage from varargin{3} to compute absolute value. If not present, the maximum value in the reference data is used.
restrict	logical	Restricted search flag. If 1, only the gamma values along the X/Y/Z axes are computed during 3D comptation. If 0 or not provided, the entire rectangular search space is computed.
res	integer	DTA resolution factor. The number of distance steps equal the resolution factor multiplied by the limit. If not provided, the factor is 100 for 1D, 50 for 2D, and 20 for 3D calculations.
limit	double	DTA limit. This number determines how far the function will search in the distance axes when computing Gamma. This also therefore specifies the maximum "real" Gamma Index value. If not provided, the function will use 2.
cpu	boolean	Set to 1 to force CPU computation.

Output Arguments

Variable	Type	Description
gamma	double array	Array of the same dimensions as `target.data` containing the computed Gamma index values