get_decision_gradient.m
2cb18aae5832master

get_decision_gradient.m
View Options

			%% This function computes a decision matrix W for TV minimization
			% If two neighbour voxels are in different areas with respect to activation
			% (e.g GM vs WM+CSF) then their finite difference is not considered in TV
			% minimization
			%
			% Inputs:
			% - Map is the probabilistic map volume
			% - sigma determines how sharp the weights are
			% - t is the number of time courses
			%
			% Outputs:
			% - wx, wy, wz are the weight volumes (along X, Y and Z)
			%
			% Implemented by Younes Farouj, 29.04.2016

			function [param] = get_decision_gradient(param)

			% dx, dy and dz are 3D matrices depicting the difference between two elements of the GM
			% map along the X, the Y and the Z directions, respectively. Zero-padding
			% is done at the end of the volumes for each case.

			[dx,dy,dz] = gradient3D_full(param.GM_map);

			wx_temp=1-abs(dx);
			wy_temp=1-abs(dy);
			wz_temp=1-abs(dz);

			param.weight_x = wx_temp;
			param.weight_y = wy_temp;
			param.weight_z = wz_temp;


			end