avoid use of mean function due to behavior in Octaves nan package.

Author: wahln

matRad/optimization/+DoseObjectives/matRad_MeanDose.m

@@ -110,15 +110,15 @@
 
     methods (Access = protected)
         function fDose = objectiveQuadraticDiff(obj,dose)
-            fDose = (mean(dose(:)) - obj.parameters{1})^2;
+            fDose = (sum(dose(:))/numel(dose) - obj.parameters{1})^2;
         end
 
         function fDoseGrad = gradientQuadraticDiff(obj,dose)
-            fDoseGrad = 2*(mean(dose(:))-obj.parameters{1}) * ones(size(dose(:)))/numel(dose);
+            fDoseGrad = 2*(sum(dose(:))/numel(dose)-obj.parameters{1}) * ones(size(dose(:)))/numel(dose);
         end
 
         function fDose = objectiveLinearDiff(obj,dose)
-            fDose = abs(mean(dose(:)) - obj.parameters{1});
+            fDose = abs(sum(dose(:))/numel(dose) - obj.parameters{1});
         end
 
         function fDoseGrad = gradientLinearDiff(obj,dose)