Generate x-z or y-z cross sections and then plot the intensities of a line drawn through the membrane(s) and cytoplasm. You could use the "Image, Stacks, Reslice" function. Or "Image, Stacks, Orthogonal Views" would also make a cross sectional view. Then draw a line through the cross section and use "Analyze, Plot Profile" to get the intensity plot. I find it helpful when doing intensity profiles to make the line at least a few pixels wide to average out noise. If you double click the straight line tool you can increase the width of the line and the intensity values of the line plot will be the mean across the line width.

You should also add a membrane marker in your staining. In this way you can quantify the colocalising signals in the membrane with non-colocalising signals in the cytoplasm.

Looking at the picture you provided, I would say that there is none of your protein in the membrane, most of it is in submembraneous vesicles.

If you have many cells to analyze I recommend an automated analysis.

Best of luck!

Just as a follow up, the method I described would work if the distribution was largely in the plasma membrane (which is what I originally assumed). Then you would see two peaks across the cross sectional line profile. If it is more associate with membranes in organelles and vesicles inside the cell, then Tilo's method would be more appropriate.