Can the anode heel effect be used to optimise radiation dose and image quality for AP pelvis radiography?