Systematic errors and susceptibility to noise of four methods for calculating anatomical dead space from the CO2 expirogram.