Gaseous respiratory acetone detection using LAPS (Light Addressable Potentiometric Sensors) is proposed as a supplementary method for the diagnosis of diabetes. First, electric simulation of the LAPS system is studied to obtain the optimum sensitivity of tested odours. Gas phase chromatography is used to obtain correlations between blood sugar levels and acetone from the breath. Then several experiments are carried out to obtain the 2-dimension static and dynamic images of respiratory acetone and air. All experimental results show that the LAPS can be very useful for the detection of respiratory odours, especially the diagnosis of diabetes.