The perfusion of the deeply different areas coexisting in ARDS lung generates two type of venus-admixure : true shunt and shunt effect. Briefly, true shunt and shunt effect are related to non-aerated and poorly-ventilated lung areas perfusion respectively. Practically, it is not possible to quantify the true shunt and the shunt-effect directly, but we can measure the "total-shunt", i.e. the sum of two kind of venus-admixured. As it is known from classical physiology, the effects of FiO2 variations on PaO2 are deeply different depending on the type of venous admixures. In the case of a patient with virtually only true shunt, any FiO2 variation will not modify the PaO2. On the other hand, in a patient with virtually only a shunt effect, increasing the FiO2 the PaO2 will progressively increase. Indeed, ventilating a patient presenting only a shunt effect (without true shunt) with pure oxygen will generate a "normal" P/F ratio. In patients ventilated with high PEEP levels it is difficult to predict based on the P/F ratio the relative amount of true shunt and shunt effect. However, patients presenting a significant shunt effect when ventilated with higher PEEP level would likely have "unstable" poorly aerated lung areas that need PEEP to remain opened. In those patients, a PEEP-weaning protocol based on the P/F ratio could induce significant alveolar de-recruitment and clinical deterioration. On the contrary, patients with less shunt effect would be likely less prone to alveolar de-recruitment and would benefice from PEEP decrease.

the investigators set up a study to test the hypothesis that patients presenting a significant shunt effect when ventilated with higher PEEP levels could be prone to significant alveolar de-recruitment at the PEEP weaning attempt. To do so, the investigator has to measure the total shunt at higher PEEP level in patients ready to be weaned from the PEEP (according to the Mercat protocol), at three FiO2 levels (i.e. 0,3, 0,6 and 1). The shunt effect is quantified as the difference in total shunt between FiO2 0,3 and .The hypothesis of this study is that the shunt effect could predict alveolar de-recruitment when weaning PEEP.

Inclusion Criteria:

• Age major than 18
• 1-2 days since the beginning of mechanical ventilation
• Moderate or severe ARDS

Exclusion Criteria:

• Patients with a story of chronic obstructive pulmonary disease (COPD)
• Patients with asthma
• Patients with neurological disease
• Patients with heart disease
• Patients with cardiopulmonary.