Forced oscillation technique: an alternative tool to define the optimal PEEP?