[en] The main purpose for studying collisions between heavy nuclei, in the 200-2000 MeV per nucleon energy range, is to determine the equation of state and the properties of dense and hot nuclear matter. The insensitiveness of the inclusive data to the equation of state has led experimental physicists to build large solid angle detectors capable of detecting simultaneously the tens of particles emitted in each event. Such measurements allow to estimate the impact parameter, the reaction plane on an event-by-event basis, and to calculate various global variables involving all particles emitted in each event. In this thesis, we study global variables which characterize the nuclear matter collective flow in a direction which is different from the direction of incident motion, for argon-nucleus collisions at 400 and 600 MeV by nucleon and for neon-nucleus collisions at 400 and 800 MeV by nucleon. The measurements have been performed with the DIOGENE detector installed at SATURNE. For the argon-beam experiments, two parallel plate avalanche counters have been used to locate the interaction point of each incoming ion with the target, which improves the reconstruction of the particle tracks in the DIOGENE central chamber. Double differential cross-sections, in the reaction plane and in the plane orthogonal to the reaction plane, are fitted with two-dimensional Gaussian distributions. Through this procedure, we get rid of geometrical acceptance effects. Several quantities, related to the collective flow (flow angle, aspect ratios, flow parameter), are thus extracted and corrected for the fluctuations of the estimated reaction plane. For argon-nucleus collisions at 400 MeV by nucleon, our results are in agreement with results obtained by other groups with different methods. For argon-nucleus collisions, the discrepancy between our experimental results and predictions of intranuclear cascade calculations is increasing with the mass asymmetry of the colliding system