Abstract : The T2K experiment is an off-axis long baseline neutrino flavour oscillation experiment. In T2K, the neutrino beam produced by an accelerator at J-PARC, Tokai, is detected using the Super Kamiokande (SK) detector, Kamioka, located 295 km away from Tokai. The accelerator produce a very pure nu_mu beam. During the propagation of neutrinos between their production point and SK they may change flavour. This effect is the neutrino flavour oscillation. The main objective of T2K is to measure some parameters describing the neutrino flavour oscillation, in particular of theta13 and deltaCP. This determination is performed by measuring the nu_mu -> nu_e oscillation probability, via the nu_e appearance search. In 2011, T2K was the first experiment to observe nu_e appearance and published a 2.5 sigma evidence that theta13 is not null. In this thesis we present the work done on the evaluation of the systematic error on the pi0 background reconstruction efficiency. The pi0 background is one of the main background sources affecting the nu_e appearance search through the detection of nu_e charged current quasi-elastic events at SK. Two different and complementary approaches have been developed. First, we have started developing a new device that partially reproduces the topology of physics events. Second, we have created a specific sample to estimate the systematic uncertainty of pi0 events reconstruction efficiency. This result was used in the official 2011 nu_e search. Furthermore, results from this sample will be used in future extended oscillation analysis. We also present the first study of the leptonic CP violation phase (deltaCP) using T2K nu_e data, assuming the recently measured value of theta13 by reactor experiments.