Abstract : Cerebral Palsy (CP) results from a non progressive lesion in the immature brain affecting 4/1000 of new-born. It results in abnormalities of muscle strength and tone (spasticity), causing contractures tending to progress with growth and bone deformities. The aim of this doctoral thesis is to assess a 3D subject specific modelization of gait analysis, bones and muscles of lower limbs in order to allow a better clinical decision making. Gait analysis was provided in a laboratory installed in Lebanon. A database of healthy children was established to better understand gait pattern of children with CP. Uncertainties on gait parameters were assessed, using repeatability study and Monte Carlo technique, in order to objectify comparison between patient's gait and healthy patterns or between pre and post patient treatment. Subject specific 3D reconstruction of lower limb bones in standing position was done by stereoradiography, using EOS® system, for CP and non CP children in order to validate faisability of this technique within children. Skeletal deformities were quantified and their uncertainty calculated. Registering technique of lower limbs with gait analysis frames was evaluated using a stereoradiographic cabin installed in Beirut. MRI acquisitions done in Beirut allowed us to obtain subject specific 3D reconstructions of muscles in lying position. Muscular geometry was quantified by calculating volumes, cross physiological areas and lengthes. 3D reconstructions of muscles in standing position was possible by combining MRI and stereoradiography data, allowing us to calculate muscle/tendon length ration. This work allows a better comprehension of gait abnormalities based on musculo-skeletal deformities for a better clinical decision making of patients with CP.