Gait analysis is clinically used as a diagnostic tool to quantify gait issues associated with neuromuscular or locomotion disorders. The variability of different spatio-temporal parameters during the gait cycles provides clinical information related to state of patients and the progress of certain diseases. Usually, gait analysis based on spatio-temporal parameters requires costly-instrumented walkways and video-based motion capture systems, which are only available in specialized clinical scenarios. Hence, Inertial sensors-based gait analysis has been proposed with promising results, but it is still considered a challenging task. In this paper, a novel method for estimating spatio-temporal parameters of gait by using an inertial sensor network is proposed. For doing that, a set of 7 inertial sensors are setting into a biomechanical model to capture both position and rotation of body segments of lower limbs. Then, gait cycles are segmented by including information from angles of the hip, knee and ankle joints, respectively. So, a quaternion-based motion analysis is carried out for defining gait phases by detecting gait events. The proposed method was validated using a camera-based motion system, respectively. The obtained results demonstrate that the estimated spatio-temporal parameters are equal to the expected values obtained by camera motion system widely used in real clinical scenarios. Therefore, the proposed method could be used in clinical applications.