S14 Abstracts / Gait & Posture 49S (2016) S1–S30 surgical plan and objectives after review of the laboratory data were used as a measure of the influence exerted by movement analysis at different levels of the UL (elbow, wrist and fingers). Results: A sample of 20 patients (mean age 36,55 years) with spastic UL was assessed. Investigations were carried out in 2 patients for elbow muscles, in 9 patients for wrist/hand muscles, and in 9 patients both seg- ments were investigated. Overall 126 muscles were investigated: 75 Muscles with DE, 38 muscles with NB and 13 muscles for both techniques. After instrumental assessment, the surgical plan changed in 14 of the 20 patients; in 41 muscles (32%) investigated the surgical plan was modified, in 32 muscles of wrist and fingers and in 9 elbow muscles The purpose of surgery changed in only 2 patients. The most frequent surgical technique to undergo change after instrumental assessment was muscle lengthening. Discussion: These preliminary results allow us to hypothesize that instrumental evaluation can influence the surgical planning of spastic UL. DE and NB seem to have a similar impact in the planning of surgery in patients with spastic paretic upper limb deformity. The higher frequency of changes in the distal as compared to the proximal segment may be compatible with the higher anatomical and functional complexity of the wrist and hand. Reference [1] Keenan MA, et al. Arch. Phys. Med. Rehabil 2003;84(2):291–6. http://dx.doi.org/10.1016/j.gaitpost.2016.07.042 Session 5 – Movement analysis with wearable sensors O21 In vivo identification of the shoulder joint centre of rotation using a magneto-inertial sensor M. Crabolu 1,* , D. Pani 1 , L. Raffo 1 , M. Conti 2 , P. Crivelli 2 , A. Cereatti 2,3 1 University of Cagliari, Italy 2 University of Sassari, Italy 3 Politecnico di Torino, Torino, Italy Introduction: The reconstruction of the human movement requires the definition of a multi-segmental kinematic model of the locomotor apparatus. In several biomedical applications, either for the purpose of assessment or to implement a rehabilitation inter- vention, a high accuracy and precision in the estimation of the joint parameters defining the kinematic model is required. In particular, the spherical joint is commonly employed to model the kinematic behaviour of several human joints. The position of the centre of rotation (CoR) can be obtained according to functional approach which entails the execution and the recording of ad hoc joint motions [1]. In this work, we investigated the accuracy in the in- vivo estimation of the shoulder joint CoR using a magneto-inertial measurement unit (MIMU). The protocol was analyzed under dif- ferent experimental conditions (different movement speeds, joint motion types and ranges of motion). Magnetic resonance imaging (MRI) was used as a gold standard. Methods: Five healthy subjects were analysed (3 males, 2 females, age: 35 ± 5 years, mass: 65 ± 5 kg). The MRI of subject’s right scapula and humerus were acquired. A phantom of the MIMU, including a MR visible spherical cavity, was attached to the humerus using a custom clip (Fig. 1c). After the MRI acquisition, the Fig. 1. Experimental setup. (a) MRI-based humerus and phantom position recon- struction; (b) a subject executing one of the stereotyped shoulder movement; (c) the MIMU, its phantom and the custom clip. phantom was replaced with the actual MIMU comprising three- axial accelerometers, gyroscopes and magnetometers (MTw2, Xsens, sample frequency: 100 Hz, dynamic accuracy: (Roll/Pitch) 0.75 deg RMS, (Heading) 1.5 deg RMS). The MIMU was used to record movement data during active shoulder stereotyped move- ments (cross-like and star-like, Fig. 1b). Data were recorded under six different experimental conditions (2 joint velocity × 2 type of motions × 2 range of motions). Three repetitions for each condi- tion were recorded for a total of 24 trials per subject. The distance d between the CoR of the shoulder and the MIMU (Fig. 1a) was esti- mated according to rigid body kinematics equations as described in a previous work [2]. Results: For each trial, the error e of the distance d, between the MIMU origin and the CoR, as estimated using the inertial data and the gold standard, was computed. The mean absolute errors and the standard deviation, computed over subjects, type of movement and ranges of motion were equal to 7 ± 6 mm for the faster movements and 12 ± 9 mm for the slow ones. Discussion: The finding of the present study indicates that magneto-inertial sensors can be used for the in-vivo functional estimation of the CoR in the human shoulder. The accuracy level was comparable to state-of-the-art optoelectronic stereo- photogrammetry techniques [3] with the advantages of using a wearable and cost effective technology. Changes in the experi- mental conditions affected both precision and accuracy of the CoR estimates. The most critical factor resulted to be the speed of the joint movement indicating that slow motions should be avoided. References [1] Cappozzo A. Hum. Mov. Sci 1984;3:27–50. [2] Crabolu M, et al. IEEE Sensors applications symposium 2016; in press. [3] Lempereur M, et al. J. Biomech 2010;43:370–4. http://dx.doi.org/10.1016/j.gaitpost.2016.07.043