From the beginning of the project, we focused on the development of the framework for the recording and analysis of multiple muscle groups during isometric contractions in stroke individuals. For this purpose, two devices for the recording of the upper and lower limb muscles were developed. The device for the upper limb consists of a flexible system made of aluminum to record flexion and extension forces of the wrist and finger muscles on both arms. The device included two force cells and the arm support was modified in order to avoid the direct contact between the EMG electrodes and the metal parts of the recording system. Similarly, a system for the recording of the lower limb forces (dorsi and plantaflexion of the ankle) was designed. The device provides the possibility to record ankle forces at different angles (neural position +- 30 degrees) and accommodate patients with different levels of plasticity. Both systems were combined with a two channels force amplifier and a 256 channels high-density EMG amplifier. Three surface EMG matrices of 64 electrodes were used for both the upper (two matrices on the flexors and one on the extensors muscles on the forearm) and the lower (soleus, medial gastrocnemius and tibialis anterior muscles) limb. A feedback of the extension/flexion force of the hand fingers or dorsi/plantaflexion force of the ankle was programmed in MATLAB and showed to the subject in real-time. In order to understand the limitations of the experimental framework and quantify possible effects of fatigue in stroke individuals, we performed a large set of protocols on a group of healthy and young individuals. Briefly, the experiments investigated the behavior of populations of motor units in a broad range of muscles, forces, and tasks. This provided a baseline for comparison with the stroke recordings. The main findings of this set of experiments showed for the first time that the neural activity of populations of motor units can be extracted in a variety of conditions and tracked across levels, sessions or after training interventions. Additionally, the combination of these innovative decomposition tools and the correlation analysis of populations of motor units spike trains showed the clear neural signature of motor control. For example, during a short session of force-matching training, we demonstrated that the control-to-neural noise ratio was improved after the repetition of the task. During the second year of the project, in acute stroke patients, we were able to track neural changes in five patients in three sessions (T0 admittance, T15 days after admittance in the clinic and T45) during the rehabilitation period. We are currently finishing the data collections in the stroke individuals. The results of the project have been disseminated in different forms. Until now, the signals recorded on the volunteers and the computational tools developed during the project have been included in several journal publications and five additional publications are in the final preparation phase. The results of the project have been presented in five international conferences and have been delivered in form of seminars in different laboratories worldwide. Finally, the project was presented in two events organized by the host institution for the general public.
