• Rate of success of the secondary functional task [ Time Frame: Third day ]
  The secondary functional task will be the one that will not be retained as a main task (primary outcome), and consist of grasping, moving and releasing either a glass (palmar grasp) or a spoon (key grip). The assessment will consist of 24 trials with the active or inactive neuroprosthesis (12 activated and 12 inactivated), the order of the trials with / without neuroprosthesis being randomized in blocks of at least three trials. In order to limit the fatigue potentially induced by the repetition of stimulations, a pause between each trial will be respected if necessary. The maximum time allowed for the completion of each test will be 1 minute. A success corresponds to a complete completion of the functional task in at least 2/3 of the trials (i.e. 8/12 trials) with the activated neuroprosthesis. The success / failure score will be assessed secondarily from video recordings by a blind evaluator of the activation or not of the neuroprosthesis.
• Comparison of the Action Research Arm Test (ARAT) performed with the inactivated (first day) and activated (day three) neuroprosthesis [ Time Frame: First and third day ]
  The ARAT assess different unimanual modes of prehension. The score is between 0 (the worst) and 57 (the best). The cotation will be performed secondarily from videos by a blind evaluator.
• Psychosocial Impact of Assistive Devices (PIADS) questionnaire [ Time Frame: Third day ]
  The PIADS assess the psychosocial impact of the use of the neuroprosthesis in the life of the person. It is composed of 3 subscales (competence, adaptability and self-esteem), each rated between -3 (worst) and +3 (best).
• Subscale "Device" from the questionnaire Quebec User Evaluation of Satisfaction with Assistive Technology (QUEST) [ Time Frame: Third day ]
  The subscale Device of the QUEST assess the satisfaction and tolerance with respect to the characteristics and use of the neuroprosthesis. The score is between 0 (the worst) and 5 the best).

Stroke is the leading cause of disability in adults. The improvement of the grasp abilities remains a challenge in the 50% of post-stroke subjects who have not recovered functional grasping due to paralysis of the finger's muscles (lack of active opening of the hand).

The use of functional electrical stimulation of the prehension muscles in order to restore grasp abilities, called grasp neuroprosthesis (GP), remained confidential in post-stroke subjects while their development was important in tetraplegic subjects. GP can provide a correct hand opening with significant functional gain, but one of the major issues corresponds to the control modalities that are not adapted to the specific impairments of post-stroke subjects.

This project proposes to assess the functional contribution of an innovative autopilot closed-loop GP targeting the extensor muscles of the fingers. The main hypothesis is that the use of GP will restore grasping abilities in subjects who have lost this ability due to post-stroke paralysis.

The main objective is to assess the impact of using an autopilot closed-loop GP on the ability to perform a standardized task of grasping, moving and releasing either a glass (palmar grasp) or a spoon (key pinch), compared to the absence of GP use.

The secondary objectives of the study are: (1) to assess the impact of the GP on unimanual grasp; (2) to assess which are the preferential modes of control; (3) to assess the psycho-social impacts of GP, and (4) to assess the subject's satisfaction and tolerance to the characteristics and use of GP.

The investigators plan to include 20 post-stroke hemiplegic subjects over a period of 9 months as part of a prospective, monocentric, multi-crossover, blinded evaluation study. Subjects will have active finger extension deficit secondary to stroke, with preservation of proximal movements. Each subject will be his own control (self-pairing).

Each subject will be evaluated three times, the protocol adding approximately 1½ hours of daily assessment to routine care already received. The first visit will collect clinical data after informed consent collection. The second visit will allow to choose the optimal mode of control of the GP among 8 modalities. The third visit will test the functional gain provided by the use of GP, by comparing the success or failure of carrying out functional tasks with inactive and active GP.

• No Intervention: Inactive neuroprosthesis
• Experimental: Active neuroprosthesis
  Intervention: Device: Functional electrical stimulation

Inclusion Criteria:

• The patient must have given free and informed consent and signed the consent;
• The patient must be an affiliate or a beneficiary of a health insurance plan;
• The patient is hospitalized as part of routine care and available for at least 3 consecutive days of follow-up during hospitalization;
• Motor deficiency of the upper limb due to a hemorrhagic or ischemic stroke;
• Stroke more than one months old;
• Inability to perform an active extension of the long fingers (opening of the hand) to voluntarily seize an empty glass with a palmar grip (grasping task in the ARAT scale), while the subject can hold the previously placed glass passively in the hand; and / or
• Inability to perform an active thumb extension to voluntarily grasp the handle of a tablespoon (flat, like a key) with a pulpo-lateral thumb-index or key-grip (grasping task in the Wolf Motor Function Scale Test), while the subject can hold the spoon previously placed passively between thumb and index;
• Ability to sit on a chair for at least 2 hours.

Exclusion Criteria:

• The subject participates in another interventional study;
• The subject is in an exclusion period determined by a previous study;
• The subject is under the protection of justice, guardianship or curatorship;
• The subject refuses to sign or give consent;
• It is not possible to give the subject enlightened information.
• The patient is pregnant, parturient, or breastfeeding;
• Patient with pacemaker;
• Unstable epilepsy;
• Unstable cardiovascular pathology (coronary heart disease, major hypertension, heart failure);
• Dermatological problems counter-indicating the application of surface electrodes;
• Musculotendinous retractions or joint stiffness of the fingers and wrist preventing passive opening of the hand sufficient to perform the functional tasks evaluated;
• Active elbow extension limited to not reaching the ipsilateral knee, the subject sitting (limitation of the approach);
• Upper limb pain limiting movements;
• Major sensory disorders corresponding to a score of the Modified Erasmus Nottingham Sensory Assessment English version of the upper limb <10/44;
• Severe aphasia with aphasia severity scale of the Boston Diagnostic Severity Aphasia Examination <= 3, indicating that there may be a clear decrease in verbal fluency or ease and speed of understanding, with no significant limitation expression or communication;
• Unilateral spatial negligence highlighted with the bell test if the difference between omissions in the left and right fields is greater than or equal to 6;
• Common extensor digitorum muscle and / or extensor thumb muscle non-stimulable with the neuroprosthesis.