New video available on the AMARSi YouTube Channel
This movie illustrates the steps involved in constructing compatible and incompatible virtual surgeries. Examples are based on the data of subject 2 (as in Fig. 2 and 3) of Berger et al. 2013 (http://www.amarsi-project.eu/pub/differences-adaptation-rates-after-virt...). The movie begins with a still image of the experimental apparatus and a subject performing the experiment (on screen caption: “The subject moves a cursor in the virtual environment by applying isometric forces to an instrumented handle”). Next (00:05), the mapping of applied force to cursor displacement is illustrated by the side-by-side animation of a force arrow (“Force”, left) and the displacement of the cursor (“Displacement”, right; on screen caption “The cursor is displaced proportionately to the magnitude and direction of the applied force.”). This is followed (00:10) by the visualization of an example of the recorded muscle activations as change in color of the muscles in a 3D musculoskeletal model (on screen caption: “Muscle activations are simultaneously recorded via surface electromyography (EMG).”). Next (00:15) the same muscle activation patterns and a few additional examples (left) are shown together with the corresponding cursor displacement (right, on scree caption: “The subjects performs a number of planar force-target reaching trials while we record EMG and end-point force.”). A still title (00:25, “The we use multiple regression to estimate a linear mapping from EMG to end-point force.”) introduces (00:28) the illustration of the EMG-to-force mapping (“Under this mapping, each muscle has an end-point force output vector.”) as muscle activations (left) and force arrows (right) for individual muscle and for the whole set of muscles (00:36, on screen caption: “Together, the muscles redundantly span the planar task-space.”). A still title (00:41, “Now we can control the virtual cursor with the EMG signals.”) introduces the illustration of EMG control (00:45, “The cursor’s movement is directly controlled by muscle activation, via the EMG-to-force mapping”) by showing muscle activations (left) and corresponding cursor displacements (right). The following title (00:55, “We use nonnegative matrix factorization to identify muscle synergies in the recorded EMG signals.”) introduces the illustration of the four synergies (01:00, “Each synergy consists of a specific balance of muscle activation.”) and their forces (01:04, “When the EMG-to-force mapping is applied, each synergy has an effective end-point force vector.”) for subject 2. Next (01:08) muscle and synergy forces are compared (“As with the direct muscle output forces, the synergy output forces span the task-space.”). The next title (01:13, “We can perform a “virtual surgery” on the subject by modifying the EMG-to-force mapping.”) introduces the illustration of the effect of a surgery on muscle output force vectors (01:18, “After the virtual surgery, the muscles have new output force vectors.”) and on synergy output force vectors (01:27, “Since the synergies combine the muscle output forces, the virtual surgery change the synergy output force vectors as well.”). The effect of the rotation in muscle space is illustrated by animating the force output vectors computed with a sequence of EMG-to-force mappings with rotation angles starting from 0 and ending to the final rotation angle. Next (01:36) the difference between compatible (“We construct two types of virtual surgery. After a compatible surgery, the synergy output vectors still span the task-space.”) and incompatible (01:45, “After an incompatible surgery, the muscle output vectors still span the task-space, but the synergy output vectors do not.”) surgeries is illustrated.