Ultrafast imaging of in vivo muscle contraction using ultrasound

Numerous monitoring techniques are commonly used to study muscular or neuromuscular function. Electromyogra- phy EMG is used to record the electrical activity of the muscle. It can be reported as the sum of action potentials propagating in a muscle's fibers.1 Mechanomyography is the recording of the muscular vibrations produced by the active muscle. It can be used as a monitor of muscle stiffness and could be related to the muscle force production.2 Unfortu- nately, all of these methods have a poor accuracy to assess local measurements and are thus not suitable for fully under- standing the underlying structure and mechanical behavior of the muscle. In order to create maps of the local response of the muscle, a few techniques have been applied to recon- struct the local velocity distributions of the muscle in three or two dimensions 3D or 2D: phase-contrast magnetic- resonance imaging can reconstruct full 3D images of the muscle motion in a stroboscopic way. From these images local strains are calculated.3 Doppler tissue imaging gives the tissues' velocity distribution in a 2D plane and allows axial strain assessments.4 Recently, ultrasound image correlations at low frame rates have also been used to track the muscle motion.5 While very promising, these techniques can only image the muscle up to a few tens of frames per second. These low frame rates cannot be considered high enough to fully visualize the transient phenomena occurring during muscle activation. Recently, ultrafast ultrasound scanners were designed by our group. Our last generation of echographic devices gives access to 2D radio frequency rf images at a few thousand hertz using a modified imaging sequence, i.e., a hundred times faster than any conventional ultrasound scanner. From these rf images, B-mode images are constructed. The cross correlation between two successive images permits us to as- sess the local axial particle velocity, and a complete movie of the axial velocity maps can be finally deduced. Such an ap- proach allows us to provide both very high spatial submil- limetric and temporal accuracy less than a millisecond sam- pling, overcoming all the respective drawbacks of the previously cited techniques. This scanner6 is a 128 multi-