Motion segmentation and estimation of active skeletal muscles in ultrasonic image sequences

The segmentation of high contrast anatomical structure from different medical images has been widely explored. However there are seldom algorithms available to reliably segment soft tissue structures of a moving musculoskeletal system with ambiguous boundaries from ultrasound image sequences. Often, ambiguous object boundaries in a single image frame can be easily resolved when dynamic effects are computed based on a sequence of frames. However, the contraction of active skeletal muscles is very complex, such as concentric contraction, isometric contraction and eccentric contraction. These inhomogeneous motion properties make them very difficult to segment. In this paper, we proposed a new algorithm to segment the active skeletal muscles by minimizing an energy function and the curve motion estimation.