Conventional colonoscopy using a flexible colono-scope remains two major limitations, including patient discomfort and nonintuitive manipulations for surgeons. To address these limitations, robotic colonoscopes have been developed to provide new approaches for conducting colon diagnosis and therapy. However, low autonomy level of most robotic colonoscopes leads to nonintuitive and difficult manipulations, which limits their applications in clinical practice. In this paper, we demonstrate semi-autonomous manipulations of an electro-magnetic actuated soft-tethered (EAST) colonoscope based on visual servo control, which aims to reduce the difficulties of colonoscope manipulation and lower the workload of surgeons. A visual servo controller is established with the magnetic actuation model and kinematic model of the tethered colono-scope. Two semi-autonomous manipulations, including region-of-interest automatic tracking and autonomous navigation with automatic polyp detection, are proposed based on visual servo control, with several image feature extraction methods developed. Experimental validation is conducted with a proof-of-concept EAST colonoscope prototype and a colon phantom. The results show that the system successfully performs visual servoing and semi-autonomous manipulations, which demon-strates the potential of applications in clinical practice.