Systems and methods for tracking a target volume, e.g., tumor, in real-time during radiation treatment are provided. The system includes a memory to store a pre-acquired 3D image of the anatomy of interest in a first reference frame and a processor, operative coupled with the memory, to receive, from an ultrasound probe, a set-up ultrasound image of the anatomy of interest in a second reference frame. The processor further to establish a transformation between the first and second reference frames by registering the set-up ultrasound image with the pre-acquired 3D image and receive, from the ultrasound probe, an intrafraction ultrasound image of the anatomy of interest. The processor further to register the intrafraction ultrasound image with the set-up ultrasound image and track motion of the anatomy of interest based on the registered intrafraction ultrasound image.