An optical image-based tracking system determines the position and orientation of objects such as biological materials or medical devices within or on the surface of a human body undergoing Magnetic Resonance Imaging (MRI). Three-dimensional coordinates of the object to be tracked are obtained initially using a plurality of MR-compatible cameras. A calibration procedure converts the motion information obtained with the optical tracking system coordinates into coordinates of an MR system. A motion information file is acquired for each MRI scan, and each file is then converted into coordinates of the MRI system using a registration transformation. Each converted motion information file can be used to realign, correct, or otherwise augment its corresponding single MR image or a time series of such MR images. In a preferred embodiment, the invention provides real-time computer control to track the position of an interventional treatment system, including surgical tools and tissue manipulators, devices for in vivo delivery of drugs, angioplasty devices, biopsy and sampling devices, devices for delivery of RF, thermal energy, microwaves, laser energy or ionizing radiation, and internal illumination and imaging devices, such as catheters, endoscopes, laparoscopes, and like instruments. In other embodiments, the invention is also useful for conventional clinical MRI events, functional MRI studies, and registration of image data acquired using multiple modalities.