While Magnetic Resonance (MR) images are invaluable because of their high degree of anatomical differentiation and their lack of ionizing radiation, they contain an inherent distortion which can exceed ten percent. Although the existence of this distortion has not been well documented by MR device manufacturers, without distortion compensation, quantitative MR measurements, regarding position, length, area and volume, are likely to be erroneous. During Phase I, we performed a feasibility study in which phantom calibration data, and data from a registration hat worn by patients were used to correct for distortion in MR images. During Phase II, we propose to validate and extend this work. This will include design, construction and testing of a special phantom and fiducial helmet, the collection of an MR and CT scan data base from human subjects, and the further development of our distortion correction algorithms. A thorough evaluation of the algorithms' performance will result in the specification of the design for the system to be implemented in Phase III.
