DETECTION OF SUBTLE BRAIN CHANGES USING SUBVOXEL REGISTRATION AND SUBTRACTION OF SERIAL MR-IMAGES

Objective: The aim of this study was to examine the potential of accurate image registration for detecting subtle changes in the brain. Materials and Methods: Isotropic T1-weighted volume images were obtained in 10 normal subjects and five patients on two or more occasions (including pre- and postcontrast studies). The images were segmented and a 3D rigid body translation and rotation technique was used with sine interpolation to precisely match the images using a chi(2)-test. The registered images and the subtraction images produced from them were used to detect changes in signal intensity, site, shape, and size of the brain. Results: Small changes due to differences in orientation of the head, growth, and development as well as inhalation of oxygen and carbogen (95% O-2/5% CO2) were observed in normal subjects. Changes were also observed in patients with minor head trauma, a meningioma, an astrocytoma, and multiple sclerosis. Differences due to contrast enhancement and surgery and/or anesthesia were also seen. Conclusion: With use of subvoxel registration, subtle changes in the brain were detected in a variety of physiological and clinical situations where differences have hitherto been difficult or impossible to detect.