Technical Note Performance Evaluation of an Automated System for Registration and Postprocessing of CT Scans Nathaniel M. Alpert, Dmitry Berdichevsky, Zakhar Levin, Venkatesan Thangaraj, Gilberto Gonzalez, and Michael H. Lev Abstract: A prototype system optimized for automatic registration of CT scans and applications such as subtraction enhancement and CT angiography was evaluated. Co-registration was performed on 50 studies of various types, including CT angio- grams. Mean registration speed was 109 s; accuracy was 2.0 mm. Examples of sub- traction enhancement and automatic extraction of the vascular tree are provided. Un- attended registration was demonstrated in routine application. Subtraction-enhanced images and extraction of the vascular tree may be useful in clinical practice. Index Terms: Computed tomography—Computed tomography, techniques—Computed to- mography, angiography. Comparison of a patient’s serial or repeat CT scans is a recurring task that could potentially be facilitated by computerized image registration. Representative of the extant methods is work by Pelizzari et al. (1), Alpert et al. (2), Woods et al. (3), Ardekani et al. (4), Wells et al., (5) and Ashburner and Friston (6). Despite the well- validated technical achievement of image registration of brain studies (7–10), few applications have achieved rou- tine clinical practice. Given that adequate computer power is essentially a commodity, what limits clinical deployment of this technology? Economic consider- ations are certainly an issue. However, it is also true that systems have not been engineered to the degree that tech- nologists and/or physicians can perform the required pro- cessing in a timely fashion. Furthermore, registration systems should provide appropriate postprocessing. For example, automatic subtraction of registered noncontrast CT scans from CT scans obtained during the passage of an iodinated contrast agent provides images with higher contrast between perfused and unperfused regions (i.e., subtraction enhancement). In this article, we present a performance evaluation of a commercial prototype (In- tegrated Medical Image Processing Systems, Watertown, MA, U.S.A.) capable of completely unattended registra- tion of brain CT scans. The evaluation addresses the speed and accuracy of registration and presents several applications of image registration with postprocess- ing that may ease diagnosis and extend the range of applications. METHODS CT Scans Fifty scan pairs were chosen retrospectively from a digital picture-archiving system (AGFA Impax RS3000 1K review station; AGFA Technical Imaging Systems, Richfield Park, NJ, U.S.A.). Twenty-five were CT angi- ography (CTA) pairs obtained with and without iodin- ated contrast material. The CTA scan protocol was de- signed to provide simultaneous high resolution vascular images of the intracranial circulation and, without addi- tional scanning sequences or contrast agent administra- tion, whole-brain 3-mm-thick axial perfused blood vol- ume-weighted images. Patients were scanned in the su- pine position with their head restrained to reduce the blurring effects of motion. Coverage was from skull base to vertex using contiguous axial slices parallel to the planum sphenoidale. Noncontrast CT scanning was per- formed using a HiSpeed Advantage helical CT scanner From the Department of Radiology, Massachusetts General Hospital, Boston (N. M. Alpert, D. Berdichevsky, Z. Levin, G. Gonzalez, and M. H. Lev), and Integrated Medical Imaging Systems, Inc., Watertown (V. Thangaraj), MA, U.S.A. Address correspondence and reprint requests to Dr. N. M. Alpert at Division of Nuclear Medicine, Massachusetts General Hospital, Fruit St., Boston, MA 02114, U.S.A. E-mail: alpert@pet.mgh.harvard.edu Journal of Computer Assisted Tomography 25(5):747–752 © 2001 Lippincott Williams & Wilkins, Inc., Philadelphia 747