AUTOMATED CORRELATION OF INTRAVASCULAR ULTRASOUND IMAGES WITH ANGIOGRAPHY

One limitation of intravascular ultrasound (IVUS) is the restriction t o viewing one cross-sectional image at a time. Computerized three-dime nsional reconstructions of IVUS images have been developed in an attem pt to overcome this limitation. These algorithms, however, are limited by artifacts from catheter movements and rotation within large vessel s. Consequently, this technique has been applied only to straight segm ents of small caliber vessels. Contrast angiography has long been the standard for vascular imaging. In order to take advantage of both cont rast angiography and IVUS, we developed a computer procedure to automa tically correlate IVUS images with their corresponding locations on co ntrast angiograms, and to display both images in a side by side format . Models of the aortic arch and aorto-ileo-femoral system were constru cted with artificial plaques located at various sites. The models were filled with iodinated contrast media and radiographic images were obt ained. Timed pull-backs were performed in both models in order to obta in sets of serial cross-sectional images. For each data set, a digitiz ed set of 75 serial IVUS images and model angiographic images were loa ded in the computer procedure. We then correlated at least one IVUS im age containing a known landmark with its position on the model angiogr am. The procedure then automatically displayed sequential ultrasound i mages along with their corresponding positions on the reference angiog ram. We analyzed the error of this algorithm as a function of the numb er of correlation points used. The maximum error was 4 mm over a total pullback distance of 130 mm (relative error of 3%). This algorithm wa s subsequently used to correlate IVUS images obtained from the aortic arch of a patient with their corresponding positions on an aortogram. Our results demonstrate that computer-based correlation of IVUS images with their corresponding positions on angiograms is accurate, may enh ance the use of IVUS to assess vascular pathology, and provides an alt ernative to three-dimensional reconstructions.