Toward region of interest computer tomography

Abstract

Computed Tomography (CT) is widely used in the modern clinical settings. In certain procedures, the region of interest (ROI) is often considerably smaller than the imaged field of view (FOV), thereby subjecting the patient to extra dose. For these procedures, we propose a method of filtered region-of-interest (FROI) CT. In this procedure, a predetermined ROI is imaged with standard x-ray intensity, while surrounding areas are imaged using a substantial lower x-ray intensity by interposing an x-ray attenuator in the beam.For the FROI-CT acquisitions in this study, a gadolinium filter with a circular central opening is placed in the x-ray beam of a standard clinical rotational angiography system. The resulting image contains a high intensity ROI, a low intensity region surrounding the ROI, and a transition region between these two. Three-dimensional reconstruction using these images would result in artifacts. Therefore, the intensities in the images are equalized prior to reconstruction. To equalize the intensities, first two images are obtained, one unfiltered and one with a filter in place. The corresponding data in the two images are used in a linear least-squares fit to determine the equalization function. The transition region is equalized using a radial filter technique, based on a comparison of the data on either side of the transition region after intensity equalization. The technique was evaluated using rotational angiographic sequences of a head phantom obtained with and without the filter in place.Differences between conventional (unfiltered) and FROI-equalized images of the head phantom were approximately 5%. Differences in reconstructed images (conventional and FROI) were 7% on average inside the reconstructed ROI. These results are comparable to those obtained for two separate standard acquisitions. A 50% dose reduction was obtained for a 50% FOV radius for the filter. These results indicate that FROI-CT can provide the physician with the image detail comparable to conventional image acquisition while reducing dose to the patient.