Algorithm for image reconstruction in multi-slice helical CT

Abstract

Efforts are being made to develop a new type of CT system that can scan volumes over a large range within a short time with thin slice images. One of the most promising approaches is the combination of helical scanning with multi-slice CT, which involves several detector arrays stacked in the z direction. However, the algorithm for image reconstruction remains one of the biggest problems in multi-slice CT. Two helical interpolation methods for single-slice CT, 360LI and 180LI, were used a starting points and extended to multi-slice CT. The extended methods, however, had a serious image quality problem due to the following three reasons: (1) excessively close slice positions of the complementary and direct data, resulting in a larger sampling interval; (2) the existence of several discontinuous changeovers in pairs of data samples for interpolation; and (3) the existence of cone angles. Therefore we have proposed a new algorithm to overcome the problem. It consists of the following three parts: (1) optimized sampling scan; (2) filter interpolation; and (3) fan-beam reconstruction. Optimized sampling scan refers to a special type of multi-slice helical scan developed to shift the slice position of complementary data and to acquire data with a much smaller sampling interval in the z direction. Filter interpolation refers to a filtering process performed in the z direction using several data. The normal fan-beam reconstruction technique is used. The section sensitivity profile (SSP) and image quality for four-array multi-slice CT were investigated by computer simulations. Combinations of three types of optimized sampling scan and various filter widths were used. The algorithm enables us to achieve acceptable image quality and spatial resolution at a scanning speed that is about three times faster than that for single-slice CT. The noise characteristics show that the proposed algorithm efficiently utilizes the data collected with optimized sampling scan. The new algorithm allows suitable combinations of scan and filter parameters to be selected to meet the purpose of each examination.