High Quality CT Image Reconstruction from Limited Projections Obtained from MiSpinner

Abstract

Computed tomography (CT) technology is widely used in the medical and industrial fields. However, a traditional CT machine costs too much for many researchers to afford it and prevents them from doing related research. To solve this problem, this paper explores the possibility of CT image reconstruction based on data from an IVIS Lumina XR machine providing X-ray projection images. Compared with traditional CT machine, it is much cheaper and widely used in animal research field. With the addition of the MiSpinner hardware that successively rotates an object through a sequence of angles, data that approximates that of a CT machine can be acquired. In this research, we begin by preprocessing the data so that it conforms to the geometry of traditional CT. Then we implement and explore various CT reconstruction algorithms to assess performance. Among the algorithms are filtered backprojection (FBP), algebraic reconstruction technique (ART), and total variation methods. The final result is satisfactory. Detail information is strengthened and artifacts caused by limited projections are removed. In addition, 3D rendering is also performed providing improvement in details of the object. This work shows the potential for using the MiSpinner hardware in the IVIS Lumina XR machine for CT research and provides an alternative choice to those people who want to do CT research economically.