R4-C.1: Advanced Multispectral CT Algorithms

Dual-energy X-ray computed tomography (DECT) is a powerful tool for non-destructive evaluation for security. In security applications where DECT is used for scanning checked luggage, many different materials may be scanned in various degrees of clutter and metal objects are common. Conventionally, fi ltered back-projection (FBP) is used for image formation and there is no unifi ed treatment of metal artifacts. In the security context, image noise and metal artifacts can be severe and lead to less reliable image estimates and object splitting, lowering the accuracy of material identifi cation. In this project, we develop a structure-preserving dual-energy (SPDE) CT reconstruction method for security, which provides enhanced estimates of material-dependent basis coeffi cient images. The proposed approach aims at mitigating metal artifacts and providing precise object localization. The basis coeffi cient images are generated jointly as the solution of a single model-based optimization problem. An auxiliary variable corresponding to the mutual boundary fi eld is estimated as well and applied to the basis coeffi cient images to improve object localization and smoothing inside the objects. In addition, metal aware data weighting is included to reduce streaks. By using such a unifi ed approach, signifi cant reduction in noise and metal artifacts can be achieved in the image formation process. These improved images, in turn, can yield lower false alarm rates and improved material and object identifi cation.