题目：Non-linear partial volume effect in CT reconstruction

报告人：刘鑫 博士

地点：南区物理光电楼1212会议室

时间：2019年3月8日（星期五）下午2:30

摘要：In CT imaging, detector bins and X-ray source focal spot have finite sizes and so do volume elements within the imaged subject sampled by the detector bins and focal spot. Within each of the volume elements sampled, the X-ray linear attenuation coefficient can be non-uniform, thus resulting in the so-called non-linear partial volume (NLPV) effect that can be described by a discrete non-linear X-ray transform. Unlike the standard X-ray transform that is a linear function of the attenuation coefficient map, i.e., the image of interest, the discrete X-ray transform with the NLPV effect becomes a non-linear function of the attenuation coefficient map. When algorithms developed for the standard, linear X-ray transform are applied to the discrete non-linear X-ray transform, they can engender the NLPV artifacts in images reconstructed. The NLPV artifacts may diminish the utility of CT images by plaguing the low-contrast anatomies in certainly clinical applications . While effort has been devoted to develop algorithms 35 to correct empirically for the NLPV artifacts , the correction for the NLPV artifact through directly inverting the discrete non-linear X-ray transform remains of theoretical and practical interest. Motivated by the fact that there has been a limited investigation of algorithms for accurate inversion of the discrete non-linear X-ray transform with the NLPV effect, we investigate and develop algorithms in the work for its accurate inversion.