Resolution enhancement of X-ray CT by spatial and temporal MLEM deconvolution correction

In multi-slice CT, there is an ongoing trend to use smaller slice thickness and faster rotation time, leading to smaller detector pixel size and shorter sampling time. In general, cross talk between adjacent detector pixels and scintillation short-term afterglow can cause resolution degradation and artifacts. Crosstalk effect becomes significant while using detector arrays with small pixels. Short-term afterglow becomes significant with high rotation speed, especially while using focal spot modulation between successive frames. In these cases of small pixels and high rotation speed, the signals of detector pixels are relatively noisy. Thus, attempt to correct cross talk and afterglow by using techniques based on inverse filters or on subtraction of constant fraction of the adjacent pixel signals, may give poor results. We developed cross talk and afterglow deconvolution correction algorithm, applied on the scanner raw data. The algorithm is based on the Maximum Likelihood Expectation Maximization method (MLEM) and is similar to the Lucy-Richardson deconvolution in image restoration. We show that the results significantly improve the image resolution of fine bone structures.