Spatially-adaptive temporal smoothing for reconstruction of dynamic and gated image sequences

In this paper we propose a method for spatio-temporal reconstruction of dynamic or gated image sequences. In a method we proposed previously, temporal smoothing in a Karhunen-Loeve (KL) transform domain was used prior to reconstruction to reduce the effect of noise. Unlike the Bayesian priors that are usually used in image reconstruction, temporal KL smoothing is a data-driven approach that takes advantage of the fact that the desired part of the data is characterized by strong inter-frame correlations, whereas the noise is uncorrelated. In this paper we improve on one of our group's previous techniques by making the temporal smoothing adapt spatially to local characteristics in the projection data. This improves the noise performance of the temporal smoothing, while significantly lessening the possibility of signal distortion. In the proposed method, spatial regions of the projection data sequence having similar time characteristics are identified by an unsupervised k-means clustering algorithm. A different KL transformation is designed for each spatial region in projection space, adapting the smoothing to the local temporal behavior. Finally, images are reconstructed from the smoothed projections by using the expectation-maximization (EM) algorithm. Experimental computer simulation results are shown that demonstrate potential improvements in image quality obtained by this technique for dynamic and gated imaging applications in brain, lesion, and cardiac imaging.