Sensor fusion in image reconstruction

A Bayesian method for ECT (emission computed tomography) image reconstruction has been developed to incorporate a priori information derived from the spatially correlated CT and MR (magnetic resonance) images. These anatomic maps, showing boundaries between regions that exhibit distinctly different characteristics, can be incorporated into the Bayesian method, thus improving the spatial resolution and noise properties. The correlated structural information can also be used as templates for deriving correcting factors for the effect of photon attenuation, thus improving the quantitative accuracy and noise properties. This novel image reconstruction method using the concept of sensor fusion has been validated by computer simulation studies. Results from these studies indicate that significant improvements in image quality can be achieved.>