Cerebral magnetic resonance image synthesis.

The authors previously described magnetic resonance (MR) image synthesis, a process that enables the investigator to manipulate imaging parameters retrospectively and generate or "synthesize" the image that corresponds to various arbitrary scanning factors. They demonstrate the validity and utility of synthetic spin-echo images in cerebral imaging. As a test of their method, spin-echo images are synthesized for echo times identical to those of the original acquired images as well as for alternate values. Subjectively, the quality of synthetic and acquired images is comparable. It is shown quantitatively for several tissue types that the reconstructed synthetic signal matches the acquired signal within the uncertainty of the acquired images. Observed and measured noise levels in the acquired and synthetic images are comparable. Because of a signal-averaging effect, the synthetic images can have a higher signal-to-noise ratio than the source images, thereby providing improved boundary definition. Applications of MR image synthesis are discussed with respect to potential reduction in scanning time. The advantages of image synthesis versus analysis of computed images are discussed.