Comparison between EPI and HASTE for ultra-fast MR imaging of the human brain.

Our purpose was to evaluate and compare the performance of ultra-fast single-shot T2-weighted sequences: echo-planar imaging (EPI) versus half-Fourier single-shot turbo spin-echo (HASTE) and to assess the usefulness of their combined reading. Comparative experiments on a phantom as well as a prospective clinical study in 47 patients were done. Axial images acquired with the following methods were compared: (a) HASTE; (b) segmented HASTE (s-HASTE); (c) single-shot spin-echo EPI (SE-EPI); and (d) gradient-echo EPI (GREEPI). Quantitative and qualitative criteria as well as lesion detectability were analyzed against the "gold standard" fast spin-echo (FSE) sequence. For contrast and contrast-to-noise ratio (CNR) between gray and white matter, GRE-EPI was best. The visibility of small markedly hyperintense lesion was best with HASTE and s-HASTE in the clinical study. Small hyperintense lesions were detected equally well with all four sequences, although all performed significantly worse than FSE. The two HASTE variants were better than the EPIs for the extraaxial lesions. The combination of the GRE-EPI and s-HASTE was judged best, and sometimes superior to the FSE image. HASTE or EPI alone cannot substitute for FSE in the screening evaluation of the brain. However, together, EPI and HASTE could provide comparable diagnostic information to that of FSE because their combination compensates for their individual limitations.