hide
Free keywords:
-
Abstract:
Conventional approaches to quantify whole brain Turn:x-wiley:07403194:media:MRM10283:tex2gif-stack-3 maps use nonlinear regression with intensive computational requirements that therefore likely limit quantitative Turn:x-wiley:07403194:media:MRM10283:tex2gif-stack-4 mapping for real-time applications. To overcome these limitations an alternative method, NumARTurn:x-wiley:07403194:media:MRM10283:tex2gif-stack-5 (NUMerical Algorithm for Real-time Turn:x-wiley:07403194:media:MRM10283:tex2gif-stack-6 mapping) that directly calculates Turn:x-wiley:07403194:media:MRM10283:tex2gif-stack-7 by a linear combination of images obtained at three or more different echo times was developed. NumARTurn:x-wiley:07403194:media:MRM10283:tex2gif-stack-8, linear least-squares, and nonlinear regression techniques were applied to multiecho planar images of the human brain and to simulated data. Although NumARTurn:x-wiley:07403194:media:MRM10283:tex2gif-stack-9 may overestimate Turn:x-wiley:07403194:media:MRM10283:tex2gif-stack-10, it yields comparable values to regression techniques in cortical and subcortical areas, with only moderate deviations for echo spacings between 18 and 40 ms. NumARTurn:x-wiley:07403194:media:MRM10283:tex2gif-stack-11, like linear regression, requires 2% of the computational time needed for nonlinear regression and compares favorably with linear regression due to its higher precision. The use of NumARTurn:x-wiley:07403194:media:MRM10283:tex2gif-stack-12 for continuous on-line Turn:x-wiley:07403194:media:MRM10283:tex2gif-stack-13 mapping in real time fMRI studies is shown.
