Version 2.2.0 released

Version 2.2.0 has been released. Go to the Download page for the most recent version.

The documentation page links to a video podcast of how to use AnalyzeNNLS (http://podcasts.imaginginformatics.ca/20081207AnalyzeNNLSTutorial.m4v) . This video was made pre version 2.1.0. Referring to that video, the following changes have been made to AnalyzeNNLS based on feedback I have received.

Version 2.1.0 - 1) Generalized Cross Validation (GCV) approach to regularization. In this version I removed the user-controlled smoothing parameter range used for regularizing the fit, which in turn creates smooth T2 distributions instead of discrete spikes. To my knowledge, GCV was first used on multiecho data by: Dula et al
J Magn Reson, Vol. 196, No. 2. (February 2009), pp. 149-156.
Version 2.1.0 - 2) The library now includes a script that converts Bruker and dicom multiecho files to MEID. MEID is AnalyzeNNLS's native format, standing for MultiEcho Image Data.
Version 2.1.0 - 3) Ran on Vista with Matlab 2008.

Version 2.2.0 - Now that the user is no longer in control of the smoothing parameter, the width of peaks have actual meaning. The geometric mean T2 width ratio (gmT2WR) is now calculated by assuming the peak of interest is normally distributed on a log scale. The T2 times for the FWHM are determined as T2long and T2short. gmT2WR = T2long/T2short.