On Tue, Feb 03, 2009 at 02:24:27PM -0600, Junqian Gordon Xu wrote: > Ping, > > If you ever need something more powerful/versatile than shell scripts > and FSLUTILS (Miscellaneous FSL Image Utilities), esp. things you need > to calculate on pixel level, here is a collection of tools (unified by > python) I use, which should achieve almost anything you want to achieve > in matlab (I understand I should refrain myself from commenting on > matlab, but ... anyway): > > [1] iPython (interactive env) > [2] PyNIfTI (file I/O) > [3] numpy (number crunching) > [4] rpy2 (interface with R for statistics) > [5] matplotlib (2D plotting) > [6] python-gnuplot (plotting) ... [X] SciPy (swiss army knife) > If your linux distro is debian, you are lucky because everything is > already packaged for you in the repo. I fully support this view. However, the are far more packages available that make the Python-based processing chain very attractive. It actually comprises the whole scientific process -- ranging from stimulus generation over data analysis to visualization. Please let me shamelessly advertise: http://www.frontiersin.org/neuroinformatics/paper/10.3389/neuro.11/003.2009/ the included 'Table 1' contains a list of interesting Python packages that extends the above list even further. Needless to mention that there is no license cost involved -- which get very interesting if you want to run stuff on a 100-node cluster... Michael -- GPG key: 1024D/3144BE0F Michael Hanke http://apsy.gse.uni-magdeburg.de/hanke ICQ: 48230050