Newsgroups: sci.techniques.mag-resonance,sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!gatech!sdd.hp.com!hplabs!hplntx!peter
From: peter@hpl.hp.com (Peter Webb)
Subject: Re: Wrap-around effect in MR images
Sender: news@hpl.hp.com (HPLabs Usenet Login)
Message-ID: <D8qE7A.AyA@hpl.hp.com>
Date: Wed, 17 May 1995 16:37:58 GMT
References: <3pc2po$l74@news.ccit.arizona.edu>
Nntp-Posting-Host: oz.hpl.hp.com
Organization: Hewlett-Packard Laboratories, Palo Alto, CA
X-Newsreader: TIN [version 1.2 PL2]
Followup-To: sci.techniques.mag-resonance,sci.image.processing
Lines: 42
Xref: glinda.oz.cs.cmu.edu sci.techniques.mag-resonance:946 sci.image.processing:14678

Te-Shen (Dickson) Liang (liang@mipl.ece.arizona.edu) wrote:
: Hi, MR experts,

:      I am working on the image analysis on the MR images. The image
:   set I have is a 124 T1-weighted slices axial head images. Because the FOV
:   doesn't extand to the whole head being imaged, what called 'wrap-around
:   effect' is very obvious in the upper and lower images.  

:      What are the reasons that cause this wrap around effects ? How to 
:   eliminate this phenomenon ? Is there any paper or book that talks 
:   about this effect ?

The wrap-around effect is aliasing, pure and simple.

In more detail, consider a spin at the edge of the FOV, and one at 3
times that distance from isocenter.  At the edge of the FOV, the first
phase encode causes 180 degrees of phase to accumulate, the second
causes 360, the third 540.  At 3 times that distance, you get 540, 1080,
1620 degrees, which, mod 360 is 180, 360, 540.  So the two points are
indistinguishable, and will show up at the same point in the image.

In fact, any spin past the edge of the FOV will experience the same
phase history as some point within it, and will thus alias onto that
part of the FOV.

The only exception is in the readout direction, where, by oversampling,
the FOV can effectively be made very large.  The extra FOV is thrown
away by digital filtering or by cropping the output of the
reconstruction.  This technique cannot be applied in the phase encode
directions, because oversampling this way requires more excitations, and
thus takes longer (another exception -- multiple average scans are
usually performed as one average with extended FOV.  The scanner
prescription software hides this from the user).

Another method sometimes used to reduce, but not completely eliminate,
wrap around is to place saturation bands at the edge of the FOV.

Probably a good place to read about wrap around and how your particular
scanner deals with it is the application guide shipped by the
manufacturer.  They usually have good tips on what to do.

Peter
