[BIC-announce] FW: Technical implications of the TIM upgrade at the UNF

[Jennifer Chew, Ms.]

FOR YOUR INFORMATION.  JENNIFER 

________________________________

From: Francine Bélanger [mailto:francine.belanger at criugm.rtss.qc.ca] 
Sent: Tuesday, April 10, 2007 7:02 PM
To: Felicia Callocchia, Ms.; Jennifer Chew, Ms.; maria.sanchez at umontreal.ca
Subject: Technical implications of the TIM upgrade at the UNF


Bonjour,
C'est la suite du premier message... pour information aux membres du CERNEC
Merci beaucoup

HI,
This is following the first message... to inform members of your groups
Thank you very much

-------- Message original -------- 
Sujet: 	[UNF Annonces] Technical implications of the TIM upgrade at the UNF	
Date: 	Tue, 10 Apr 2007 17:47:56 -0400	
De: 	Rick Hoge <r.hoge at umontreal.ca> <mailto:r.hoge at umontreal.ca> 	
Pour: 	annonces at unfweb.criugm.qc.ca	



Technical implications of the TIM upgrade at the UNF

The upgrade of our Siemens Trio scanner to the TIM platform has important implications for users in two contexts.  First, the new hardware is better in many respects, allowing image acquisition with higher sensitivity and using techniques not available on the current system.  This is a good thing, and the reason we have worked so hard to get the upgrade.  The second important issue is the concern that subtle changes in image characteristics could increase variance or introduce bias in longitudinal or cross-sectional studies in which sessions are performed both before and after the upgrade.

If you don't have any studies that will span the upgrade (and the ability to do retrospective analyses against old data is not critical), the situation is quite simple:  when planning studies after the upgrade you will have the option of choosing new acquisition methods that will offer higher sensitivity and spatial resolution, or reduced scan time.    If you *are* in the midst of a longitudinal study, or if the ability to perform retrospective analyses against old data is important, then you will need a way to minimize any scanner-specific characteristics that affect your image data.

If it is extremely inconvenient for you to avoid acquiring data for a single study in a period that spans the upgrade, there are a number of measures you can take to minimize the impact of scanner version on your data:

1)  The first and most obvious step is to use same pulse sequence and parameter settings in all data that are acquired, both before and after the upgrade.  This will mean forgoing improved sequences available after the upgrade, but this is a fact of life in longitudinal (and multi-centre) studies.  

2)  If you can replicate the original sequence timing characteristics after the upgrade, then you are left with the effects of scanner hardware that has changed in the upgrade.  The most important hardware changes will be the improvements in gradient linearity and RF homogeneity after the upgrade.  To minimize the impact of the upgrade, you have to apply corrections for gradient non-linearity and RF inhomogeneity on all data (from before and after the upgrade) that will be analyzed.  These corrections will be based on calibration measurements we will perform on the UNF scanner before and after the upgrade.  Note that if you are not combining data from before and after the upgrade, these corrections are less important.

3)   You should try to adapt your schedule for subject recruitment and scheduling to minimize potential confounding effects.  For example it would clearly better to avoid scanning all the patients in a study before the upgrade, and all the controls after.  You should treat scanner version as another experimental variable, and follow factorial design principles to minimize bias.

4)  The final thing we can do is to acquire control data (anatomic and functional) in a small group of subjects immediately before and after the upgrade, to apply statistical tests to estimate the magnitude of the upgrade effect.  Doing this properly requires some work, as characterization of test-retest reliability for comparison against the upgrade effect will require two sessions pre-upgrade and two afterwards.  The minimum number of subjects required for this is probably five or six, and we would include some standard structural scans as well as DTI, fMRI, and possibly spectroscopy.  If you are able to include scanner version in a factorial design for your study, you will be able to estimate the "upgrade effect" in your own data.  Getting input on what kinds of acquisition and paradigm to test will help us prioritize scans in the limited time that will be available for this.

There are a number of recent publications out that examine the impact of scanner upgrades or replacements on imaging data:

Friedman et al.  Reducing interscanner variability of activation in a multicenter fMRI study: controlling for signal-to-fluctuation-noise-ratio (SFNR) differences.  Neuroimage. 2006 Nov 1;33(2):471-81.

Han et al.  Reliability of MRI-derived measurements of human cerebral cortical thickness: the effects of field strength, scanner upgrade and manufacturer.  Neuroimage. 2006 Aug 1;32(1):180-94.

Rick Hoge, Julien Doyon





-- 
Francine Bélanger
Coordonnatrice administrative de l'UNF/RNQ/REPRIC
Centre de recherche, Institut universitaire de gériatrie de Montréal
4565, chemin Queen Mary, Montréal (Québec) H3W 1W5
Tél.:	  (514) 340-2800 p. 4785
Téléc.:   (514) 340-3530
Courriel: francine.belanger at criugm.rtss.qc.ca
Web :	  www.unf-montreal.ca
	  www.rsmnq.ca/repric