Email discussion lists for the UK Education and Research communities

FSL@JISCMAIL.AC.UK

View:

 Message: [ First | Previous | Next | Last ] By Topic: [ First | Previous | Next | Last ] By Author: [ First | Previous | Next | Last ] Font: Proportional Font

Options

Subject:

Poisson regression bias

From:

Date:

Sun, 6 Jan 2019 03:07:01 +0000

Content-Type:

text/plain

Parts/Attachments:

 text/plain (23 lines)
 ```Hi everyone, So I was hoping someone could help me make sense of this problem. I came across this paper that discusses how the FSL probabilistic DTT may yield bias tractography relating to the physical distances between brain regions. This paper went a step further to figure out at what distance the bias becomes negligible. It was found that brain regions separated more than 17.4mm the bias was negligible. But for distances shorter than 17.4 mm, a bias did exist in the structural connectivity strength (distance related false positives). They then go on to discuss: how they employ a Poisson regression-based statistical adjustment that yields measures of SC adjusted for the physical distances between region locations. Specifically, they apply a model that assumes that the number of DTT streams Sij connecting regions i and j follows a Poisson distribution with the mean μ(Sij|gij) dependent on the physical distance gij between these regions, i.e. Sij|gij~Poi(μ(Sij|gij)). They estimate and subsequently adjust for the association between the physical distances and the DTT counts using the effect α1 in the log-linear model: log(μ(Sij|gij))=α0+α1gij So, it is assumed that each structural connectivity strength is adjusted for physical (geometric) distance to reduce the potential impact of false structural connections Now my question is: How can I adjust my measures of SC for physical distance between regions? Does the poisson regression (log-linear model) naturally take care of this bias? Or do I include the the 17.4 mm somehow in my model in order to correct for it? Any help would be greatly appreciated. Thanks. ######################################################################## To unsubscribe from the FSL list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=FSL&A=1```