Hi
> I have the following questions regarding selection of significant
> components from TICA output. The common approach is to select
> components with "standard subject scores not driven by outliers and
> mean subject score significantly different from zero". My question is:
>
> 1) How are the subject scores given in melodic report calculated?
These are the loadings after calculating a rank-1 approximation to the
matrix of all subject dependent time courses for a given component
(see the technical report http://www.fmrib.ox.ac.uk/analysis/techrep/tr04cb1/tr04cb1/index.html
)
> What does it mean?
These encode the effect size per subject for the associated spatio-
temporal process shown in the spatial map and the time course.
> Are they already standarized?
Nope, otherwise you'd loose important information such as relative
average strength between compoents.
> If they are not standarized, do we need to standarize them?
>
Depends what you'd want to do with them. By default I'd argue that no,
no standardisation necessary.
>
> 2) How can I detect if standarized subject score are driven by
> outliers?
>
The boxplot is a useful way to look at this - in fact, components are
ordered according to the median effect size (centre horizontal line in
the boxplot). Outlier components have almost all subect mode values
close to zero with a few exceptions. This results in a boxplot which
is a small box around 0 and a few red dots (strong outliers). Non-
outlier components show proper boxes which are not centered on zero.
hth
Christian
_______________________________________________
Christian F. Beckmann, DPhil
Senior Lecturer, Clinical Neuroscience Department
Division of Neuroscience and Mental Health
Imperial College London, Hammersmith Campus
Rm 419, Burlington Danes Bldg, Du Cane Road, London W12 0NN, UK
Tel.: +44 (0)20 7594 6685 --- Fax: +44 (0)20 7594 6548
Email: [log in to unmask]
http://www.imperial.ac.uk/medicine/people/c.beckmann/
Senior Research Fellow, FMRIB Centre
University of Oxford
JR Hospital - Oxford OX3 9DU
Tel.: +44 (0)1865 222551 --- Fax: +44 (0)1865 222717
Email: [log in to unmask]
http://www.fmrib.ox.ac.uk/~beckmann
|