In <2961288583.0.p00907 at psilink.com>
"Richard E. Cytowic MD" <p00907 at psilink.com> writes:
> There is rather a lot of inter-subject variability not only in
>gross brain anatomy, but in functional mapping (electrical). On the
>face of it, you propose to measure what cannot be measured.
Thank you for your suggestion. I expected a response like yours.
I searched for the papers you mentioned, but they were so old that
our library(it's a large library, though..) didn't have. But I found
another article by Ojemann "Cortical language localization in left, dominant
hemisphere", 1989, Journal of Neurosurgery 71:316-326, which notes
the variability of language localization determined by intraoperative
stimulation. According to that article, their distributions were mosaic-
patterned, varied regardless of gyration.
But actually, intersubject averaging is so widely accepted in PET
activation studies, that without it, no new activated foci can be
detected. As a matter of course, as you imagine, resolution is not
good(as large as 8-10mm). What I actually am about to do is non-linear
transformation. It means that transforming the brain PET image
non-linearlly(or distorting) before averaging so that each corresponding
point is overlapped more precisely. Currently, only linear
transformation is done. We expect by non-linear transformation, the
ability of detection (or signal/noise ratio) is increased compared with
linear transformation, and want the theoretical background.
Tetsuji Rai MD trai at twics.co.jp, 74610.1537 at compuserve.com
Department of Radiology, Hospital attached to University of Tokyo
Home: 5-12-21-301, Toyotamakita, Nerimaku, Tokyo 176, Japan