Neuroscience 2003 Abstract
| Presentation Number: | 862.19 |
|---|---|
| Abstract Title: | MRI of the perinatal human brain: automated segmentation and volumetry. |
| Authors: |
Nishida, M.*1
; Makris, N.1
; Kennedy, D. N.1
; Fischl, B.2
; Caviness, V. S.1
; Grant, E.2
1Neurol., Mass. Gen. Hosp., Charlestown, MA 2MA, 149 13th Street Room 6015, 02129, |
| Primary Theme and Topics |
Techniques in Neuroscience - Staining, tracing and imaging techniques |
| Secondary Theme and Topics | Development<br />- Axonal and Dendritic Development |
| Session: |
862. Staining, Tracing, & Imaging Techniques: Brain Imaging Poster |
| Presentation Time: | Wednesday, November 12, 2003 10:00 AM-11:00 AM |
| Location: | Morial Convention Center - Hall F-I, Board # VV39 |
| Keywords: | brain development, infant, automated segmentation |
In the perinatal period, human brain growth is most rapid and the parameters of growth are greatly sensitive to deleterious influences upon brain development. Standard MRI-based morphometric techniques are of limited application at this age because the perinatal brain is minimally myelinated and lack the marked differences in tissue contrast upon which standard approaches depend. The goal of the present study is to develop an accurate, reliable and reproducible method for segmentation and volumetric analysis of the minimally myelinated brains in the range of 34 to 42 weeks gestational age. The emphasis is placed upon the forebrain where highly reliable criteria allow segmentation of the external contour, the separate boundaries of the hippocampus and amygdala and en bloc basal nuclear structures including caudate, diencephalon, striatum and pallidum. It is intended that a series of perinatal brains segmented in terms of these boundaries will serve as a training set for the automated segmentation algorithm, Free Surfer. However, training data sets must be reliable. An initial reliability exercise has involved two separate analyses of a single normative 34 week premature infant. The average pixel overlap for corresponding structures was 84% (median 86.4). (For reference values > 80% are considered extremely reliable.) The slope of the regression correlating voxel overlap for the same brain was 1.03 (r^2=0.99) while mean volume differences between segmented structures was not significantly different from zero (Wilcoxon Signed Ranks, W+ = 57, W- = 63, N = 15, p <= 0.8904). These findings suggest that we have developed a reliable method for manually segmenting perinatal brains and that manually segmented perinatal brains are suitable as training sets for Free Surfer.
Supported by K23NS42758-01, NS34189, Fairway Trust, Giovanni Armenise Harvard Found. for Advanced Scientific Research
Sample Citation:
[Authors]. [Abstract Title]. Program No. XXX.XX. 2003 Neuroscience Meeting Planner. New Orleans, LA: Society for Neuroscience, 2003. Online.
Copyright © 2003-2025 Society for Neuroscience; all rights reserved. Permission to republish any abstract or part of any abstract in any form must be obtained in writing by SfN office prior to publication.
