Biomedical Engineering Reference
In-Depth Information
Anovel technique is used to compute the 3Ddistancemap as a shape descriptor
of the WM. The distribution of this distance map is then used as a statistical feature
that allows the discrimination between the two groups. The preliminary results
were very promising and motivated us to push the shape analysis even further and
for more specific brain structures, particularly for the CC.
We also use our new nonrigid registration technique, which is based on scale
space and curve evolution theories [69], to devise a new classification approach by
analyzing the deformation fields generated from registering CCs onto each others.
The remainder of this chapter is organized as follows. Section 2 describes
the data sets and how they were collected, as well as the MRI protocol. The
image processing approaches used in the analysis and the results are introduced in
Section 3. Finally, we present our results and conclude with a discussion section.
2. SUBJECTS AND IMAGE ACQUISITION
Two types of brain data are used in this study: postmortem proton density
(PD)-weighted MRI data, and in-vivo T1-weighted MRI metadata. The latter will
be referred to as the savant data in this chapter.
Data and Specimens
Postmortem brains were obtained from the Autism Tissue Program (ATP).
Diagnosis for each patient was established by the Autism Diagnostic Interview-
Revised (ADIR). Postmortem brains from 14 autistic patients (mean interval be-
tween death and autopsy: 25.8 hours) and from 12 controls (mean interval between
death and autopsy: 20.4 hours) were used in this study.
Data Collection
Aquick description of the postmortemMRI data used in this study is presented
in this section. For more details, one can refer to [48]. In order to optimize white-
gray matter substance contrast in formalin-fixed brains, a PD-weighted imaging
sequence was used. The method employed a 1.5-Tesla GE MRI system to scan
brains that were placed within a special device that avoids dehydration during
the scanning procedure. The scan time lasted an average of 50 minutes and 29
seconds. A Fast Spin Echo (FSE) technique with a long Repetition time (TR) of
6700 ms and a short Echo time (TE) of 8 . 23 ms and an Echo train of 4 render
coronal images with excellent gray-white matter contrast. The technique allows
the full coverage of brains with 114 slices, 1 . 6 mm thick, with no inter-slice gap. A
high-resolution K-space dataset consisting of 256 points in the phase and Zero-fill
interpolation (ZIP) 512 points in the frequency both encode direction. An in-plane
resolution of 625 microns
312.5 microns, a field of view of 16 × 16 cm 2 , and a
number of repeated excitations (NEX) of 7 with a bandwidth of 62.5 were used.
×
Search WWH ::




Custom Search