-
Notifications
You must be signed in to change notification settings - Fork 1
/
Maddah_Notes.lyx
194 lines (160 loc) · 5.17 KB
/
Maddah_Notes.lyx
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
#LyX 1.6.2 created this file. For more info see http://www.lyx.org/
\lyxformat 345
\begin_document
\begin_header
\textclass article
\use_default_options true
\language english
\inputencoding auto
\font_roman default
\font_sans default
\font_typewriter default
\font_default_family default
\font_sc false
\font_osf false
\font_sf_scale 100
\font_tt_scale 100
\graphics default
\paperfontsize default
\use_hyperref false
\papersize default
\use_geometry false
\use_amsmath 1
\use_esint 1
\cite_engine basic
\use_bibtopic false
\paperorientation portrait
\secnumdepth 3
\tocdepth 3
\paragraph_separation indent
\defskip medskip
\quotes_language english
\papercolumns 1
\papersides 1
\paperpagestyle default
\tracking_changes false
\output_changes false
\author ""
\author ""
\end_header
\begin_body
\begin_layout Title
Notes from Mahnaz Maddah's Thesis
\end_layout
\begin_layout Section
Quantitative Analysis of Diffusion MRI
\end_layout
\begin_layout Enumerate
Region of interest (ROI)-based methods
\begin_inset Newline newline
\end_inset
Non-time efficient (require user interaction)
\begin_inset Newline newline
\end_inset
ROI size, shape, number, and location not only affect the measured quantities,
but also influence the significance of the group analysis.(cite Kanaan 2006)
\end_layout
\begin_layout Enumerate
Voxel-based methods
\begin_inset Newline newline
\end_inset
Datasets compared voxel-by-voxel.
Simple but alignment very critical.
\begin_inset Newline newline
\end_inset
Registering scalar fields like FA does not employ all of the information
in the data and will not provide the most accurate analysis.
\begin_inset Newline newline
\end_inset
Advantages: 1.user-independent, 2.
whole-brain analyses.
\begin_inset Newline newline
\end_inset
Disadvantages: smoothing for statistical validity reduces resolution, significan
t group differences do not necessarily lie within an anatomical tract.
Reference to an anatomical atlas is hindered by the low resolution of the
obtained difference map by the limited resolution of the atlas itself (cite
Kannaan 2006).
\end_layout
\begin_layout Enumerate
Tract-oriented methods.
\begin_inset Newline newline
\end_inset
Tract-oriented methods offers advantages over ROI-based since it reveals
local variations of the fiber integrity which are lost when the quatitative
parameters are averaged over the entire fiber tract in ROI-based methods.
\begin_inset Newline newline
\end_inset
Clustering could group trajectories to single
\begin_inset Quotes eld
\end_inset
anatomical
\begin_inset Quotes erd
\end_inset
fiber tracts.
\begin_inset Newline newline
\end_inset
The principal benefit is that observed differences are due to differences
in the properties of specific tracts rather than differences in the overall
anatomy/shape of the individual brains.
\end_layout
\begin_layout Section
Problems with Tract-oriented methods
\end_layout
\begin_layout Enumerate
Regardless of the nature of tractography methods, whether deterministic
or probabilistic, the output trajectories often have discontinuities due
to the presence of noise and image imperfections.
\end_layout
\begin_layout Enumerate
Even with an ideal tractography algorithm and incredibly efficient preprocessing
of the data to remove noise artifacts, lesions and other brain abnormalities
may cause discontinuities in the trajectories.
\end_layout
\begin_layout Enumerate
Defining similarity between trajectories is not trivial.
The similarity between three-dimensional curves (trajectories) is not uniquely
defined and depends on the application.
A good strategy is to require the similarity measure to use both spatial
and shape information from the whole trajectory.
The point correspondence could be facilitated by the use of landmarks that
contain most of the information for the shape and location of trajectory.
Such landmarks can be specified using local extrema
\begin_inset CommandInset citation
LatexCommand cite
key "deriche1990dcm"
\end_inset
or minimum description length
\begin_inset CommandInset citation
LatexCommand cite
key "DavisTMI02"
\end_inset
.
\begin_inset Newline newline
\end_inset
\end_layout
\begin_layout Enumerate
An unsupervised algorithm is not guaranteed to produce the clusters of interest
for a given application (fear of over or under-clustering).
Mahnaz Maddah in her thesis uses a supervised clustering algorithm in DTI
data that benefits from anatomical information.
By such a tract-based analysis, she was able to identify a significant
drop in FA in the vicinity of the lesion (cite Mahnaz's thesis figure p.31),
without knowing where the lesion is located a priori (cite all Mahnaz's
papers and thesis).
The point correspondence between the trajectories is built using a distance
map and a Voronoi diagram on the same space.
The anatomical prior is given by an anatomical atlas or by a Dirichlet
distribution that controls the impact of the atlas.
By using an atlas, the correspondence between clusters is different subjects
is automatically known.
\end_layout
\begin_layout Standard
\begin_inset CommandInset bibtex
LatexCommand bibtex
bibfiles "/home/eg01/Documents/diffusion_review/devel/diffusion"
options "plain"
\end_inset
\end_layout
\end_body
\end_document