-
Notifications
You must be signed in to change notification settings - Fork 6
/
iotaroadmap.txt
144 lines (80 loc) · 4.35 KB
/
iotaroadmap.txt
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
Project Group Iota
##### Project General Information
Data:
(1) We have 4 groups of participants: schizophrenia patients, patients'
siblings, normal people, and normal people's siblings.
(2) We have 4 tasks: R, 1B, 2B, 3B.
(3) We have 4 ROIs: FP, CO, CER, DMN.
Original Paper Objectives:
1. Within-network connectivity:
(a) For each participant, find connectivity within each ROI on different tasks.
2.Between-network connectivity:connectivity:
(a) For each participant, find connectivity between each ROI on different tasks.
Our Project Objectives: 1.Simplified assumptions:
(a) Since we do not have the knowledge about partition the brain into ROIs, we
will simply focus on the entire brain (We can ask Matthew for help for this
part).
(b) Since we do not have the knowledge of comparing different subjects who have
different brain shapes, we will simply visually compare brain images and
approximate.
2. We will examine two subjects, one subject with schizophrenia and one healthy
subject.
##### Tentative Roadmap
General to do list:
- Get convolved time course and save as convolved.txt.
- Pick one subjects and go full length analysis. Or pick one Schizophrenic
participant and one healthy participant.
- Make design matrix using convolved.txt and run regression on BOLD signals to
get the estimated betas. (Note: the betas are the scaling of the BOLD signals
in response to the stimulus.) We can do t-tests on the betas (However, keep in
mind the assumptions we use.) and see which of the betas are significant.
- Find correlations between average voxels in each ROI (Before we figure out
ROIs, we can try to find correlation between each voxels as from lecture 12,
Correlation per voxel, in 2D). Measure mean signal. Make a histogram of the
values of correlations (values from 0 to 1), and then
determine a cutoff threshold to determine which of the correlations are
significant.
- Use Graphical Lasso to visualize covariance and precision of data.
##### To_Do_chunk_1 (Weeks 11/2 - 11/13):
1. Make convolved.txt using notes from lecture Day 13 (reference:
http://www.jarrodmillman.com/rcsds/lectures/convolution_background.html).
2. Make design matrix using convolved.txt and run regression on 2D BOLD signals.
Question: what are we regressing on? -- We are regressing the BOLD signals with
the convolved time course. If the betas are large, then the BOLD signal is
strong in response to the stimulus. (reference: see introduction of this about
convolution: http://practical-neuroimaging.github.io/on_convolution.html)
3. Make progress slides for 11/12 progress presentation.
4. Go to office hours on Friday morning 11/6 for project questions.
5. Project-related questions will be asked on Gitter.
Task Assignment:
-Zeyu is responsible for making weekly plans, and assign tasks. Will be helping
with other teammates with code writing or reviewing as well as slides making.
-Jie and Qingyuan are responsible for writing code for convolved time course. If
there are questions, ask Zeyu or Yunchuan. -Yunchuan is responsible for making
the progress presentation report.
##### To_Do_chunk_2 (Weeks 11/16 - 11/27):
1. Substitute raw data with filtered data newly uploaded.
2. Review and revise and comment code for reproducibility and clarity.
3. Use lectures on to get our high resolution hemodynamic for each of the
condition files to be used for design matrix in linear modeling.
4. Improve t-map and p-map and completely finish linear modeling.
5. Do t-test on different task levels.
-Zeyu is responsible for reviewing and revising code, and writing draft paper.
-Jie does the time series analysis and finish code for p-map.
-Qingyuan improves the p-map and write tests for code.
-Everyone is responsible for any questions that anyone have posted through
issues and pull requests.
##### To_Do_chunk_3 (Weeks 11/30 - 12/14):
1. Explore get_affine for ROIs and ANOVA and f-test.
2. Explore Lasso and PCA.
3. Finish time series analysis.
4. Prepare for final presentation on 12/3
##### To_Do_chunk_4 (After talking with Jarrod and JB)
1. Block Design:
a. Beta plot
b. No significant p-value detected, try to choose top 50s or 100s p-value
and regenerate the plot.
2. Time series:
a. DCF method. Gaussian design.
b. Ask Matthew for his notes on Time-series analysis.
c. Threshold p-value: look at residuals of edge points.