-
Notifications
You must be signed in to change notification settings - Fork 8
/
README
158 lines (114 loc) · 5.26 KB
/
README
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
readme
This directory contains various algorithms for image reconstruction and
other inverse problems such as image restoration and image registration.
It also contains code for related applications including MRI pulse design.
--- GETTING STARTED
After installing the toolbox, use matlab's "path" functionality to put
the top level directory in its path (or launch matlab from that directory).
Then run the file 'setup.m' and all the subdirectories will be added to
the path. You may find it convenient to read setup.m and customize it.
I recommend running and examining some of the files in the example/ directory
or any of the many ..._example.m files around, such as
emission/eml_osem_example.m
Many example files prompt you to hit enter before continuing, so you and I
can see the output of each stage before proceeding.
To change this behaviour, execute 'prompt run'
(See utilities/prompt.m for help.)
Some of examples may require the binary program "wt" or "op" that are part
of Aspire. You can also get Aspire for free by following the instructions
on my web pages. There are also a couple of mex files that you may need -
wtfmex and f3dmex for example. I distribute these only in linux/mac formats;
see mex directory.
Part of my motivation for creating these files is to accompany a book on
image reconstruction that I am currently writing. If you have any problems
with these m-files, or any suggestions whatsover, I welcome your input!
---------------------------------------------------------------------------
Jeff Fessler, http://www.eecs.umich.edu/~fessler/
---------------------------------------------------------------------------
--- Subdirectories (in alphabetical order):
align:
image registration tools
blob:
SPECT reconstruction with blob basis functions (not recommended)
contrib:
algorithms contributed by others. these directories are not added to
the path by setup.m so the user must modify the path to use them.
contrib/ppcd:
test routines comparing WLS-CD, WLS-GCD, WLS-PPCD
(these are mostly for internal UM use)
ct:
polyenergetic CT routines (beam hardening, dual energy, etc.)
data:
data for examples
doc:
see the pdf file within for some introductory documentation.
emission:
algorithms for Poisson emission tomography PET/SPECT/ Poisson regression
eml_ emission maximum likelihood
eql_ emission quadratically penalized likelihood
epl_ emission penalized likelihood
example:
example(s) of usage. there are more examples in other directories too.
running any of these examples is a good place to start!
fbp:
filter-backproject reconstruction, including 2D parallel and fan-beam
and 3D Feldkamp (FDK) cone beam reconstruction
freemat:
work in progress, towards making the code run with freemat
general:
some algorithms that work for generic image reconstruction problems
graph:
graphics routines
mex:
MEX (matlab executables), including some C99 source code
mri:
MR image reconstruction
mri-rf:
MR pulse design tools, including Spectral-spatial pulse design
for phase precompensatory slice selection. (more to come)
nufft:
non-uniform FFT toolbox
penalty:
functions associated with regularization
systems:
system matrices and system matrix object classes
If you are interested in edge-preserving image restoration
for a shift-invariant blur model with additive gaussian noise,
then start with systems/Gblur_test.m and example/restore_example.m
For tomography, consider starting with systems/Gomo2_strip.m, which
is used in many of the examples.
transmission:
algorithms for Poisson transmission tomography
tml_ transmission ML
tql_ transmission quadratically penalized likelihood
tpl_ transmission penalized likelihood
utilities:
useful matlab utilities for image reconstruction algorithms.
wls:
algorithms associated with the weighted least squares (WLS)
cost function and penalized versions thereof
pwls_ penalized weighted least squares
qpwls_ quadraticaly penalized weighted least squares
Most algorithms also include a test routine...
--- Additional notes:
Raymod Muzic has matlab routines for reading ECAT files available:
www.nuclear.uhrad.com/comkat
(I have not yet tried them myself.)
There is one known issue with Matlab7. It can store sparse matrices
only as doubles with is fine, but if you do "S * x", where S is a
sparse matrix and x is a vector of class single, Matlab7 gives an
error message rather than politely upgrading x to a double.
The object Gsparse.m provides a work around for this.
Complain to Matlab loudly that they should fix this annoyance...
Windows users:
Some of the subdirectories of the "systems" directory contain "links"
to m-files in other directories. (These are created using "ln -s" in
unix.) These links are also in the .tar file as soft links. But these
links may not be recognized by Windoze, resulting in various error messages.
They work fine on Mac OSX since that fantastic operating system is unix
"under the hood." I recommend that you avoid using Windows.
But if you insist, then you will have to figure out how to fix those links
or copy the appropriate m-files into the appropriate directories.
Another problem is that apparently windoze is case insensitive.
I believe I have purged most of the m-files that had capitalized names now.
Nevertheless, at this point it would be better to just go buy a Mac...