Fast Flash Cards

Front 1

Preprocessing

Back 1

Before running FAST an image of a head should first be brain-extracted, using BET. The
resulting brain-only image can then be fed into FAST.

Front 2

Number of input channels

Back 2

if there is only 1 input image (ie you are not carrying out multi-channel segmentation) then leave this at 1 - otherwise set appropriately.

Front 3

Image type

Back 3

Aids the segmentation in identifying which classes are which tissue type. This option is not used for multi-channel segmentation.

Front 4

Output image(s) basename

Back 4

Output images will have filenames derived from this basename.

Front 5

Binary segmentation: All classes in 1 imge will have filename

Back 5

<basename>_seg.hdr

Front 6

Number of classes

Back 6

to be segmented. Normally 3 (GM, WM, CSF) but if there is very poor grey/white contrast you may want to reduce this to 2. Alternatively if there are strong lesions showing up as fourth class you may want to increase this. If you are segmenting T2s you may need to select 4 so dark non-brain matter is processed correctly (this is not a T1 problem as CSF and dark non-brain look similar).

Front 7

5 Output images:

Back 7

Partial volume maps, binary segmentation: single image, binary segmentation: one image per class, restored input and bias field.

Front 8

Partial volume maps

Back 8

A (non-binary) partial volume image for each class, where each
voxel contains a value in the range 0-1 that represents the proportion of that class's
tissue present in that voxel. This is the default output.

Front 9

Binary segmentation: single image

Back 9

This is the "hard" (binary) segmentation, where
each voxel is classified into only one class. A single image contains all the necessary
information, with the first class taking intensity value 1 in the image, etc.

Front 10

Binary segmentation: one image per class

Back 10

This is also a hard segmentation output;
the difference is that there is one output image per class, and values are only either 0 or 1.

Front 11

Restored input:

Back 11

This is the estimated restored input image after correction for bias
field.

Front 12

Bias field:

Back 12

This is the estimated bias field.

Front 13

Bias field (definition)

Back 13

spatial intensity variations, or RF inhomogeneities.

Front 14

Use a-priori probability maps

Back 14

tells FAST to start by registering the input image to standard
space and then use standard tissue-type probability maps (from the MNI152 dataset) instead of
the initial K-means segmentation, in order to estimate the initial parameters of the classes.
This can help in cases where there is very bad bias field. By default the a-priori probability
maps are only used to initialise the segmentation; however, you can also optionally tell FAST
to use these priors in the final segmentation - this can help, for example, with the
segmentation of deep grey structures.

Front 15

Use file of initial tissue-type means

Back 15

tells FAST to use a text file with mean intensity values
(separated by newlines) for the starting mean values of the different classes to be segmented.
This is then used instead of the automated K-means starting parameter estimation.

Front 16

fast command-line

Back 16

Type fast to get usage. This is used for both single-channel and multi-channel segmentation
program.

Front 17

fast syntax

Back 17

fast [options] file(s)

Front 18

file(s)

Back 18

image, or multi-channel set of images, to be segmented.

Front 19

-S <n>

Back 19

number of image channels

Front 20

--channels+<n>

Back 20

number of image channels

Front 21

-t <n>

Back 21

type of image (n=1 for t2, 2 for t2, 3 for PD)

Front 22

--type=<n>

Back 22

type of image (t1, t2, PD)

Front 23

-o <base>

Back 23

basename for outputs

Front 24

--out=<base>

Back 24

basename for outputs

Front 25

-n <n>

Back 25

number of tissue-type classes

Front 26

--class=<n>

Back 26

number of tissue-type classes

Front 27

-B

Back 27

output estimated bias field

Front 28

-b

Back 28

output restored image (bias-corrected image)

Front 29

-H <v>

Back 29

MRF beta value for main segmentation phase (increasing this
gives spatially smoother segmentations)

Front 30

--Hyper=<v>

Back 30

MRF beta value for main segmentation phase (increasing this
gives spatially smoother segmentations)

Front 31

--Hyper=<v>

Back 31

MRF beta value for main segmentation phase (increasing this
gives spatially smoother segmentations)

Front 32

-R <v>

Back 32

MRF beta value for mixeltype (increasing this gives spatially
smoother mixeltypes and hence PVEs - NB: mixeltype is the
classification of what tissue-types are non-zero in a voxel)

Front 33

--mixel=<v>

Back 33

MRF beta value for mixeltype (increasing this gives spatially
smoother mixeltypes and hence PVEs - NB: mixeltype is the
classification of what tissue-types are non-zero in a voxel)

Front 34

-a <standard2input.mat>

Back 34

use prior probability maps for initialisation (must specify FLIRT
transform to standard space)

Front 35

-P <file>

Back 35

use prior probabilitiy maps at all segmentation stages

Front 36

--Prior

Back 36

use prior probabilitiy maps at all segmentation stages

Front 37

-s <file>

Back 37

file specifying initial tissue -type means

Front 38

--manualseg=<file>

Back 38

file specifying initial tissue-type means

Front 39

FAST can estimate the tissue volume for a given class. For the most accurate quantification...

Back 39

...we recommend using the partial volume estimates.

Front 40

How can the actual volume of tissue be calculated?

Back 40

From the corresponding partial volume map by summing up all the values. This can be done using fslstats and then multiplying the mean value by the volume (in voxels)

Front 41

For example, for an image called structural_bet that fast was run on, the tissue volume of tissue class 1 can be found by running the following:

Back 41

vol=`$FSLDIR/bin/fslstats structural_bet_pve_1 -V | awk '{print $1}'`

mean=`$FSLDIR/bin/fslstats structural_bet_pve_1 -M`

tissuevol=`echo "$mean * $vol" | bc -l`

echo $tissuevol

This prints the total tissue volume in voxels (which is also stored in the variable tissuevol). Note that to get the volume in mm3 just replace the {print $1} with {print $2} in the first line above.

Front 42

Type of image (eg T1, T2, PD) - note spaces in fast

Back 42

fast3: -t<n>
fast4: -t <n> (note space: t 2 not t2 as in fast3)
or --type=<n>

Front 43

Number of tissue-type classes to estimate

Back 43

fast3: -c <n>
fast4: -n <n> or --class=<n>

Front 44

Basename for all output files

Back 44

fast3: -od <base>
fast4: -o <base> or --out=<base>

Front 45

Output separate image for each (hard) segmentation class

Back 45

fast3: -os
fast4: -g or --segments

Front 46

output 1 probability map (of hard seg) per class

Back 46

fast3: -op
fast4: -p

Front 47

output bias corrected (restored) image

Back 47

fast3: -or
fast4: -B

Front 48

Output estimated bias field

Back 48

fast3: no equivalent
fast4: -b

Front 49

Output bias field correction image (reciprocal of the estimated bias field)

Back 49

fast3: -ob
fast4: no equivalent

Front 50

Suppress segmentation outputs

Back 50

fast3: -n
fast4: no equivalent

Front 51

Input file with manual starting values for class intensities

Back 51

fast3: -m <file>
fast4: -s <file> or --manualseg=<file>

Front 52

2D segmentation mode (for 3D images)

Back 52

fast3: -2
fast4: no equivalent

Front 53

number of main-loop iterations

Back 53

fast3: -i <n>
fast4: -O <n> or --fixed=<n>

Front 54

Bias field smoothing : n is iterations, m is FWHM in mm - rough guide is m = 2n^0.5

Back 54

fast3: -l <n>
fast4: -l <m> or --lowpass=<m>

Front 55

disable automatic parameter updating

Back 55

fast3: -p
fast4: no equivalent

Front 56

Main MRF beta value (initial segmentation calculation)

Back 56

fast3: -b <v>
fast4: -H <v> or --Hyper=<v>

Front 57

Output dilated bias correction field (extra n iterations of smoothing)

Back 57

fast3: -oba <n>
fast4: no equivalent

Front 58

Enable partial volume classification

Back 58

fast3: -e
fast4: (default)

Front 59

Do not do partial volume classification

Back 59

fast3: (default)
fast4: --nopve

Front 60

Output partial volume (PVE) images (one per class)

Back 60

fast3: -ov
fast4: (default)

Front 61

MRF beta value for PVE classification (fast3) or mixeltype in in PVE classification (fast4) - these are related but not exactly equivalent

Back 61

fast3: --b <v>
fast4: -R <v> or --mixel=<v>

Front 62

Disable parameter updating during PVE classification

Back 62

fast3: --p
fast4: no equivalent

Front 63

Turn on verbose/diagnostic output (fast3 has 6 levels, fast4 is binary)

Back 63

fast3: -v<n>
fast4: -v or --verbose

Front 64

Use prior probability maps for initialisation (fast4 requires a FLIRT matrix whereas fast3 requires extra argument -ap)

Back 64

fast3: -a
fast4: -a <standard2input.mat>

Front 65

Path and filename prefix for where to find pre-registered prior probability mpas (fast4 requires transformation matrix with -a instead)

Back 65

fast3: -ap <prefix>
fast4: no equivalent

Front 66

Use prior probability maps throughout calculations (must be used with -Ap for fast3 or -a for fast4)

Back 66

fast3: -A
fast4: -P or --Prior

Front 67

Path and filename prefix for where to find pre-registered prior probability maps

Back 67

fast3: -Ap <prefix>
fast4: no equivalent

Front 68

Path/filenames for alternative prior images

Back 68

fast3: nearly equivalent to -ap or -Ap
fast4: -A <prior1> <prior2> <prior3>

Front 69

Loop iterations during initial bias-field removal phase

Back 69

fast3: no equivalent
fast4: -I <n> or --iter=<n>

Front 70

MRF beta value during initial bias-field removal phase

Back 70

fast3: no equivalent
fast4: -f <v> or --fHard=<v>

Front 71

Number of segmentation-initialization iterations

Back 71

fast3: no equivalent
fast4: -W <n> or --init=<n>

Front 72

Do not remove bias field

Back 72

fast3: no equivalent
fast4: -N or --nobias

Front 73

Number of input channels (images)

Back 73

Mfast: -s <n>
Fast4: -S <n> or --channels=<n>