User Tools

Site Tools


jvs:fscan:manual:chapter2:roi

This is an old revision of the document!


fScan Reference Manual, Chapter 2 (CmdLineOptions): roi

ROI -- Define and manipulate Regions of Interest

usage:

  • -roi POINT [NAME] x y z
  • -roi BLOCK [NAME] x0 y0 z0 nx ny nz
  • -roi BLOCK [NAME] x0 y0 nx ny
  • -roi BLOCK [NAME] x0 nx
  • -roi CIRCLE [NAME] x0 y0 z0 rad
  • -roi CIRCLE [NAME] x0 y0 rad
  • -roi CIRCLE [NAME] x0 y0 z0 radx rady radz
  • -roi ARROW [NAME] x0 y0 z0 x1 y1 z1
  • -roi ARROW [NAME] x0 y0 x1 y1
  • -roi ARROW [NAME] x0
  • -roi TEXT [NAME] textstring x0 y0 z0 x1 y1 z1
  • -roi TEXT [NAME] textstring x0 y0 x1 y1
  • -roi MASK datasetname
  • -roi FILE roifile
  • -roi RX refimage (not yet implemented)
  • -roi MEAN flag
  • -roi CORREL flag refROI binsize corrROI name thresh
  • -roi CLUSTER refimage flag name
  • -roi MERGE refimage flag name minvalley minsize mindist
  • -roi STATS refimage flag showflag mode name value nlevels level0 dlevel

Adding new ROIs

Modifying ROIs

cluster ROIs

ROI's can be added to the current data set by explicitly specifying the ROI type, location, and size. The list above shows the various ROI types and how they are specified. The meaning of he arguments is:

  • NAME - specify a name for this ROI (no spaces or punctuation)
  • X0,Y0,Z0 - specify the starting location of the ROI as upper left corner of the first slice. The default origin is 0,0,0.
  • NX,NY,NZ - specify the size of the ROI. The default size is 1x1x1 (i.e. a point) if X0,Y0,Z0 are specified, otherwise it is the entire volume.
  • RAD,RADX,RADY,RADZ - specify radii for circular ROIs
  • COLOR - specify color for ROI (24-bit RGB value)

The RX type of ROI (when implemented) creates a slice prescription ROI on the current data set based on slice orientation in REFIMAGE.

The CLUSTER type ROI assumes that the current data set is an ROI mask image (in which regions are defined by voxels with the same intensity value). The CLUSTER command uses the input mask image to create an ROI list (name=NAME) of CLUSTER type ROIs where each cluster is defined by its mask value. For each CLUSTER, the center-of-mass location and the radius along the 3 principal axes are stored. By default, the new ROI list is attached to the input mask data set.

  • REFIMAGE, find peak intensity value and voxel location.
  • FLAG: (no values yet)
  • NAME: name of ROI list. If omitted, the list is linked to input mask images.

For manipulating existing ROIs:

The ROI command can also be used to manipulate existing ROI's.

-roi CLUSTER refimage flag name

	Assumes roi mask image as input.
Create an ROI list (name=NAME) of clusters where each
  cluster is defined by its mask value. For each cluster,
  store center-of-mass and radius along principal axes.
If REFIMAGE, find peak intensity value and voxel location.
By default, ROI list is attached to input mask images. 
FLAG: (no values yet)

NAME: name of ROI list. If omitted, list is linked to 
        input mask images.

-roi MERGE refimage flag name minvalley minsize mindist

The MERGE command assumes that the current image is an ROI mask.

MERGE edits the ROI list. Merge adjacent clusters if:

1) distance bet peaks < MINDIST 
2) stat value in valley between peaks is within MINVALLEY percent of lower peak
3) smaller cluster contains fewer than MINSIZE voxels

FLAG: 2 - do not modify input mask image values when editting ROI list
 (i.e. if merging ROI clusters, change merged mask values)

NAME: name of ROI list. If omitted, list is linked to input mask images.
MINVALLEY: (default= 60% of lower peak maximum)
MINSIZE: (default= 5 voxels)
MINDIST: (default= 40 mm)

-roi STATS refimage flag showflag mode name value nlevels level0 dlevel

Calulate stats in input data by ROI

REFIMAGE - dataset to use as ROI mask(s)
FLAG - stats options (select one value 0-7; add other values)
     0 - default stats (3 = Mean,Count,StDev)
     1 - include mean
     2 - include counts, and mean
     3 - include stdev, and above
     4 - include range, and above
     5 - include ctr loc, and above
     6 - include peak val&loc, and above
     7 - include intensity weighted ctr loc, and above
   010 - mean time course
   020 - calculate laterality index if available
   040 - calculate stats omitting zero values
  0100 - report ROI volumes in mm3 (voxel count X voxel volume)
  0200 - include ROI's with only zero values (usually ignored)
SHOWFLAG - Report generation flags
     1 - tab-separated columns in text output
         (default is space-separated)
     2 - comma-separated columns in text output
  0100 - save as XML text file
  0200 - save as tab-delimited text file
  0400 - plot ref time series
 01000 - save as ref time series
 02000 - generate AMPLE statistics
 04000 - ignore thresholds (default: use threshold if active)
010000 - Watch flag (?)

MODE
     0 - default: 1 if active ROI, 2 if this has glossary,
         3 if prev has glossary, 4 if prev is atlas
     1 - block ROI's (active or all)
     2 - mask values in current data
     3 - mask values in prev data set
     4 - atlas areas
     8 - threshold (can combine with 1-4)
NAME - save stats in file NAME, if specified
VAL: if > 0, only calculate for this mask value
NLVLS: # threshold levels to calc separately (def=1)

If input is a Surface:

FLAG
  1 - surface area of AREA ROI or path length of contour ROI
  

-roi CORRELATE flag ref_roi binsize corr_roi name thresh

Generate new data set (named NAME) containing multiple correlation maps based on correlating the time course for each voxel within the REFROI in the current data set with each voxel in the CORR_ROI. The 2 ROI's can be the same; by default they are the entire data set volume.

FLAG can be used for specifying options:

1 - generate neighbor correlation map. Correlates each voxel time course with
    its 26 neighbors and saves the mean correlation coefficient. 
2 - do not mask out low intensity values (by default CORRELATE assumes the -m command)
4 - use REFROI as CORR_ROI as well

REFROI can be the name of an ROI, or it can be:

0 - use the default: 5 if FLAG&
1 - use the time course at the current voxel as reference
2 - use the current point ROI time course as reference
3 - use every voxel within the current box roi as a separate reference (this generates 
    1 map per ref voxel; the ref voxel in each  map will have correl value of 1.0)
4 - use every voxel in all active ROI's in this data set
5 - use every voxel in the volume
<0 - Negative ROINUM - specify any active ROI by the negative of its ROINUM\\

Default is 3 if there is a box ROI, otherwise it is 1

BINSIZE is how many voxels in each dimension to bin together for reference (default=1).

CORR_ROI is the ROI to use as a mask for which voxels to test. By default,CORR_ROI is the entire data set. Like REFROI, CORR_ROI can be the name of an ROI, or it can be:

0 - default. same as 5
1 - use REFROI as CORR_ROI
3 - use the current box roi as CORR_ROI 
4 - correlate with voxels in all active ROI's in this data set
5 - correlate with every voxel in the data set volume (default)
<0 - Negative ROINUM - specify any active ROI by the negative of its ROINUM

NAME is the name of the output data set; default adds suffix roicorrel to current data set name.

THRESH is the auto-correlation threshold for counting neighboring voxels as positively correlated (default=0.3)

See Also:
fScan Home, fScan Manual, CmdLineOptions, Manual Help

jvs/fscan/manual/chapter2/roi.1433618616.txt.gz · Last modified: 2015/06/06 19:23 by voyvodic