Preprocessing settings for scans - cerr/CERR GitHub Wiki

This field contains nested JSON objects specifying scan identifiers and pre-processing parameters. Settings for multiple scans can be input using nested lists. Note : Exported CT images are converted to HU (offset subtracted by default).

• identifier : Identifiers are used to determine the series to be exported.
  Supported identifiers:

  • imageType : "CT SCAN" or "MR SCAN"
  • seriesDescription :
  • scanType :
  • scanNum :
  • scanDate : first" or "last"
  • assocStructure : (list of structure names or "none").

  Derived images (filtered/registered) generated using this pipeline can be selected by setting the "filtered" or "warped" flags (default: 0 (off)).

          "identifier": {"identifier_name": "value",  "warped" : 1,  "filtered" : 1}

For further details, see getScanNumFromIdentifiers.m

• required : Indicate if scan volume is strictly required ("yes"— this is the default) or optional ("no").

These include parameters for image pre-processing ("filter", "register", "resample", "crop", "resize", "view", and "channels"), described below.

• filter : Dictionary specifying "imageType" (any of the supported image filters) along with required parameters.

• register : Dictionary of parameters for image registration. Required sub-fields include:baseScan, movingScan, (registration) tool, andmethod. inputCmdFile and copyStr may also be specified. Base and moving scans are specified through supported identifiers, with the optional addition of a binary filtered flag (default:0) if using the result of the previous "filter" stage.

  • baseScan: Dictionary specifying "identifier" and (optional) "filtered" flag.
  • movingScan: Dictionary specifying "identifier" and (optional) "filtered" flag.
  • tool: Define registration tool: "plastimatch","elastix","ants", or "custom"
  • method: "align center", "rigid", "bspline", "demons plastimatch"
  • inputCmdFile: Path to text file containing registration commands.
  • copyStr: List of structures to deform from moving to base scan frame of reference.
    

    Example:
    "register" : {"baseScan": {"identifier":{"name":"value", "filtered":1}},
    "movingScan":{"identifier":{"name":"value", "filtered":0}},
    "tool": "plastimatch",
    "method": "rigid",
    "inputCmdFile":"plastimatch_command/plm_reg_settings_with_starting_transform.txt",
    "copyStr":["Str1","Str2"]}

• view : Transform image orientation to axial, sagittal or coronal orientations. Note: Input scans are assumed to have axial orientation.

• required : Indicate if scan volume is strictly required ("yes" by default) or optional ("no").

• view : List of output image orientations. Supported options: "axial" (default), "sagittal", "coronal" or combination.
  E.g. Single view: "view" : ["axial"],
  Multi-view: "view" = ["axial","sagittal"],

• resample : Dictionary of resampling parameters with fields:

  • resolutionXCm, resolutionYcm, resolutionZCm: Voxel dimensions in cm.
  • method : Supported methods include "sinc", "linear", "cubic", "spline", "makima", "nearest", and "none". (default).

• crop : Dictionary of parameters for cropping. Valid fields include "method", "params", and "operator".

  • method: Valid cropping method (see table). May contain a list of supported methods, which are applied sequentially.
  • params: Dictionary of parameters required for selected cropping method.
  • operator: Supported values for chaining cropping methods are "" (empty/none), "union" and "intersection".

  For descriptions and examples of the cropping methods listed below, see section on pre-processing methods. For further details, see getMaskForModelConfig.m.

To save the ROI mask(s) returned by the selected cropping method(s) to planC (including auto-generated patient outline), include the following under "params":

      "saveStrToPlanCFlag": 1,  
      "outStrName": "roi_out" 

5. resize : Dictionary of parameters for resizing with fields:

   • methods: Supported methods include "pad2d", "pad3d", "padorcrop2d", "padorcrop3d", "bilinear", "bicubic", "sinc" and "none" (default). padorcrop determines the appropriate operation from input and output image dimensions.
   • size : Output dimensions [numRows, numCols]

6. channels : Dictionary of parameters for populating image channels. For each channel, the following fields may be specified:

   • imageType: Valid options include "original" (default) or any of the supported image filters along with required parameters.
   • slice : Supported options include "current", "current-n" for the 'n'th slice above the current slice, and "current+n", for the 'n'th slice below the current slice.
     

     Syntax:
     "channels":{"imageType": validType, "slice": validSlice}

   Optional input:

   • "intensityOutsideMask: Replace intensities outside the region of interest (structure used in cropping) with user-specified value. E.g.: "intensityOutsideMask" : {"val": 0},

   Examples

• Configuration for single channel output
  • With original image (default)
    "channels": {"imageType":"original", "slice":"current"}
  • With image texture
    "channels": {"imageType":{"Wavelets":{"Wavelets":"Coif", "Index":"1", "Direction": "HLH"}}, "slice":"current"}
• Configuration for multi-channel output
  • 2.5D
    "channels": [{"imageType":"original", "slice":"current-1"},
{"imageType":"original", "slice":"current"},
{"imageType":"original", "slice":"current+1"}]
  • Combination of original and derived images (same slice)
    "channels": [{"imageType":"original", "slice":"current"},
{"imageType":{"Wavelets":{"Wavelets":"Coif", "Index":"1", "Direction": "HLH"}},
"slice":"current"}]