JSON settings for ROE (models) - cerr/CERR GitHub Wiki

The functional forms of several dosimetric models are distributed with CERR. Associated model parameters are specified through JSON format files. This includes parameters required for bio-normalization of planned doses, i.e., the fraction number or fraction size used in developing the model.

JSON fields

General

  • Name : Name of the model
  • Type : May be "TCP", "BED" , or "NTCP".

For fractionation correction

  • fractionCorrect: "Yes" or "No"
  • correctionType: "fsize" or "nfrx"
  • fractionSize: Fraction size of modelled dataset (required for correctionType: fsize)
  • stdNumFractions: No. fractions of modelled dataset (required for correctionType: nfrx)
  • abRatio: Radiosensitivity (α/β) in Gy

Functional form

  • function: Function name defining model form.
    Model forms distributed with CERR are detailed here. Custom implementations should follow the established function signature:
   outcome = functionName(paramS, doseBinsV, volBinsV)

Model parameters

Each parameter is specified through 4 fields

  • val: Parameter value or function name to evaluate. Set t 1 for constant parameters.

  • weight : Coefficient. Set to fixed value for constant parameters.

  • type : An array of values specifying data types of value & weight inputs, respectively.
    May be "string", "bin"(for binary) or "cont" (for numeric inputs).

  • cteg: Specify supported values (e.g.: cteg =[0,1]) for binary type variables.

  • desc: Map labels to descriptions (e.g.: desc =["No","Yes"]) for binary type variables only.

  • Syntax for structure-specific parameters

   "structure":{
                  "structName1": {
                                   "param1":{
                                              "val": val
                                              "weight": weight
                                              "type": [val_type, weight_type]
                                              "cteg": [cteg1, cteg2]
                                            }
              }

Example

JSON settings for models distributed with ROE are available here.

E.g. Esophagitis model [1]
{
	"name": "Esophagitis (Huang)",
	"type": "NTCP",
	"stdNumFractions": 35,
	"fractionCorrect": "Yes",
	"correctionType": "nfrx",
	"abRatio": 10,
	"parameters": {
		
		"concurrentChemo": {    
				"val": 1,
				"weight": 1.5,
				"type": ["bin", "cont"],
				"cteg": [0,1],
				"desc": ["No", "Yes"]
                          },


		
 		"structures": {
			       "Esophagus":{
                			  "meanDose": {
							"val": "calc_meanDose",
 							"weight": 0.0688,
							"type": ["string", "cont"],
							"cteg": ["calc_meanDose","calc_meanDose"]
                                    	               }
				          }
			},

                "constant": { 
				"val": 1,
				"weight": -3.13,
				"type": ["cont", "cont"],
				"cteg": [0, 1]
		            }
	               },
	"function": "logitFn"
}

References

  1. Huang, E. X., Hope, A. J., Lindsay, P. E., Trovo, M., El Naqa, I., Deasy, J. O., & Bradley, J. D. (2011). Heart irradiation as a risk factor for radiation pneumonitis. Acta Oncologica, 50(1), 51-60.