ROE architcture - cerr/CERR GitHub Wiki
Components
1. Segmentations and dose distributions
CERR archives contain RT/Image objects stored in a MATLAB cell array variable named 'planC'. The ROE plug-in extracts relevant information (dose distribution, segmentations of normal structures and tumors, etc) from CERR archives.
Dose bins and volume histograms are computed within ROE using the following CERR functions:
[dosesV,volsV] = getDVH(structNum,planNum,planC);
[doseBinsV,volHistV] = doseHist(dosesV,volsV,binWidth);
Note: binWidth = .05 by default.
2. Protocols, model parameters, and clinical constraints
Treatment protocols, dosimetric model parameters, and clinical criteria are defined via JSON format files.
For further information see JSON settings for ROE or sample files distributed with CERR.
3. Dosimetric models
Predicted outcomes (TCP/BED and NTCP) are calculated using user-defined Matlab functions. Several such functions including LKB, linear, and logistic forms are distributed with CERR.
The following syntax should be used to define custom functions:
outcome = functionName(paramS, doseBinsV, volBinsV)
where paramS
is a dictionary of model parameters generated from the JSON-format model parameters.
Parameter values can be accessed using
paramVal = paramS.parameterName.val;
Note: For models based on multiple structures, dose and volume bins are passed as cell arrays.
outcome = functionName(paramS, doseBinsC, volBinsC)