Independent dose calculation in carbon therapy

[ni484call5]

Hello,

I'm trying to import log file, recorded by nozzle while beam on in carbon therapy, to matRad and recalculating dose in TG119 phantom. The dose calculated from log file by matRad will be compared with origin treatment planning by the method of 2 dimension gamma index. For the purpose mentioned above, I've built base data for specific machine matching machine.data.Z, depth, sigma etc with log file. And next, I guess i should convert log file to stf and use matRad_calcDoseDirect.m. While I can't figure out how to do that? Or you will suggest me to do that by importing log file through matRad_importDicom.m?

regards, nicholas

[wahln]

You are on the right track.

• Building the base data is the most important step and there you should make sure that this is really dosimetrically equivalent. One still undocumented tip: What you can also do on our current dev branch is to put the beam model at the nozzle as an emittance parameterization directly into the base data and use TOPAS for recalculation. Let me know if this is of interest.
• To run matRad_calcDoseDirect, in addition to the phantom you need two things: a pln struct and an stf struct. The pln should be quite easy to set up using the examples, as it contains mainly meta-information. The stf is a bit more complicated. There's two options:
  • If you have an RTPlan (before the log file), you can start by importing this via our DICOM import (I can't guarantee that it works due to some vendor-specific issues sometimes). This at least will give you the structure of the stf. Then, depending on the log file contents, you could change the respective variables (i.e. spot positions, MUs, etc.) directly in the stf.
  • You could also build the stf from the log file directly. There's information about the data structure here: https://github.com/e0404/matRad/wiki/The-stf-struct . The link is not yet up-to-date with our dev branch, which allows some more variables (i.e. to manage the MU calibration if present in the machine file). Basically, you would need to iterate over all spots in your log-file and derive the correct positions in the ray coordinate system for each spot (all coordinates relative to isocenter, which makes it more simple with BEV data from the log file).

Happy to help you with this along the way if you encounter obstacles or wonder about a specific thing / variable from the log-file to correctly set-up in the stf.

[github-actions[bot]]

This issue was automatically marked as stale because it has been sitting there for 14 days without activity. It will be closed in 14 days if no further activity occurs.

[ni484call5]

Thanks your reply.

As your first suggestion, I replace value, spot position and mu, in plan with the one recorded in log file. Then, import the modified RT plan into .mat file with the import function in matRadGUI and calculating dose with the .mat file. The process works well.

Next step, In order to increase app efficiency and save time, I try to import only original RTPlan and CT, no RTStruct, into matRad by dicom import function in matRadGUI. The .mat file is created well. Then, using matRad_calcDoseDirect.m with it, the .mat file,

1. In the first case, an patient RT plan, the dose distribution shown on matRad is broken.

2. In the second case, an 10*10*10 uniform 2Gy physical dose deposited in water tank, the error shows up

   
   Warning: ray does not hit patient. Trying to fix afterwards... 
   Error using matRad_computeSSD>matRad_closestNeighbourSSD (line 100)
   Could not fix SSD calculation.

Error in matRad_computeSSD (line 77) stf(i).ray(j).SSD = matRad_closestNeighbourSSD(rayPos_bev, SSD, rayPos_bev(j,:));

Error in matRad_calcDoseInit (line 148) stf = matRad_computeSSD(stf,ct);

Error in matRad_calcParticleDose (line 39) matRad_calcDoseInit;

Error in matRad_calcDoseDirect (line 74) dij = matRad_calcParticleDose(ct,stf,pln,cst,calcDoseDirect);

Error in nL_excutePlan (line 12)

I'll try to figure out the importance of RTStruct in the process mentioned above and the possibility to skip it while import RTPlan and calculate dose.

regards,
nicholas

[wahln]

matRad uses the contours to internally figure out where to calculate results. I would suggest to at least have a contour for the whole patient body and a target contour. I've never seen the behavior you observe in your dose distribution, but it can be very well related to missing contours. Also rays missing the patient can be related to that.

Also I suggest to run your stuff with our current "dev" branch.