Only first spot got calculated while analyzing IDD

[ni484call5]

Note

matRad version: "Blaise" v2.10.1(matRadHITC-7a6004d9), copied from dev branch in 20240817.

Hello, I'm working on analyzing the IDD calculated by matRad with our own beam data.

I'll describe the workflow briefly and the issue encountered.

1. create IDD, comprise of $\frac{0.04MU}{spot} * \text{100spots}$ at central axis with SSD water tank
2. export dicom from our treatment planning system, including rtplan, ct, contouring, rtdose.
3. import files shown above to matRad and transform to .mat file with matRad_importDicomWidget.m
4. calculate dose with matRad_calcDoseForward.m.

In the experiment, I noticed that only first spot got calculated. In other word, the variable, resultGUI.w, showed 4.0000e-08 rather 1x100 double, comprised of [4.0000e-08; 4.0000e-08; ...], shown in stf.ray.weight.

I guessed the issued caused by [matRad_doseEngineBase] (https://github.com/e0404/matRad/blob/dev/matRad/doseCalc/%2BDoseEngines/%40matRad_DoseEngineBase/matRad_DoseEngineBase.m)

stf(i).numOfBixelsPerRay(j) is 1, causing k=1 and only first weight, stf(i).ray(j).weight(k), tranfer into calculation.

line 212
for k = 1:stf(i).numOfBixelsPerRay(j)
    counter = counter + 1;
    w(counter) = stf(i).ray(j).weight(k);
end

I'm trying to solve the problem by patch variable like script shown below.

stf.numOfBixelsPerRay = 100;
stf.totalNumOfBixels = 100;
stf.ray.focusIx = ones(1,100);
...

Is it correct or other suggestion for this issue?

I really appreciate you read this.

regards,
nicholas

[wahln]

When you do a forward calculation with matRad_calcDoseForward(...), we avoid storing the spot-specific information. Spots will directly be added up. Still, the output is weird. I think it might rather be an issue of the DICOM import, that it doesn't import the spot geometry into the stf correctly, and while patching it might work, I would rather check if we can see what is going on during DICOM import of the RTPlan.

Could you tell me a bit about the plan geometry? How many energy layers, how many spots laterally?

[ni484call5]

Thanks you reply.

Pseudo-water tank locate at center and surrounded by air. One beamlet from vertical beam line aimed at isocenter, x=0 and y=0, and SSD was selected. The beamlet comprise of 0.04MU * 100 spots and only one energy, 138.8 MeV, was selected. That means
stf.ray.weight = 1x100 double, comprised of [4.0000e-08; 4.0000e-08; ...].

On my opinion, with the setup discussed above, the two variables cause only first one spot got calculated, in term of weight, in matRad_doseEngineBase.

1. stf.numOfRays = 1: because only one beamlet aiming at (0,0)
2. stf.numOfBixelsPerRay = 1: because only one energy at (0,0)

line 212
for k = 1:stf(i).numOfBixelsPerRay(j)
    counter = counter + 1;
    w(counter) = stf(i).ray(j).weight(k);
end

Thanks your suggestion! I'll go back to inspect import process, maybe some DICOM UID haven't be patched by myself correctly.

regards,
nicholas

[wahln]

Do I understand correctly that you have 100 spots in the same direction with the same energy, so they are all on top of each other?

[ni484call5]

Yes!

[wahln]

Okay, so it is indeed weird that stf.numOfBixelsPerRay(j) == 1

Retracing your copy date as 17th of August last year, I identified this commit as the point where you branched off for your developments: https://github.com/e0404/matRad/tree/01b0d9db822c24d29c6743b21a5ac634e22e82a1

This is the part in the DICOM import on this branch that probably goes wrong:
https://github.com/ni484call5/matRad/blob/5b175b9d0d3bade26924aced63c8a7421f7aab1e/dicom/matRad_importDicomSteeringParticles.m#L202-L217

There, a unique operation is run on the energies intended as cleanup, which will remove your duplicate energies from this ray. I think the original idea did not consider a scenario like yours where you would superimpose identical spots. In your case, this just returns the single energy, and then subsequently only recognizes a single beamlet on this ray. Probably, we do not need the cleanup in
https://github.com/ni484call5/matRad/blob/5b175b9d0d3bade26924aced63c8a7421f7aab1e/dicom/matRad_importDicomSteeringParticles.m#L204

You could try removing this and see if the import correctly imports the RTPlan to include all spots.