CVT3D makes incomplete spheres on complex shaped targets

[varian-ma]

Do we need to look at tweaking the margin or structure mask sampling?

[japnp]

Hi Anthony,

We revised the target it takes in from PTV to ptvRetract. This should make sure partial spheres are not created at boundaries. If two spheres are getting joined into one sphere still, that very likely has to do with the sampler. We have a check for that in the python version with LHS but not in C# (I wasn't sure if it was necessary because I though Voronoi tessalation should take care of it but if not we can add a check for that). Changes have been pushed; can you please check it on your end and let us know what you want?

We downloaded the dicom and tried. Spheres are generated well within the structure. We'll need to check with Ilias and make sure we're not doing retraction twice (I don't think we are but it would like to double check).

Thanks!

[varian-ma]

@japnp Is the size of ptvRetract relative to the position of the full spheres slider?

[japnp]

Hi Anthony,

Yes. So the ptvRetract is how much we retract the original PTV before we make spheres. So on slider we can allow partial spheres. If that slider is at 1, we choose ptvRetract such that there are no partial spheres.

Thanks, Japan

---

From: Anthony Magliari @.> Sent: Monday, October 14, 2024 12:11 PM To: Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper @.> Cc: japnp @.>; Mention @.> Subject: Re: [Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper] CVT3D makes incomplete spheres on complex shaped targets (Issue #24)

@japnphttps://github.com/japnp Is the size of ptvRetract relative to the position of the full spheres slider?

— Reply to this email directly, view it on GitHubhttps://github.com/Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper/issues/24#issuecomment-2411693388, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABVKXBWTJODKQT5QWWETKBLZ3PUKFAVCNFSM6AAAAABOSRD2IWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJRGY4TGMZYHA. You are receiving this because you were mentioned.Message ID: @.***>

[varian-ma]

My last 3DCVT built doesn't give partial structures but instead they get merged together. I can try latest in main (or just larger spacing).

[japnp]

Hi Anthony,

If they're getting merged together, we'll have to put in a check for that like we did for the python version. We can put that in this evening and push it back.

Thanks, Japan

Sent from my T-Mobile 5G Device Get Outlook for Androidhttps://aka.ms/AAb9ysg

---

From: Anthony Magliari @.> Sent: Monday, October 14, 2024 1:52:20 PM To: Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper @.> Cc: japnp @.>; Mention @.> Subject: Re: [Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper] CVT3D makes incomplete spheres on complex shaped targets (Issue #24)

My last 3DCVT built doesn't give partial structures but instead they get merged together. I can try latest in main (or just larger spacing). image.png (view on web)https://github.com/user-attachments/assets/8a7e3285-ca42-46e0-beff-a03fd6967757

— Reply to this email directly, view it on GitHubhttps://github.com/Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper/issues/24#issuecomment-2411893057, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABVKXBVTAFSUDWCWHYY26ADZ3QAFJAVCNFSM6AAAAABOSRD2IWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJRHA4TGMBVG4. You are receiving this because you were mentioned.Message ID: @.***>

[japnp]

Hi Anthony,

We added a check for merged spheres so that should not be an issue anymore. We pushed it to CVT3D branch. We also changed the build to .NET 4.6.1. We changed it back from ptvRetract to PTV. If you prefer to keep ptvRetract as the target (this is correct but using PTV as target may also be okay). Please check it on your end and let me know what edits you'd prefer and whether you want to switch back to ptvRetract or not. Plmk if you have any further questions or concerns.

Thanks!

[varian-ma]

I think it's OK to merge to main but take the trailing 2 out of the build name. I feel like 3DCVT needs a PTVretract but a much smaller one. You can see with 13mm radius and 30mm spacing the original number was 41 seeds, your check brought it down to 39 and no spheres are connected but now there is an occasional partial sphere. Perhaps a very minor margin for CVT3D?

[japnp]

Hi Anthony,

Sounds good, it'll be done this evening! 😊

Thanks, Japan

Sent from my T-Mobile 5G Device Get Outlook for Androidhttps://aka.ms/AAb9ysg

---

From: Anthony Magliari @.> Sent: Tuesday, October 15, 2024 11:42:23 AM To: Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper @.> Cc: japnp @.>; Mention @.> Subject: Re: [Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper] CVT3D makes incomplete spheres on complex shaped targets (Issue #24)

I think it's OK to merge to main but take the trailing 2 out of the build name. I feel like 3DCVT needs a PTVretract but a much smaller one. image.png (view on web)https://github.com/user-attachments/assets/90351c76-c409-49f6-b1c1-3e19a903f5aa You can see with 13mm radius and 30mm spacing the original number was 41 seeds, your check brought it down to 39 and no spheres are connected but now there is an occasional partial sphere. Perhaps a very minor margin for CVT3D?

— Reply to this email directly, view it on GitHubhttps://github.com/Varian-MedicalAffairsAppliedSolutions/MAAS-SFRThelper/issues/24#issuecomment-2414382419, or unsubscribehttps://github.com/notifications/unsubscribe-auth/ABVKXBXXJN4CZ5CD5TO3SKDZ3UZV7AVCNFSM6AAAAABOSRD2IWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMJUGM4DENBRHE. You are receiving this because you were mentioned.Message ID: @.***>

[japnp]

Hi Anthony, we wrapped this up. Just one thing to keep in mind - users can use the the slider control how much margin they want to give ptvRetract and that will also determine the number of spheres being created. Everything else is done.