SBRT technical discussion

[scarbrtj]

---

Members don't see this ad.

How do you know that the 1 mm drift is real. When dealing with mm measurements, could this just be part of the variability in taking multiple measurements? If I measured a stable phantom to the mm fifty times wouldn’t I make some “errors”and get some “drift” in a couple cases?

You don’t know if it’s real or not. But like Rummy said you go to war with the army you have not the army you want. In theory inside the Varian system it’s actually measuring to 0.1 mm precision ie an extra significant digit. Exactrac actually spits out numbers to 0.01mm precision as I recall.

 

[communitydoc13]

In theory inside the Varian system it’s actually measuring to 0.1 mm precision ie an extra significant digit. Exactrac actually spits out numbers to 0.01mm precision as I recall.

My guess is 1 mm is well within 1 standard deviation of a "person's" uncertainty in repeatedly aligning an idealized fiducial where there is no uncertainty in the image itself. Just look at 1 mm on a ruler. Now look at that projection on a CBCT with fiducial artifact. Never mind things like fiducials can themselves move or your slice thickness on planning CT (1 mm on best of scans) or slow acquisition of CBCT.

 

[communitydoc13]

I did not advocate single-fraction as definite treatment for prostate cancer. I was merely making the point in what scenario internal movement would be a limiting factor in safely applying SBRT.

Never meant to imply you were advocating for this and I agree that the stakes are higher for single fraction. (I would want gated treatment with real time tracking on any single fraction body SBRT.)

 

[scarbrtj]

In theory inside the Varian system it’s actually measuring to 0.1 mm precision ie an extra significant digit. Exactrac actually spits out numbers to 0.01mm precision as I recall.

My guess is 1 mm is well within 1 standard deviation of a "person's" uncertainty in repeatedly aligning an idealized fiducial where there is no uncertainty in the image itself. Just look at 1 mm on a ruler. Now look at that projection on a CBCT with fiducial artifact. Never mind things like fiducials can themselves move or your slice thickness on planning CT (1 mm on best of scans) or slow acquisition of CBCT.

I don’t disagree with any of these or aforementioned misgivings per se. But in my clinic, with my equipment, with my therapists, using their eyeballs on my patients, this was the data. As an aside, the IGRT acceptance and QA on the machine somewhat requires accepting that sub mm IGRT precision (and accuracy I guess?) is possible.

 

[Gfunk6]

I can comment a bit on the CyberKnife tracking algorithm. It can track both internal (target) movement and external (chest wall) movement. The former is tracked by oblique x-rays either based on an implanted fiducial or a distinct radiographic appearance of the target relative to background (e.g. Xsight Lung). The latter is tracked by three LEDs placed on top of the anatomically relevant area of the chest wall - those LEDs can either be attached via a customized velcro vest or the use of form fitting compression garments.

Typically when a CK fraction starts:

1. You optimize the xray technique based on the patient for optimal imaging
2. Line up the spine on both oblique views
3. Line up the target on both oblique views
4. Run the Synchrony model. This looks at the movement of the three LEDs on the chest wall and builds and individualized model of the patient's breathing. Generally there are 8-14 points of the patient's respiratory cycle based on establishing a peak, valley, and points in-between.
5. Beam on starts.
6. During treatment x-rays can be taken every 20-60 seconds to verify #3 and #4 remain accurate. If they are out of tolerance, the CK automatically disengages and awaits operator intervention.

 

[communitydoc13]

I don’t disagree with any of these or aforementioned misgivings per se. But in my clinic, with my equipment, with my therapists, using their eyeballs on my patients, this was the data. As an aside, the IGRT acceptance and QA on the machine somewhat requires accepting that sub mm IGRT precision (and accuracy I guess?) is possible.

I think we are on the same page here. Thank you for your data and that outlier is interesting. I usually assume that lateral shifts are smallest in the pelvis.

My understanding is that the winston lutz and other QA is to ensure fidelity with rotation along the treatment axis as well as fidelity between the imager and the treatment iso. The main sources of uncertainty are in the image itself and interpretation of it.

 

[scarbrtj]

Thank you for your data and that outlier is interesting.

Uh yeah. You look back and think "what the heck was that about?" Makes you wonder. Gather enough data and you see rare "IGRT gremlins" like that. Ultimately, does it matter? And that's the art of radiation oncology.