Reply To: Vertical shifts from radiographic images

Forums Deep Inspiration Breath Hold Vertical shifts from radiographic images Reply To: Vertical shifts from radiographic images

Marko Laaksomaa

Hi Jose

Purpose of this text is to raise the discussion in the hospitals and share thoughts.

Some additions to my former text, now how to handle systematic errors concerning vertical shifts from orthogonal images, if we do not want to do those tricks 1-3 daily. In the point 1, where we have systematic isocenter error in both sternum and vertebra, and there has been a need to apply couch shifts patient in BH: we need new FB and BH reference surfaces and new couch vertical value for the upcoming setups (if the user uses additionally the planned couch vertical value for the setup). In point 2 where we ask patient to take more or less air into lungs based on mostly due to only sternum displacement in vertical direction, we need new BH surface and not necessary new FB surface. Still, I would always acquire new FB surface when we have to acquire new BH surface. Otherwise, there is a risk that there is something different between the FB and BH surfaces taken at different days, and we need to setup the patient also in BH in the upcoming fractions (do corrections on patient position or posture based on AlignRT BH surface after FB setup). This is what we would like to eliminate as often as possible. In the point 3 where we shift the couch due to mostly vertebra vertical error only, we need to acquire new couch vertical value (if the user uses the planned couch vertical for the setup as well) and certainly new FB surface. Still, I would always acquire also new BH surface when we have to acquire new FB surface. Otherwise…see text above.

To say it (too) shortly concerning BHL, in AlignRT DIBH cases errors in the sternum vertical we correct with the new surface with asking patient to take more or less air into lungs and errors in the spine vertical we correct with the couch shifts.

In the first fraction with AlignRT DIBH we begin the image evaluation accordingly: we acquire only LAT image and no matter what we shift the couch in the online match to meet the vertebra in vertical and LNG directions in the images. Then we ask patient to take BH to the window/level she sees on the monitor (DICOM surface) and acquire second LAT image. Now the vertebra is in its correct location in the image and we begin to check the sternum location. If it is too dorsally, we ask patient to take more air into lungs as much as the sternum vertical error in the image is. For this, we would like to have adaptive threshold feature, at least to only have a possibility to acquire new BH surface for the treatment with that feature. After we are satisfied and possibly acquired that new reference BH surface, we go on to the AP+LAT images. Once the images are “perfect”, patient chest wall is relaxed back to setup posture, possible couch shifts are taken into account, we acquire new FB gated setup surface (where we see entire arm tomorrowJ) and if we did not stick with the DICOM, we acquire new BH surface. We do not have FB DICOM from CT, and BH DICOM we take with the RPM guidance at the CT. We tested this FB+BH DICOM combination, but that often we finally found ourselves using VRT surfaces at some point of treatment that we considered it easier to create the FB VRT surface on the linac, at least so far.

In the first fraction, there are situations where we first shift the couch based on vertebra match in the LAT image and still based on AP+LAT images shift the couch out of that location again in LNG direction to make compromise match between sternum and the ribs. This is because rather often there is small displacement errors in between the vertebra and sternum in LNG and we cannot do anything for this patient pitched posture, except decisions in the matching location. Additionally, vertebra alone is typically the weakest possible location to finally match LNG direction, concerning the accuracy in the tangential image ribs LNG with DIBH. Therefore, if I think about the workflow where we would daily begin with the LAT image and vertebra match and see where the sternum is after that, I do not feel it practical, even if it may slightly improve the realized BHL accuracy. One reason is that there is a risk that we may begin, due to 1-3 mm shifts in the AP+LAT images, begin to daily acquire more images, acquire new surfaces for the treatment and go to record mode to get former surfaces active… The main reason is that nowadays with the AlignRT setup, the errors between vertebra and sternum are usually small enough in vertical direction in DIBH (in our latest data only 7% more than 4 mm). In another hospital with different workflow, with AlignRT, results can be of course different.

If in the weekly LAT- images users action limit allows less than 3 mm displacement in both vertebra and sternum location in vertical direction, this includes one risk. If in the images, the vertebra locates in the image acquisition position 2.9 mm too ventrally and sternum 2.9 mm too dorsally, BHL in the images is 5.8 mm too shallow, which error we should correct. How? Lower the couch 3 mm based on vertebra match and ask patient to take 3 mm more air than in the previous reference surface…

You had one question, where you wondered is the reason for BHL displacement in the LAT DIBH images due to wrong posture in the patient chest wall in FB or some kind of errors in (the workflow with) the system, which guides the patient to slightly wrong BHL. Based on the offline evaluation it is difficult to know, is the original reason for displacement errors between vertebra and sternum in LAT image that patient chest wall is more relaxed (too dorsally) or more tensed (too ventrally) at the CT in FB than it is in FB setup at the linac, leading to systematic errors in the realized BHL. We see these kind of systematic errors in the images also in the group of the FB breast patients sometimes. Or, is the reason something else in the workflow concerning the BH itself. If with the individual corrective actions you can get your LAT image repeatedly to match (both vertebra and sternum to their correct locations), this is most important. On the other hand, I feel that with the AlignRT you get much more supporting information how to correct setup errors based on offline images, if we compare it with the conventional tattoo setup. Needs to correct BHL after first fractions are rare in our case, but we need to know what to do once those appear. I think we will always face also these systematic BHL errors in the images occasionally and therefore there will always be needs to readjust the reference surfaces or planned couch vertical values at some point of treatment, typically during the first fractions, regardless how good our entire workflow is.

As noticed here on the forum, there are differences between the units on how to evaluate the succeeding of the BHL as a goal to spare the heart. One checks the diaphragm level and heart location in CBCT images, other checks the heart location in tangential images, third checks the displacement between vertebra and sternum in the LAT images, forth mostly relies on the setup workflow and the AlignRTsystem only…I have tried to demonstrate something about one way to do it.

Have a nice week-end