Bootstrap Framework 3.3.6

[marko_laaksomaa]

Hi

At the moment we acquire daily tangential and weekly additional AP+LAT images in DIBH n+ and mastectomy cases with AlignRT, usually there is no surprises due to exceeding of IGRT tolerances in the weekly AP+LAT images and we can mostly continue with tangential images for a week. In tangential images we have (together with ribs+ breast) also vertebra and shoulder joint daily visible so we can approximately daily evaluate the accuracy for the lymph node area and BHL. In those images pretty typical situation seems to be such that there is -0.1 - -0.4 cm couch shift needed for the vertebra in LNG (vertebra locates too cranial in the first place in comparison to reference image), but ribs LNG is ok and thereby there is no need to apply the couch shifts. At the treatment, some of those patients takes BH in the latest treatment fields some mm too much towards cranial direction in LNG based on AlignRT deltas, and based on findings earlier in the published study with RPM. As a conclusion based on findings in practice, for some DIBH patients there are situations in practice that we should leave BH LNG delta 1-3 mm too caudally from zero at the BH setup to optimize the treatment accuracy (then also vertebra LNG would be in correct location in the first place) and to eliminate the risk that patient exceeds the LNG threshold (4 mm) with taking the BHs too cranial at the end of the fraction. The reason for this may be that the patient has possibly taken 3-5 BHs  more at the CT just before CT was acquired than she takes at the setup process. Therefore at the setup we can not copy that "pitch" which appears only after patient has taken several BHs. The other reason (for other patients) for the vertebra displacement in LNG may be that BHL in the images is 1-3 mm too shallow in the first place, when vertebra in LNG usually remains at least as much too cranial in the images, but that is another story already. One thing that I have noticed is that if we leave 1° roll and 2 mm LAT error to the "same" direction at the BH setup with AlignRT, there can be 5 mm displacement in the images in LAT. These are such some may think insignificant things that may be workflow related issues and things that you do not have, but I tried to bring something to think about here to the forum again from the practice.

Have a nice week

Marko

[marko_laaksomaa]

... lately we have had couple mastectomy DIBH patients, where there has been needs to correct systematic error in the arm position. I give you one example about the arm position correction. With the FB and BH reference surfaces we have had everything in good accuracy in the images (vertebra, sternum, BHL, ribs in all directions), but the shoulder joint has been systematically daily 6-9 mm too caudally. At the setup we now positioned the arm more cranially in comparison to earlier FB reference surface (with treatment capture) and after all the FB deltas were 0, we took a new FB reference surface as a possible candidate for the upcoming setups (arm now more cranially). In the images there was 6 mm LNG error in the isocenter, but the arm position was ok (in comparison to th1 also). 6 mm couch shift was done in LNG, but we did not acquire new reference surfaces after couch shifts, because the bolus was used and setup is done without bolus. On the next day we knew that the patient posture (also arm) is good with the FB and BH surfaces that we acquired yesterday, but the LNG isocenter is not. So we did setup normally with the yesterdays FB and BH surfaces, but after that shifted the couch such that LNG delta was 6 mm out of zero in FB reference surface (shifted the couch based on that isocenter error that we had yesterday), took a new FB reference surface in that location and asked patient to take BH. When the LNG delta went that 6 mm out of zero again in yesterdays BH reference surface (as it should go), we took a new BH reference surface. In the images everything was ok!

What I want say with this is that correct arm position can individually be very important also when we want to optimize LNG isocenter accuracy in mastectomy DIBH with AlignRT. Arm position affected on isocenter even if the ROI was created such that it mostly covers the chest wall and not the surface structures around the shoulder. Interesting! It is much possible that with some of our other patient story is different. Individual adjustment and succeeding is what we have experienced anyway again with AlignRT.

BR Marko

[marko_laaksomaa]

…Some additional thoughts about DIBH LNG. One thing that we used to do earlier with AlignRT DIBH was to do small couch adjustments in LNG (and LAT) at time patient takes a BH at the setup to reach zero errors in deltas in BH. This workflow includes a risk that we shift vertebra out of its correct location in LNG if we trust the patient and her ability to reproduce the daily BH at the setup. One improvement that we have done in n+ and mastectomy DIBH to keep the eye on the vertebra accuracy in LNG (and thereby going to more safely to tangential images only + weekly AP+LAT workflow) has been to write down the BH amplitude values on the paper at the treatment room. We have a list of DIBH patients inside treatment room, to know the correct chest wall movement in VRT+LNG in DIBH. Then at the time we setup the patient on the first fractions, we write down both vertical and longitudinal delta movement between FB and BH surfaces from the fractions when images have been ok (BHL, isocenter etc.). This LNG value on the list tries to demonstrate the correct vertebrae location in LNG at the setup. At the setup, to optimize accuracy in the vertical direction, we use the planned couch vertical (vertebrae matched couch value) to keep the vertebra location daily correct and we use as well AlignRT FB surface vertical (zero error), to keep the BH amplitude correct. If there is a need to acquire new reference surfaces for the setup: the more patient has taken BHs before we capture the references for the setup, the more there is a risk that those reference surfaces will fail for the next day setup (in the images). These are some of improvements, which have led to clear improvements in the AlignRT DIBH- setup accuracy.

If we do small couch shifts based on how patient takes a BH in LNG without understanding what we do, there is a risk that we will shift the vertebra out of its correct location in that direction. For the chest wall accuracy this thing has not much to do with. To say it exaggerated, it seems possible to accidentally setup DIBH patient in FB with BH field, all the BH field deltas inside thresholds and in the tangential image MID_PTV chest wall accuracy may seem good enough, if vertebrae is not visible in the images and only lateral part of the ribs and breast is visible. This chest wall accuracy is what we have known and noticed to be excellent in the first place with AlignRT due to predefined BH- reference surface. The challenge in the workflow has been to improve the accuracy to the vertebrae and arm position (lymph node area) with DIBH.

This is an example what sometimes happens with DIBH: we have reference BH amplitude at the list (chest wall delta movement between FB and BH based on the accurate reference surfaces, which have led to accurate posture and isocenter position in the images) in VRT -1.0 cm and in LNG 1.0 cm. Next we setup patient to FB with the AlignRT accuracy of 1mm/1° and with plan couch vertical. Then we start monitoring BH field and for reason or other deltas are now  -1.0 cm in vertical (as it should be) and 0.5 cm in LNG (as it should not be). What we nowadays do is that we manually shift the couch/patient in LNG 0.5 cm caudally (to reach 1.0 cm LNG delta value at the BH field) before patient takes any BHs (now on the background FB isocenter goes with 5 mm offset in LNG). Usually with this workflow deltas goes to zero at the time patient takes a BH and vertebra locates in the images with better accuracy than without the shift based on the list. If we do not do this small couch shift in LNG to correspond the value at the list before patient takes the first BH at the setup, it is much possible that she will do BH similarly anyway and such that we will finally shift the couch that same amount (5 mm) in LNG, patient being in BH (which was the case earlier). Anyway, we should be somehow wiser than the patient and not let her do anything she wants in LNG and follow that with the couch shifts at the BH setup with a risk that we will shift the vertebrae out of its correct location in LNG.

Some days ago there was a patient, whose BH amplitude at the list was -1 cm VRT and +1 cm LNG. For reason or other she did a breath hold to delta -1 cm LNG (too cranially) starting from those named values. In the images vertebra location was ok in LNG, but sternum location in the images was 1 cm too cranially (as supposed). At the setup we did not trust the patient and her ability to reproduce the BH, we trusted the 1) AlignRT and 2) value on the list. We finally asked patient not to move her chest wall in LNG that much at the BH and she changed the way to take a BH kindly without couch shifts and tangential image was ok. We have been able to acquire kV- tangential images for a year now without problems. We have not used MV tangential images for a year.

Some days ago there was another problematic DIBH patient, which was imaged with tangential image before treatment and there was everything ok in the images (vertebrae and chest wall accuracy good). During the treatment, LNG delta went to -1cm in BH, too cranially, and we imaged tangential image at that point in the middle of fraction. Indeed, there was 0.8 cm need to shift the couch in LNG, based on chest wall match. Since we earlier during the fraction tried to guide the patient how to do BH in LNG direction without any improvements, we decided to shift the couch in LNG, based on chest wall and accepted the 8 mm weaker accuracy in the lymph node area. The rest of the fraction went well with excellent accuracy on the chest wall, but in theory with 8 mm weaker accuracy in the lymph node area.

Finally about the DIBH setup, if we have good FB and BH reference surfaces taken from CT or taken based on image guidance (and plan vertical value), the BH amplitude and thereby vertebrae location should remain daily the same in both vertical and longitudinal directions without the need for any small couch corrections (LNG, LAT) at the FB/BH setup or without any lists, which mostly is the case in our hospital also.

[marko_laaksomaa]

...on last week we had a whole breast (WB) DIBH- patient, which has been imaged with tangential images for while. Vertebra was not visible in those images this time. We noticed that the heart was shifted 3-4 mm towards treatment field, but the ribs and breast location was very accurate without couch corrections. We decided to find out what was the reason for heart displacement.

On the next day we acquired AP+LAT images. On those images we noticed that there was + pitch on the sternum. If we think about the optimal matching location for the tangential image, it is not surprise that the tangential image accuracy has been excellent. There were no need for the couch shift in LAT (ribs ok in LAT), VRT (in this case caudal part of the sternum ok), or LNG (in the midway of sternum and the ribs). Still there was 2-3 degree pitch on the sternum and vertebra located such that there was a need to raise the couch vertical (-3mm/-4 mm). Vertebra located also too cranially (5-7mm).

What we can do now is to acquire new reference surface for the FB setup when couch plan vertical value is +4 mm lower and  patient shifted 5-7 mm caudally on the couch and ask the patient to take a BH to the same BH surface, since it led to the excellent accuracy on the tangential image ribs. Why we want to shift the vertebra caudally and lower the couch?.. to achieve better BHL in the images. The other possibility is to setup the patient with same reference surfaces as previously and do the vertebra match in the online match and shift the couch based on vertebra (lower the couch and shift patient caudally), ask the patient to take a BH to same BH surface and hope that she will take a BH in LNG to zero error in thresholds with the vertebra match, since the BH surface itself was good for the chest wall and breast.

In practice we did not do either correction strategy finally. It was difficult for the patient to take a BH to that BH surface and lowering the couch would lead to more pitched position on the sternum in the practice and looked like that there is risk that patient begins to lift her back, when the result in heart sparing may even get weaker in this case. Since the heart location in the tangential images was inside our IGRT tolerances, we accepted 3-4 mm shift on the heart border towards treatment field finally. It was anyway interesting to see that the reason for that was too shallow BHL and pitch on the sternum in the images.

The other thing that we learned was that since without tangential images and SGRT, some units usually acquire mostly AP+LAT images and matching location in vertical direction is not typically on the caudal part of the sternum as in this case and we/they may at least randomly or even systematically have incorrect matching location if offline match is not daily used. Thereby it can be said that in this kind of individual case SGRT setup with possible tangential image verification can be at least individually quicker and more accurate than the the tangential image accuracy after daily couch corrections based on AP+LAT images in the online match.

VMAT, whole breast + lymph nodes and mastectomy DIBH patients would have needed new planning CT with this kind of displacements in our unit, but since the breast, ribs and heart were inside tolerances we can continue with the tangential images with this WB DIBH patient.

Marko

[marko_laaksomaa]

... Now the case is AlignRT mastectomy DIBH patient with bolus. In the AP+LAT images sternum was systematically 4 mm too dorsally in comparison to reference sternum and vertebra was 5 mm too caudally in comparison to reference image vertebra. In the tangential images chest wall remained 2mm too dorsally, so the total error of the sternum vertical error was not visible on the chest wall in the tangential images, in fact only half of it was. Anyway, the problem in the images is that chest wall is too dorsally, there is 5 mm displacement in lymph node are in LNG, both indicating also too shallow BHL in the images.

We decided to correct this LNG error of the vertebra and too dorsal sternum. With our current workflow there is two ways to do this correction, 1) based on image guidance to acquire new reference surfaces for the setup or 2) evaluate the corrections based on offline review and create the reference surfaces before image guidance. We decided to do the latest, since typically bolus is on the patient skin at the time patient image guidance is performed and new reference surfaces for the next day setup are possible to achieve at the end of the fraction, with doing the "reverse" bolus action in comparison to setup. This is not considered practical, since the posture of the patient may have changed during the fraction.

Therefore we shifted the couch at the setup 4 mm more caudally from the AlignRT FB LNG delta and acquired new reference for the FB setup there. Then we asked patient to take 4 mm more air in vertical direction than on the previous BH reference surface and acquired new BH reference surface there (LNG delta went to zero, as we wanted, since there was no need for corrections in that direction based on chest wall). After these adjustments, posture of the patient was very good in the images and tomorrow we continue with the tangential images. Of course there would have been a way to go with the previous reference surfaces, shift the couch based on vertebra match in caudally in longitudinal direction in the online match and ask patient to take 4 mm more air than previous level was, acquire new surfaces for FB and BH setup, go inside treatment room, ask patient to BH, set the bolus to the just acquired BH surface, change the field to bolus+treatment field, acquire BH reference for the treatment with bolus there, verify the action with tangential image.

With the action we did, BHL is not too shallow anymore, sternum is in its correct location in vertical and there is no displacement in vertebra LNG. Hope these our findings helps someone in the cases of systematic corrections.

Marko

[marko_laaksomaa]

I noticed irritating writing error in my text today January 21, 11:41 AM on the second line. The vertebra has located at the setup images, in acquisition position, systematically too cranially in comparison to vertebra in the reference image, not too caudally as written.

[marko_laaksomaa]

Hi

This is breast+lymph nodes breast DIBH, treated with AlignRT. In the Image 1 accuracy in the tangential image ribs and breast looks good in the image acquisition position. Image 2 is exactly the same image as 1. but now I show you vertebra and shoulder joint visible. There is 7 mm offset in the vertebra and shoulder joint in LNG. Situation was systematic based on Offline review. Since this is lymph+ DIBH, tangential image vertebra should be inside 5 mm in LNG. On the same day, we acquired orthogonal images right after that tangential image to see the reason for LNG offset in vertebra. In the orthogonal Image 3 in the acquisition position, there is 7 mm displacement in the spine and shoulder joint in LNG, just like we noticed from open kV-tangential image. When we match those same orthogonal images based on vertebra (Image 4), we see that sternum is 2 mm too dorsally, indicating slightly too shallow BHL. This is also visible in the Image 5, still same tangential image is now matched on vertebra. Finally, we were not able to totally solve this LNG discrepancy between tangential image ribs and vertebra in the upcoming fractions. We had to create setup reference surfaces which led to location where both vertebra and ribs were daily inside 5 mm in LNG in tangential image, so we created setup surfaces which in practice led to acceptable accuracy in both supraclavicular- and tangential fields.

The purpose of this case is to show, that matching location in orthogonal images which leads to good accuracy in the tangential image may sometimes be surprisingly individual, like in this case and therefore we need a combination of AP+LAT+tangential during the first fractions until the systematic error in the images is solved. On the other hand I wanted to show that relying only on good tangential image ribs accuracy (Image 1) may lead to 7 mm systematic uncertainties in the lymph node area, or would we accept the accuracy in the Image 3? Third thing to say is that with daily open kV-tangential image, used for quick verification only (Image 2), we were able to detect the 7 mm offset in the lymph node (vertebra and shoulder joint) LNG and get even some tips that the BHL is too shallow.

Have a good week at work

Regards Marko