By marko_laaksomaa 
-
-
- Thu Sep 06, 2018 11:50 am
#574
Hi all
I created a small poster kind of letter to this forum about what I have find out from the RTTs view.
Purpose: FB n0- breast patients setup errors were compared between AlignRT (A) n=17 and laser setup (L) n=17, number of orthogonal image pairs 400 (179+221) in total. Methods: The estimated errors were pitch (errors between MID_PTV vertebra and sternum, errors between th1-4 - th10 in vertical), rotation (errors between th1 and th10 in LAT) and arm position (errors between th1 and shoulder joint in LNG and LAT), residual errors for the named structures with A setup only and with daily IGRT, isocenter corrections after A/L- setup and tangential image accuracy (chest wall, breast). For the setup 1 mm and 1 degree accuracy was demanded with A. In group L four tattoos were used for the setup: mid- line tattoos on the sternum and 20 cm caudally, lateral tattoos below breasts on the rigid surface. In total 86 errors were compared in VRT,LNG and LAT directions (43 systematic and 43 random). Opposite T-Roi was used in A including small medial portion of both breast and excluding the diapraghm. Position of the arm was checked and when needed corrected with the aid of treatment capture at the setup process. Results: 23% (20/86) of the estimated errors were better (p<0.05) with A than with L, none of the errors were better with L. From those 20 improved errors, the type of the errors were in 80% of the cases random and in 20% systematic. Random errors for the isocenter corrections in online match were in LNG and LAT directions with A 0.22 cm,0.21 cm and with L setup 0.39 cm,0.34 cm. Residual errors with only A setup, without IGRT, exceeded 5mm in 14% of the fractions in th1 area, 7 mm in shoulder joint area was exceeded in 13% of the fractions. 1% of the MID_PTV ribs in tangential images were out of tolerance 5mm in LNG and 4 mm in LAT/VRT. Conclusions: Possibility that only tangential image (without the aid of orthogonal image match) is daily in tolerances is much better with A setup than with L setup due to smaller isocenter variation. Also random errors in the patient posture changes were found partly lower with A in the investigated groups. To minimize systematic posture errors with A, offline review is recommended and permanent reference surface for the setup should not be taken at the fraction there is displacement errors of the structures in the images. AlignRT is more reproducible method to setup FB n0 patients than conventional tattoo based setup. For the A breast patients treated with lymph nodes at our unit, weekly IGRT with the thresholds is needed to verify/correct the combined effect of the isocenter- and displacement errors of the bony structures related to lymph nodes.
Sometimes earlier we discussed about pitch, which causes residual errors for the spine in LNG direction. As a free discussion I could say that I have noticed three different kind of pitch in the images with and without AlignRT. First one is seen in the vertebra. There is displacement error between cranial and caudal part of spine in LAT- image in vertical. This seems to be possible to correct with AlignRT- setup such that patient moves in CC direction and mostly AlignRT notices this kind of pitch in the sternum, as someone here mentioned earlier. The other case is such that there is not displacement error in the vertebra in LAT kV- image, but there is pitch in the sternum which is seen also in AlignRT. This has something to do with the different relaxation between CT and setup and correction may be difficult. The third case is that entire chest wall and the arm is shifted in CC without numerical or visual pitch in the AlignRT. Fixation seems important in this as mentioned here. The correction strategy usually needs offline review evaluation during the first fractions. If in theory 2 out of 3 error types in pitch is visible with AlignRT, I agree that also this is as an advantage for SGRT. The realized spine accuracy in LNG is what I am interested in other units results with AlignRT setup. In n0 breast this knowledge may have negligible significance, but in n+ treatments we consider this important as spine and arm position correlates with the lymph nodes. For comparison with us, I challenge you to offline match the th1-4 (and shoulder joint) from about 20 patients and add those results in Excell to existing isocenter correction-results to find out the local accuracy of the structure with AlignRT setup (without IGRT). It is all done in one day.
marko.laaksomaa@pshp.fi
Best regards Marko
I created a small poster kind of letter to this forum about what I have find out from the RTTs view.
Purpose: FB n0- breast patients setup errors were compared between AlignRT (A) n=17 and laser setup (L) n=17, number of orthogonal image pairs 400 (179+221) in total. Methods: The estimated errors were pitch (errors between MID_PTV vertebra and sternum, errors between th1-4 - th10 in vertical), rotation (errors between th1 and th10 in LAT) and arm position (errors between th1 and shoulder joint in LNG and LAT), residual errors for the named structures with A setup only and with daily IGRT, isocenter corrections after A/L- setup and tangential image accuracy (chest wall, breast). For the setup 1 mm and 1 degree accuracy was demanded with A. In group L four tattoos were used for the setup: mid- line tattoos on the sternum and 20 cm caudally, lateral tattoos below breasts on the rigid surface. In total 86 errors were compared in VRT,LNG and LAT directions (43 systematic and 43 random). Opposite T-Roi was used in A including small medial portion of both breast and excluding the diapraghm. Position of the arm was checked and when needed corrected with the aid of treatment capture at the setup process. Results: 23% (20/86) of the estimated errors were better (p<0.05) with A than with L, none of the errors were better with L. From those 20 improved errors, the type of the errors were in 80% of the cases random and in 20% systematic. Random errors for the isocenter corrections in online match were in LNG and LAT directions with A 0.22 cm,0.21 cm and with L setup 0.39 cm,0.34 cm. Residual errors with only A setup, without IGRT, exceeded 5mm in 14% of the fractions in th1 area, 7 mm in shoulder joint area was exceeded in 13% of the fractions. 1% of the MID_PTV ribs in tangential images were out of tolerance 5mm in LNG and 4 mm in LAT/VRT. Conclusions: Possibility that only tangential image (without the aid of orthogonal image match) is daily in tolerances is much better with A setup than with L setup due to smaller isocenter variation. Also random errors in the patient posture changes were found partly lower with A in the investigated groups. To minimize systematic posture errors with A, offline review is recommended and permanent reference surface for the setup should not be taken at the fraction there is displacement errors of the structures in the images. AlignRT is more reproducible method to setup FB n0 patients than conventional tattoo based setup. For the A breast patients treated with lymph nodes at our unit, weekly IGRT with the thresholds is needed to verify/correct the combined effect of the isocenter- and displacement errors of the bony structures related to lymph nodes.
Sometimes earlier we discussed about pitch, which causes residual errors for the spine in LNG direction. As a free discussion I could say that I have noticed three different kind of pitch in the images with and without AlignRT. First one is seen in the vertebra. There is displacement error between cranial and caudal part of spine in LAT- image in vertical. This seems to be possible to correct with AlignRT- setup such that patient moves in CC direction and mostly AlignRT notices this kind of pitch in the sternum, as someone here mentioned earlier. The other case is such that there is not displacement error in the vertebra in LAT kV- image, but there is pitch in the sternum which is seen also in AlignRT. This has something to do with the different relaxation between CT and setup and correction may be difficult. The third case is that entire chest wall and the arm is shifted in CC without numerical or visual pitch in the AlignRT. Fixation seems important in this as mentioned here. The correction strategy usually needs offline review evaluation during the first fractions. If in theory 2 out of 3 error types in pitch is visible with AlignRT, I agree that also this is as an advantage for SGRT. The realized spine accuracy in LNG is what I am interested in other units results with AlignRT setup. In n0 breast this knowledge may have negligible significance, but in n+ treatments we consider this important as spine and arm position correlates with the lymph nodes. For comparison with us, I challenge you to offline match the th1-4 (and shoulder joint) from about 20 patients and add those results in Excell to existing isocenter correction-results to find out the local accuracy of the structure with AlignRT setup (without IGRT). It is all done in one day.
marko.laaksomaa@pshp.fi
Best regards Marko