- Email: [email protected]
EP-1629: Lung tumor tracking using CBCT-based respiratory motion models driven by external surrogates
S882 ESTRO 36 _______________________________________________________________________________________________
Lastly we will present the early follow up data (average follow up 1.5 years) of rate and type of complications observed and contrast it to our non-tracked SBRT population. This will indicate if SBRT tracking does prevent over-radiation of sensitive structures. References: [1] Salter BJ et al., 3D Transperineal Ultrasound Image Guidance Methods for Prostate SBRT Radiotherapy Treatment, Radiotherapy and Oncology, 115, S460. Figure 1: Example of the tracking of one patient in one dimension and the subtraction of table movement. Figure 2: Number of events a corrective positional shift was required during treatment per patient. Each blue dot represents one of the 35 patients. Only 10 patients did not require a corrective action to bring the PTV within tolerance levels (3mm or less).
EP-1628 Analysis of prostate SBRT treatments using 3D transperineal ultrasound image guidance methods M. Szegedi1, C. Boehm1, B. Ager1, V. Sarkar1, P. RassiahSzegedi1, H. Zhao1, L. Huang1, J. Huang1, A. Paxton1, F. Su1, J. Tward1, B. Salter1 1 University of Utah Huntsman Cancer Hospital, Radiation Oncology, Salt Lake City, USA Purpose or Objective A 2nd generation 3D ultrasound image guidance (USIG) system (Clarity, Elekta Inc), that allows for transperineal (TP) localization and intra-fractional tracking of the prostate has been used in SBRT of the prostate at our institution. We have analyzed 35 patients (175 fractions) regarding the localization and tracking performance of our USIG based prostate SBRT protocol. Material and Methods Our clinical workflow for prostate SBRT (5 fractions of 7.25 Gy each) involves setting the patient up based on skin tattoos and using TP localization for image guidance. A trans-abdominal (TA) ultrasound study (BAT, Nomos Inc) is also performed to independently check the patient’s position once TP-USIG-based shifts are applied. A detailed description of our workflow has been presented before [1]. Once the TP-based alignment has been approved by both a physicist and physician with extensive USIG experience, TP-based tracking is initiated. During the treatment, the beam is manually switched off for any migrations greater than 3 mm in any direction. If this migration occurs for more than 5 seconds, the patient’s position is re-adjusted before treatment resumption. For all 175 treatments in the present cohort, the tracking data was analyzed to determine the number of incidents and duration the target’s excursion was greater than 3 mm. Further we evaluate the potential for partial PTV miss, by subtracting couch movement from target movement shown in Figure 1, showing the potential excursion if no corrective action was taken and contrast this with the PTV margins used.Results Figure 2 shows the number of instances where the position of a patient had to be corrected. Only 10 of the 35 patients did not require any corrective action. In two patients (cases 29 and 32), the position had to be corrected more than 20 times over the five fractions. Conclusion With more than 70% of the patients analyzed requiring repositioning, it is clear that intra-fractional tracking should be used when treating with a hypo-fractionated approach, where large excursions should be avoided.
EP-1629 Lung tumor tracking using CBCT-based respiratory motion models driven by external surrogates A. Fassi1, A. Bombardieri1, G.B. Ivaldi2, M. Liotta3, P. Tabarelli de Fatis3, I. Meaglia2, P. Porcu2, M. Riboldi1, G. Baroni1 1 Politecnico di Milano, Dipartimento di Elettronica Informazione e Bioingegneria, Milano, Italy 2 Istituti Clinici Scientifici Maugeri, Radiation Oncology Department, Pavia, Italy 3 Istituti Clinici Scientifici Maugeri, Medical Physics Division, Pavia, Italy Purpose or Objective The aim was to investigate the use of time-resolved (4D) Cone-Beam CT (CBCT) to build a patient-specific respiratory motion model driven by a surface-based breathing surrogate. The proposed approach was applied for the real-time intra-fraction tracking of lung tumors. Material and Methods The study included two lung cancer patients treated with stereotactic body radiotherapy. Two CBCT scans, acquired at the beginning and at the end of the first treatment fraction, were analyzed for each patient. Seven passive markers were positioned on anatomical landmarks of the patients' thoraco-abdominal surface. Markers 3D coordinates were continuously acquired during all CBCT scans through an optical tracking system (SMART-DX 100, BTS Bioengineering), synchronized with the acquisition of CBCT projections. A breathing surrogate was obtained from the trajectory of all surface markers. The external surrogate was used to reconstruct the 4D CBCT using the motion-compensated algorithm [1]. A deformable respiratory motion model [2] was built from the 4D CBCT of the first scan. The breathing phase and amplitude given as input to the motion model were estimated from the external surrogate. The accuracy of the proposed tracking approach was evaluated on both the first and the second CBCT scan, after compensating for baseline shifts. Tumor positions estimated in 3D with the motion model were projected at the corresponding angle and compared to the real tumor trajectory semi-automatically identified on CBCT projections. [1] Rit S et al, Med Phys 2009;36:2283-96. [2] Fassi A et al, Phys Med Biol 2015;60:1565-82.
S883 ESTRO 36 _______________________________________________________________________________________________
Results Tumor visibility on CBCT images was limited to a small range of projection angles, corresponding to about 5 seconds of each CBCT scan. Table 1 reports the tracking errors measured along the vertical image dimension, i.e. the projection of the superior-inferior tumor motion, and along the horizontal image dimension, combining anteroposterior and medio-lateral tumor motion. The absolute tracking error, averaged over all CBCT scans of all patients, measured 1.0 ± 0.9 mm and 0.8 ± 0.7 mm for the horizontal and vertical image dimensions, respectively. The comparison of real and estimated tumor trajectories along the vertical dimension is depicted in Figure 1. A significant correlation (p-value < 0.005) was found between real and estimated tumor motion, with correlation coefficients higher than 0.75 and 0.69 for the horizontal and vertical dimensions, respectively. Conclusion We developed and tested a novel approach for intrafraction lung tumors tracking, using CBCT-based motion models driven by an external breathing surrogate obtained from non-invasive optical surface imaging. Compared to CT-based motion models, the proposed method does not need to compensate for inter-fraction motion variations that can occur between planning and treatment phases. The validation of the proposed approach on a wider patient population is currently ongoing. EP-1630 The impact of DIBH on dose to the junction in loco-regionally treated left-sided breast patients M. Van Hinsberg1, I. De Bree2, E. Osté2, D. De Ronde2 1 Zuidwest Radiotherapeutisch Instituut, Klinische Fysica, Vlissingen, The Netherlands 2 Zuidwest Radiotherapeutisch Instituut, RTT, Vlissingen, The Netherlands Purpose or Objective Irradiating loco-regional left sided breast cancer patients using a field matching technique is common practice in clinical routine. Possible under- and overdosages at the junction are normally reduced by the natural breathing of the patient. Adding a (deep inspiration) breathhold strategy introduces extra risks of under- and overdosage in this region. A modified conventional treatment technique for the irradiation of loco-regional left-sided breast patients combined with a voluntary DIBH technique has been introduced in our institute. The objective of this pilot is
to validate that this technique is safe, robust and well tolerated by the patient. Material and Methods The standard conventional irradiation technique is adapted to become less sensitive to different intrafractional DIBH levels. A less steep penumbra at the junction is obtained by creating several subbeams with different MLC-positions. In order to reduce the number of beams, a selection of these beams are grouped to one beam using the Field in Field (FiF) forward IMRT functionality of Varian Eclipse TM TPS. This resulted in treatment plans with 6 to 9 beams. The study consists of the following steps: 1 The field match is validated using film measurements in a phantom. 2 Dose distributions and relative movements for 6 patients were assessed in vivo. For this a strip of film (with 6 mm buildup) was placed on the patient skin at the junction. In addition markers were placed on the skin above, on and below the junction, next to the film. Relative movements of the markers in each DIBH were determined using MVimages. 3 The relative movements of the marker measured in each DIBH were used to position the beams in the corresponding location in the TPS, resulting in a realistic dose distribution for each fraction and for the total treatment. 4 The in-vivo film measurements were used as a validation for the skin dose distribution calculated as explained in 3. Results The measured relative marker positions of all patients are shown in fig.1.
Fig. 1 Marker positions relative to the fraction mean. Distribution mean 0.4 mm; sd 2mm; range; -7 to 8 mm. A negative sign means the patient has moved in caudal direction. Analysing the marker positions of each patient shows no trend between the DIBH level of the first and last beam. No hyperventilating nor exhaustion occurred, indicating that DIBH with 6 to 9 beams is well tolerated. Planned dose-values and actual, total dose values for the CTV are presented in table 1:
Lastly we will present the early follow up data (average follow up 1.5 years) of rate and type of complications observed and contrast it to our non-tracked SBRT population. This will indicate if SBRT tracking does prevent over-radiation of sensitive structures. References: [1] Salter BJ et al., 3D Transperineal Ultrasound Image Guidance Methods for Prostate SBRT Radiotherapy Treatment, Radiotherapy and Oncology, 115, S460. Figure 1: Example of the tracking of one patient in one dimension and the subtraction of table movement. Figure 2: Number of events a corrective positional shift was required during treatment per patient. Each blue dot represents one of the 35 patients. Only 10 patients did not require a corrective action to bring the PTV within tolerance levels (3mm or less).
EP-1628 Analysis of prostate SBRT treatments using 3D transperineal ultrasound image guidance methods M. Szegedi1, C. Boehm1, B. Ager1, V. Sarkar1, P. RassiahSzegedi1, H. Zhao1, L. Huang1, J. Huang1, A. Paxton1, F. Su1, J. Tward1, B. Salter1 1 University of Utah Huntsman Cancer Hospital, Radiation Oncology, Salt Lake City, USA Purpose or Objective A 2nd generation 3D ultrasound image guidance (USIG) system (Clarity, Elekta Inc), that allows for transperineal (TP) localization and intra-fractional tracking of the prostate has been used in SBRT of the prostate at our institution. We have analyzed 35 patients (175 fractions) regarding the localization and tracking performance of our USIG based prostate SBRT protocol. Material and Methods Our clinical workflow for prostate SBRT (5 fractions of 7.25 Gy each) involves setting the patient up based on skin tattoos and using TP localization for image guidance. A trans-abdominal (TA) ultrasound study (BAT, Nomos Inc) is also performed to independently check the patient’s position once TP-USIG-based shifts are applied. A detailed description of our workflow has been presented before [1]. Once the TP-based alignment has been approved by both a physicist and physician with extensive USIG experience, TP-based tracking is initiated. During the treatment, the beam is manually switched off for any migrations greater than 3 mm in any direction. If this migration occurs for more than 5 seconds, the patient’s position is re-adjusted before treatment resumption. For all 175 treatments in the present cohort, the tracking data was analyzed to determine the number of incidents and duration the target’s excursion was greater than 3 mm. Further we evaluate the potential for partial PTV miss, by subtracting couch movement from target movement shown in Figure 1, showing the potential excursion if no corrective action was taken and contrast this with the PTV margins used.Results Figure 2 shows the number of instances where the position of a patient had to be corrected. Only 10 of the 35 patients did not require any corrective action. In two patients (cases 29 and 32), the position had to be corrected more than 20 times over the five fractions. Conclusion With more than 70% of the patients analyzed requiring repositioning, it is clear that intra-fractional tracking should be used when treating with a hypo-fractionated approach, where large excursions should be avoided.
EP-1629 Lung tumor tracking using CBCT-based respiratory motion models driven by external surrogates A. Fassi1, A. Bombardieri1, G.B. Ivaldi2, M. Liotta3, P. Tabarelli de Fatis3, I. Meaglia2, P. Porcu2, M. Riboldi1, G. Baroni1 1 Politecnico di Milano, Dipartimento di Elettronica Informazione e Bioingegneria, Milano, Italy 2 Istituti Clinici Scientifici Maugeri, Radiation Oncology Department, Pavia, Italy 3 Istituti Clinici Scientifici Maugeri, Medical Physics Division, Pavia, Italy Purpose or Objective The aim was to investigate the use of time-resolved (4D) Cone-Beam CT (CBCT) to build a patient-specific respiratory motion model driven by a surface-based breathing surrogate. The proposed approach was applied for the real-time intra-fraction tracking of lung tumors. Material and Methods The study included two lung cancer patients treated with stereotactic body radiotherapy. Two CBCT scans, acquired at the beginning and at the end of the first treatment fraction, were analyzed for each patient. Seven passive markers were positioned on anatomical landmarks of the patients' thoraco-abdominal surface. Markers 3D coordinates were continuously acquired during all CBCT scans through an optical tracking system (SMART-DX 100, BTS Bioengineering), synchronized with the acquisition of CBCT projections. A breathing surrogate was obtained from the trajectory of all surface markers. The external surrogate was used to reconstruct the 4D CBCT using the motion-compensated algorithm [1]. A deformable respiratory motion model [2] was built from the 4D CBCT of the first scan. The breathing phase and amplitude given as input to the motion model were estimated from the external surrogate. The accuracy of the proposed tracking approach was evaluated on both the first and the second CBCT scan, after compensating for baseline shifts. Tumor positions estimated in 3D with the motion model were projected at the corresponding angle and compared to the real tumor trajectory semi-automatically identified on CBCT projections. [1] Rit S et al, Med Phys 2009;36:2283-96. [2] Fassi A et al, Phys Med Biol 2015;60:1565-82.
S883 ESTRO 36 _______________________________________________________________________________________________
Results Tumor visibility on CBCT images was limited to a small range of projection angles, corresponding to about 5 seconds of each CBCT scan. Table 1 reports the tracking errors measured along the vertical image dimension, i.e. the projection of the superior-inferior tumor motion, and along the horizontal image dimension, combining anteroposterior and medio-lateral tumor motion. The absolute tracking error, averaged over all CBCT scans of all patients, measured 1.0 ± 0.9 mm and 0.8 ± 0.7 mm for the horizontal and vertical image dimensions, respectively. The comparison of real and estimated tumor trajectories along the vertical dimension is depicted in Figure 1. A significant correlation (p-value < 0.005) was found between real and estimated tumor motion, with correlation coefficients higher than 0.75 and 0.69 for the horizontal and vertical dimensions, respectively. Conclusion We developed and tested a novel approach for intrafraction lung tumors tracking, using CBCT-based motion models driven by an external breathing surrogate obtained from non-invasive optical surface imaging. Compared to CT-based motion models, the proposed method does not need to compensate for inter-fraction motion variations that can occur between planning and treatment phases. The validation of the proposed approach on a wider patient population is currently ongoing. EP-1630 The impact of DIBH on dose to the junction in loco-regionally treated left-sided breast patients M. Van Hinsberg1, I. De Bree2, E. Osté2, D. De Ronde2 1 Zuidwest Radiotherapeutisch Instituut, Klinische Fysica, Vlissingen, The Netherlands 2 Zuidwest Radiotherapeutisch Instituut, RTT, Vlissingen, The Netherlands Purpose or Objective Irradiating loco-regional left sided breast cancer patients using a field matching technique is common practice in clinical routine. Possible under- and overdosages at the junction are normally reduced by the natural breathing of the patient. Adding a (deep inspiration) breathhold strategy introduces extra risks of under- and overdosage in this region. A modified conventional treatment technique for the irradiation of loco-regional left-sided breast patients combined with a voluntary DIBH technique has been introduced in our institute. The objective of this pilot is
to validate that this technique is safe, robust and well tolerated by the patient. Material and Methods The standard conventional irradiation technique is adapted to become less sensitive to different intrafractional DIBH levels. A less steep penumbra at the junction is obtained by creating several subbeams with different MLC-positions. In order to reduce the number of beams, a selection of these beams are grouped to one beam using the Field in Field (FiF) forward IMRT functionality of Varian Eclipse TM TPS. This resulted in treatment plans with 6 to 9 beams. The study consists of the following steps: 1 The field match is validated using film measurements in a phantom. 2 Dose distributions and relative movements for 6 patients were assessed in vivo. For this a strip of film (with 6 mm buildup) was placed on the patient skin at the junction. In addition markers were placed on the skin above, on and below the junction, next to the film. Relative movements of the markers in each DIBH were determined using MVimages. 3 The relative movements of the marker measured in each DIBH were used to position the beams in the corresponding location in the TPS, resulting in a realistic dose distribution for each fraction and for the total treatment. 4 The in-vivo film measurements were used as a validation for the skin dose distribution calculated as explained in 3. Results The measured relative marker positions of all patients are shown in fig.1.
Fig. 1 Marker positions relative to the fraction mean. Distribution mean 0.4 mm; sd 2mm; range; -7 to 8 mm. A negative sign means the patient has moved in caudal direction. Analysing the marker positions of each patient shows no trend between the DIBH level of the first and last beam. No hyperventilating nor exhaustion occurred, indicating that DIBH with 6 to 9 beams is well tolerated. Planned dose-values and actual, total dose values for the CTV are presented in table 1: