- Email: [email protected]
Virtual simulation in patients with breast cancer
Radiotherapy and Oncology 59 (2001) 267±272
www.elsevier.com/locate/radonline
Virtual simulation in patients with breast cancer Andre Buchali a,*, Dirk Geismar b, Margit Hinkelbein b, Lorenz Schlenger b, Kathleen Zinner b, Volker Budach b b
a Klinik fuÈr Strahlentherapie, Ruppiner Kliniken GmbH, Neuruppin, Germany Klinik und Poliklinik fuÈr Strahlentherapie, UniversitaÈtsklinikum ChariteÂ, Campus Berlin-Mitte, Berlin, Germany
Received 11 July 2000; received in revised form 5 February 2001; accepted 8 February 2001
Abstract Background: Investigation of the feasibility and effectiveness of virtual simulation in patients receiving radiotherapy of the breast. Methods: Twenty-three patients were included in the study. All of them underwent a course of postoperative tangential breast irradiation. The patients were prospectively randomised into two groups. Group A patients (n 11) received a conventional computed tomography based treatment planning, group B patients (n 12) a virtual simulation. The results of both treatment planning procedures were compared. Results: The treatment planning was feasible in all patients. The time expenditure could be reduced from a median of 45.0 to 16.5 min and from 55.0 to 32.0 min for the technician and physician, respectively, using virtual simulation. Furthermore the treatment planning for the patient could be reduced from a median of 45.0 min in two sessions to 16.5 min in one session. The image quality of the digital reconstructed radiographs was satisfying compared to the simulation ®lms. The incidence and extension of set-up corrections for the patients at the ®rst treatment were comparable in both groups. The time interval between the planning CT and the ®rst treatment could be reduced by 31% using virtual simulation due to the omission of the conventional simulation. Conclusion: The virtual simulation is a feasible tool for the treatment planning of patients undergoing tangential irradiation of the breast. Compared with the conventional simulation procedure virtual simulation is superior regarding to the precision of patients marking, the quality of the reference images and, the time expenditure for the patients and medical staff. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Radiotherapy; Treatment planning; Breast; Virtual simulation
1. Introduction Virtual simulation is a modi®cation of radiation treatment planning, which allows omitting the simulator. The term was ®rst described by Sherouse et al. 1990 [24]. It represents a process of radiation treatment planning which includes the de®nition of the target volumes according to ICRU50 [11,12], the organs at risk, an isocentre, the size, shape and arrangement of the beams, the documentation of the treatment parameters, the generation of digital reconstructed radiographs (DRR's) and the patients marking. This process is based on the bilateral communication between a CT scanner with a laser positioning system, a workstation and a treatment planning system. The presence of the patient is not needed for virtual simulation after acquisition of the volumetric CT data set and isocentre marking on the patients skin [2].The aim of this investigation was to test the feasibility of virtual simulation in patients receiving radiotherapy of the breast. The main subjects to investigate * Corresponding author.
were: (1) the treatment planning using the isocenter, de®ned immediately after CT scan, without shift, (2) the comparison of the DRR's with the simulation ®lms, (3) the practicability of the avoidance of conventional simulation, (4) the comparison of set-up corrections at the ®rst treatment fraction despite the respiratory movements of the patients and (5) the comparison of the time expenditures for the patients and medical staff. 2. Material and methods Twenty-three patients were included in the investigation. Nineteen patients suffered from invasive breast cancer after conserving therapy (n 17) or radical mastectomy (n 2). Four patients suffered from ductal carcinoma in situ (DCIS) after breast conserving therapy. All patients included in this investigation received a postoperative tangential breast or chest wall irradiation. They all had a low risk for a regional recurrence and received breast or chest wall irradiation alone without identi®cation of lymphatic targets.
0167-8140/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0167-814 0(01)00322-X
268
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
The patients were irradiated in supine position on a head support. A cranial arm holder was used for a stable and comfortable arm positioning. For the planning CT study, a spiral CT (PQ 2000, Picker International, St. David, PA) was used to generate the scans with a slice thickness of 5 mm covering a volume from the upper rim of the head of the humerus to 3 cm below the lower rim of the breast using CTparameters of 130 kV and 125 mA. For all patients the clinical and planning target volumes as well as the ipsilateral lung and in left sided tumours the heart were outlined using the AcQSime (Picker International, St. David, PA) software. The extension of the Clinical Target Volume (CTV) was calculated according to the visible breast tissue in the axial CT slices and to the outer breast contour. For the Planning Target Volume (PTV) a margin of 8 mm was added. The ipsilateral lung was contoured automatically using a threshold technique (the threshold for the lung tissue was 2300 HU).The data were transferred to the planning system CADPLANe (Varian-Dosetek) for 3-D treatment planning in all patients. An experienced physicist generated all plans and calculated their dose distributions. A standard technique was used with two tangentially, collimated, wedged ®elds for 6 MVphotons. The patients were randomised according to their month of birth into two planning procedures. Group A and B patients were either conventionally (n 11) or virtually (n 12) planned. Group A patients did not got any marking at the skin after planning CT scan. After contouring and physical treatment planning a conventional simulation was performed to verify the isocenter and the beams of the treatment plan (Fig. 1). The superior-inferior position of the isocenter was found using the a±p and lateral topograms of the planning CT scan, showing the position of the individual axial CT slices. The a±p and lateral isocenter positions were found according to bony landmarks like the vertebral bodies, the chest wall and the sternum. The precise isocenter de®nition was controlled using the comparison of the treatment ®elds with the beamseye-views. Group B patients remained positioned after acquisition of the CT data set. The data were immediately transferred to
the workstation VoxelQ (Picker International, St. David, PA). The isocentre was de®ned using the AcQSime software in the centre according to the cranio-caudal extension of the outer breast contour and nearly the chest wall. The position of the CT couch was changed according to the calculated superior-inferior, right-left and anterior-posterior translations and the isocentre was marked at the patients skin using a laser system [2]. Additionally the patients were marked at the chest wall inferior of the breast to increase the reproducibility of the positioning. After physical treatment planning the plan was retransferred to the workstation VoxelQ and DRR's of the irradiation ®elds were reconstructed using the AcQSime software (Fig. 1). We used a standard window with a level of 150 HU and a window of 600 HU, followed by an individual ®ne tuning of the contrast and brightness for the reconstruction of DRR's. The DRR's were used as reference images for the ®rst treatment session. The setup veri®cation at the ®rst treatment was done using an electronic portal imaging device (Varian medical systems). For the determination of the accuracy the distances between the ®eld borders and the chest wall, the inferior and anterior breast contour as well as the angle between the posterior ®eld border and the chest wall were measured. The results of both treatment planning procedures were compared. All values given here are Median values, 25% Quartile and 75% Quartile. For statistical comparison of the time expenditures for the medical staff and patients the t-test for independent samples was used. A P , 0:05 was considered statistically signi®cant. 3. Results Of 23 patients included in this investigation, 11 were conventionally (group A) and 12 virtually (group B) planned. The generation of the volumetric CT data set followed by the transfer to the workstation needed a median time period of 10.0 min. The contouring of the CTV, PTV and ipsilateral lung needed a median of 19.0 min (Fig. 1).
Fig. 1. Flow sheet of virtual and conventional CT-based treatment planning including the duration of the different steps (Median, 25% (Q1/4) and 75% (Q3/4) Quartile in minutes).
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
269
Table 1 Comparison of resulting total time expenditure for the patients and medical staff with virtual and conventional planning procedure a
Number of patients
Total duration for the Physician Technician Patient Whole planning procedure a
Conventional simulation
Virtual simulation
11
12
P-value
Median (min)
Q1/4: Q3/4 (min)
Median (min)
Q1/4; Q3/4 (min)
55.0 45.0 45.0 65.0
47.0; 61.5 40.0; 56.5 40.0; 56.5 58.0; 78.0
32.0 16.5 16.5 43.0
30.0; 37.5 13.8; 20.3 13.8; 20.3 38.3; 47.5
,0.005 ,0.001 ,0.001 ,0.001
Median, 25% (Q1/4) and 75% (Q3/4) Quartile, t-test for independent samples.
In group A, a conventional simulation was performed after CT-guided 3-D treatment planning. The median duration was 35.0 min for the patient, a physician and a technician (Fig. 1). In three patients differences up to 10 mm were observed between planned and simulated isocentre. For virtually simulated patients (group B) additionally 7.0 min were necessary after the CT-scan for the patient, a physician and a technician for the de®nition of the isocentre and patients marking (Fig. 1). A 3-D treatment plan was generated after contouring using the previously de®ned isocentre. This isocentre could be used in all virtually simulated cases without the necessity for corrections. After transmission of the treatment plan to the workstation, a physician needed a median of 9.5 min for the generation of the DRR's in virtual simulation (Fig. 1).The resulting median time expenditures for the treatment planning were signi®cantly reduced by means of virtual simulation from 45.0 to 16.5 min (P , 0:001, t-test for independent samples) and from 55.0 to 32.0 min (P , 0:005) for the technician and physician, respectively. Similarly the median durations for the patient and the whole planning procedure were reduced from 45.0 to 16.5 min (P , 0:001) and from 65.0 to 43.0 min (P , 0:001), respectively (Table 1). We measured the homogeneities of the dose distributions of the Planning Target Volume as a reference parameter for an acceptable position of the isocentre. The minimum doses were in median 88.4 and 89.8% (no signi®cant) and the maximum dose 105.2 and 106.3% (no signi®cant) with conventional and virtual simulation, respectively. A comparison of a simulation ®lm and a DRR is shown in Figs. 2 and 3. The quality of simulation ®lms and DRR's was scored into adequate, suf®cient or insuf®cient according to the visualisation of the chest wall, the outer breast contour and the central lung distance. Two physicians did scoring of each ®lm independently; the more worsening result of the two physicians was used for the comparison. The simulation ®lms were scored adequate, suf®cient or insuf®cient in 6, 5 and 0 of 11 patients, respectively, with conventional simulation. With virtual simulation, DRR's were evaluated adequate, suf®cient or insuf®cient in 11, 0 and 1 of 12 patients, respectively (no signi®cant). The crucial point
was the overexposed outer breast contour (®ve patients) with conventional simulation. The whole breast was incompletely visualised in one patient after virtual simulation, caused by a caudally insuf®cient extension of the CT scan, which clearly differed from the protocol parameters. Clips within the tumour bed could only be detected in one of ®ve patients with DRR's without contouring of them, whereas they could be detected in all of ®ve patients with conventional simulation. The omission of the conventional simulation for virtually planned patients resulted in a shorter planning-CT to ®rst treatment period, which was reduced by 31% compared with conventional treatment planning procedure. Repeated veri®cations at the simulator after the ®rst treatment setup were not necessary in any patient. An electronic portal imaging device was used for the veri®cation at the ®rst treatment fraction as described above. The rotational error (differences of the angles of the chest wall to the posterior ®eld border) with conventional simulation/DRR and portal ®lms was lower than 18 in all cases. The distances between the inferior breast contour and inferior ®eld border were ,4 mm in all cases and not corrected. An on-line correction of the patients was only necessary according to differences of more than 3 mm of the distance between the chest wall and the posterior ®eld border (central lung distance) in two of 11 patients with conventional and two of 12 patients with virtual simulation, respectively. In additionally three conventionally simulated patients corrections were necessary because of imprecise reference images due to differences between the planned and simulated isocentre. 4. Discussion The virtual simulation is a feasible tool for radiation treatment planning in breast cancer patients. Time-consuming procedures involved herein were the CT scan and the data transfer. The median duration of this treatment planning step for the patients was 17 min. This is comparable with an image acquisition time of ,10 min and duration for the patient of 20 min as described by Mah et al. [16]. Earlier publications showed longer scanning
270
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
Fig. 2. Comparison of a stimulation ®lm of a conventionally and a DRR of another virtually simulated patient. The irradiation technique consisted of two isocentric, tangential, wedged, asymmetric ®elds.
and transfer times due to former CT scanner generations and networks [19,22]. The time expenditures were signi®cantly reduced for the technician and physician with virtual simulation. Furthermore, this method made the time schedule easier for the patients compared with conventional breast treatment planning. Likewise virtual simulation has improved staff ef®ciency and resource utilisation [16,20]. Virtual simulation as a tool for radiation treatment planning can be of particular value for departments with limited staff and ®nancial resources. A distinct variation of the virtual/CT simulation procedure is to mark the whole treatment ®elds at the patients skin, not only an isocentre. Using this procedure, the contouring of the target volumes and the subsequent generation of the treatment plan is necessary immediately after data transfer. This variation requires extended immobilisation periods of the patients (30±90 min) on the CT couch, who are not able to lie such a long time motionless [8,18,20±
Fig. 3. Comparison of a stimulation ®lm of a conventionally and a DRR of another virtually simulated patient. The irradiation technique consisted of two isocentric, tangential, wedged, asymmetric ®elds.
22]. Furthermore, the CT-scanner is not available for other use. We used isocenter skin markers only, which can be calculated within a few minutes after CT scan. The AcQSime system is not capable of treatment edge marking. We investigated the homogeneities of the dose distributions within the PTV as a reference parameter for an acceptable position of the isocentre in virtually planned patients. We did not found differences between the two planning procedures concerning this parameter. The de®nition of the isocentre immediately after CT scan proved as practicable in our investigation and could be used for physical treatment planning in each patient without the need for corrections. The isocenter position proved not as critical for the treatment plan in our investigation. A homogenous dose distribution could be improved by changes of asymmetric collimators, wedges, ®eld weights and gantry and collimator angles in cases with a sub optimal isocenter position. On the other hand, in these cases the physicist could be changed the isocenter. If the true isocenter had been differed
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
from the CT simulation isocenter, the software has been calculated the shift using anterior-posterior, right-left and superior-inferior translations [16]. In comparison with the simulation ®lms, the contrast between chest wall and the outer contour of the breast was lower in DRR's caused on the use of a standard window and an individual ®ne-tuning of the contrast and brightness levels. This leads to an increased discrimination of chest wall structures, the central lung distance and the outer breast contour. The quality of DRR's was more similar to that of electronic portal images in comparison to simulation ®lms. Clips, which were positioned in the tumour bed for the boost irradiation, could not be envisaged in DRR's [3]. In these patients, an additionally contouring of the clips was necessary, which lasts additionally 1±2 min. The resolution of the DRR's is limited by the voxel size of the CT scan and by 512 £ 512 pixels. The slice width and thickness as well as patient's movement during the CT scan procedure further degrades the DRR's [9,17]. The acquisition of the scan volume should be performed without rest [18]; we needed an interval of about 5 s between the acquisitions of the two spirals with a scanning time of approximately 35 s each. The precision of the calculated divergence in the DRR's depends on the slice thickness and was found to be less than 1.0 mm [17].Due to an inadequate generated scanning volume virtual simulation and DRR's were insuf®cient in one patient. The omission of a conventional simulator appointment after planning CT using virtual simulation seems to reduce the delays before treatment. A reduction of the postoperative treatment interval leads to better local tumour control ®gures [6]. Additionally a postoperative interval of less than 6 weeks is often a prerequisite for patients to be eligible for breast cancer trials. From this perspective the reduction of the postoperative interval due to virtual simulation can be of some value. Moreover the schedule for the patient could be done easier using virtual simulation. The initial set-up check of the patients was done using an electronic portal imaging device. Being aware of uncertainties in the judgement of on-line portal ®lms up to 5 mm [10], the set-up of the patients has been corrected according to the simulation results in terms of the central lung distance and breast contour, if any distance difference between simulation and electronic portal image had been more than 3 mm. This was necessary in two virtually and conventionally simulated patients, each. A correction was also needed in three conventionally simulated patients, in whom differences between the planned and simulated isocenter were found. The range was in the order of described simulation errors of 6.4 ^ 3.9 mm (mean and 1 SD) and ^5 mm (in 78% of the patients) in different cancers, respectively [15,23]. These errors can compromise the treatment success, if not detected immediately and can be avoided using virtual simulation only. Virtual simulation has the potential to eliminate the trial and error of conventional simulation [24], but it cannot avoid the variances between the
271
marked and treatment isocenter, which seems to be comparable between the two groups. In contrast to this, the accuracy of patients marking with CT simulation was measured in different settings and found to be between 1 and 2.2 mm [8,17,20±22]. The degree of the corrections at the ®rst irradiation was in the order as others reported for interfractional set-up variations (^4.3 mm (1 SD) and 0.1 ^ 4.8 mm (Mean ^ 1 SD)) [1,14]. The set-up error between simulation, however, and the ®rst treatment session can be larger than interfractional variations [7]. Das et al. found a reduction of the ®rst set-up error after a conventional simulation by introducing a 2nd simulation including a sub sequential correction prior to the ®rst treatment fraction [5]. This corresponds to a reduction of the incorrect simulated isocentre at the 1st simulation as well as to respiratory movements, which have a major impact on the precision of the ®rst set-up. The CT scan represents the momentary respiratory situation of the patient during the scan. It can be vary considerable from the ®rst treatment situation. On the other hand, the spiral CT-scan lasts longer than one respiratory cycle in our investigation. Consequently, the rib cage and breast contour movements were recorded in the volumetric CT data set, producing the DRR's, which can be probably reduced by means of respiratory gating and/or mask ®xations [4,13]. Butker et al. have not found respiratory movements to be a signi®cant obstacle in virtual simulation [3]. The patient's set-up with virtual simulation according to the marked isocentre and additional marks in the lower chest wall region was adequate in this investigation. A CT simulation with subsequent treatment planning leads to patients marks indicating the treatment portals on the body surface. The disadvantages were discussed above. The mark of the whole treatment ®eld with conventional simulation leads not to a reduction of the set-up errors. As other authors have shown, no necessity exists for an additional conventional simulation after virtual simulation for the patients [3,8,25]. Galvin estimated 70% of the patients qualifying for the CT simulation [8]. 5. Conclusion The virtual simulation is a feasible tool for the treatment planning of patients undergoing irradiation of the breast. Compared with the conventional simulation procedure virtual simulation is superior regarding to the precision of patients marking, the quality of the reference images and, the time expenditure for the patients and medical staff. References [1] Boehmer D, Feyer P, Harder C, et al. Veri®cation of set-up deviations in patients with breast cancer using portal imaging in clinical practice. Strahlenther Onkol 174: Suppl. 1998;II:36±39. [2] Buchali A, Dinges S, Koswig S, et al. Virtual Simulation ± ®rst
272
[3] [4]
[5] [6] [7]
[8] [9] [10] [11] [12] [13] [14]
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272 clinical results in patients with prostate cancer. Strahlenther Onkol 1998;174:88±91. Butker EK, Helton DJ, Keller JW, Hughes LL, Crenshaw T, Davis LW. A totally integrated simulation technique for three-®eld breast treatment using a CT simulator. Med Phys 1996;23:1809±1814. Creutzberg CL, Althof VG, Huizenga H, Visser AG, Levendag PC. Quality assurance using portal imaging: the accuracy of patient positioning in irradiation of breast cancer. Int J Radiat Oncol Biol Phys 1993;25:529±539. Das IJ, Cheng CW, Fosmire H, Kase KR, Fitzgerald TJ. Tolerances in setup and dosimetric errors in the radiation treatment of breast cancer. Int J Radiat Oncol Biol Phys 1993;26:883±890. Dubey AK, Recht A, Come S, Shulman L, Harris J. Why and how to combine chemotherapy and radiation therapy in breast cancer patients. Recent Results Cancer Res 1998;152:247±254. Fein DA, McGee KP, Schultheiss TE, Fowble BL, Hanks GE. Intraand interfractional reproducibility of tangential breast ®elds: a prospective on-line portal imaging study. Int J Radiat Oncol Biol Phys 1996;34:733±740. Galvin JM. Is CT simulation the wave of the future? Med Phys 1993;20:1565±1657. Galvin JM, Sims C, Dominiak G, Cooper JS. The use of digitally reconstructed radiographs for three-dimensional treatment planning and CT-simulation. Int J Radiat Oncol Biol Phys 1995;31:935±942. Herman MG, Abrams RA, Mayer RR. Clinical use of on-line portal imaging for daily patient treatment veri®cation. Int J Radiat Oncol Biol Phys 1994;28:1017±1023. Hess C-F, Christ G, Jany R, Bamberg M. Dose speci®cation at the 'ICRU-reference point`: consequences for the clinical practice. Strahlenth Onkol 1993;169:660±667. ICRU-Report No. 50. Prescribing, recording and reporting photon beam therapy. 1992. Kubo HD, Hill BC. Respiration gated radiotherapy treatment: a technical study. Phys Med Biol 1996;41:83±91. Lirette A, Pouliot J, Aubin M, Larochelle M. The role of electronic portal imaging in tangential breast irradiation: a prospective study. Radiother Oncol 1995;37:241±245.
[15] Lohr F, Schramm O, Schraube P, et al. Simulation of 3D-treatment plans in head and neck tumors aided by matching of digitally reconstructed radiographs (DRR) and on-line distortion corrected simulator images. Radiother Oncol 1997;45:199±207. [16] Mah K, Danjoux CE, Manship S, Makhani N, Cardoso M, Sixel KE. Computed tomographic simulation of craniospinal ®elds in pediatric patients: improved treatment accuracy and patient comfort. Int J Radiat Oncol Biol Phys 1998;41:997±1003. [17] McGee KP, Das IJ, Sims C. Evaluation of digitally reconstructed radiographs (DRRs) used for clinical radiotherapy: A phantom study. Med Phys 1995;22:1815±1827. [18] Nishidai T, Nagata Y, Takahashi M, et al. CT simulator: a new 3-D planning and simulating system for radiotherapy: Part 1. Description of system. Int J Radiat Oncol Biol Phys 1990;18:499±504. [19] Perez CA, Purdy JA, Harms W, et al. Design of a fully integrated three-dimensional computed tomography simulator and preliminary clinical evaluation. Int J Radiat Oncol Biol Phys 1994;30:887± 897. [20] Ragan DP, Forman JD, He T, Mesina CF. Clinical results of computerized tomography-based simulation with laser patient marking. Int J Radiat Oncol Biol Phys 1996;34:691±695. [21] Ragan DP, He T, Liu X. Correction for distortion in a beam outline transfer device in radiotherapy CT-based simulation. Med Phys 1993;20:179±185. [22] Ragan DP, He T, Mesina CF, Ratanatharathorn V. CT-based simulation with laser patient marking. Med Phys 1993;20:379±380. [23] Rosenman J, Sailer SL, Sherouse GW, Chaney EL, Tepper JE. Virtual simulation: initial clinical results. Int J Radiat Oncol Biol Phys 1991;20:843±851. [24] Sherouse GW, Bourland JD, Reynolds K, McMurry HL, Mitchell TP, Chaney EL. Virtual simulation in the clinical setting: some practical considerations. Int J Radiat Oncol Biol Phys 1990;19:1059±1065. [25] Valicenti RK, Waterman FM, Corn BW, Curran WJ. A prospective, randomized study addressing the need for physical simulation following virtual simulation. Int J Radiat Oncol Biol Phys 1997;39:1131± 1135.
www.elsevier.com/locate/radonline
Virtual simulation in patients with breast cancer Andre Buchali a,*, Dirk Geismar b, Margit Hinkelbein b, Lorenz Schlenger b, Kathleen Zinner b, Volker Budach b b
a Klinik fuÈr Strahlentherapie, Ruppiner Kliniken GmbH, Neuruppin, Germany Klinik und Poliklinik fuÈr Strahlentherapie, UniversitaÈtsklinikum ChariteÂ, Campus Berlin-Mitte, Berlin, Germany
Received 11 July 2000; received in revised form 5 February 2001; accepted 8 February 2001
Abstract Background: Investigation of the feasibility and effectiveness of virtual simulation in patients receiving radiotherapy of the breast. Methods: Twenty-three patients were included in the study. All of them underwent a course of postoperative tangential breast irradiation. The patients were prospectively randomised into two groups. Group A patients (n 11) received a conventional computed tomography based treatment planning, group B patients (n 12) a virtual simulation. The results of both treatment planning procedures were compared. Results: The treatment planning was feasible in all patients. The time expenditure could be reduced from a median of 45.0 to 16.5 min and from 55.0 to 32.0 min for the technician and physician, respectively, using virtual simulation. Furthermore the treatment planning for the patient could be reduced from a median of 45.0 min in two sessions to 16.5 min in one session. The image quality of the digital reconstructed radiographs was satisfying compared to the simulation ®lms. The incidence and extension of set-up corrections for the patients at the ®rst treatment were comparable in both groups. The time interval between the planning CT and the ®rst treatment could be reduced by 31% using virtual simulation due to the omission of the conventional simulation. Conclusion: The virtual simulation is a feasible tool for the treatment planning of patients undergoing tangential irradiation of the breast. Compared with the conventional simulation procedure virtual simulation is superior regarding to the precision of patients marking, the quality of the reference images and, the time expenditure for the patients and medical staff. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Radiotherapy; Treatment planning; Breast; Virtual simulation
1. Introduction Virtual simulation is a modi®cation of radiation treatment planning, which allows omitting the simulator. The term was ®rst described by Sherouse et al. 1990 [24]. It represents a process of radiation treatment planning which includes the de®nition of the target volumes according to ICRU50 [11,12], the organs at risk, an isocentre, the size, shape and arrangement of the beams, the documentation of the treatment parameters, the generation of digital reconstructed radiographs (DRR's) and the patients marking. This process is based on the bilateral communication between a CT scanner with a laser positioning system, a workstation and a treatment planning system. The presence of the patient is not needed for virtual simulation after acquisition of the volumetric CT data set and isocentre marking on the patients skin [2].The aim of this investigation was to test the feasibility of virtual simulation in patients receiving radiotherapy of the breast. The main subjects to investigate * Corresponding author.
were: (1) the treatment planning using the isocenter, de®ned immediately after CT scan, without shift, (2) the comparison of the DRR's with the simulation ®lms, (3) the practicability of the avoidance of conventional simulation, (4) the comparison of set-up corrections at the ®rst treatment fraction despite the respiratory movements of the patients and (5) the comparison of the time expenditures for the patients and medical staff. 2. Material and methods Twenty-three patients were included in the investigation. Nineteen patients suffered from invasive breast cancer after conserving therapy (n 17) or radical mastectomy (n 2). Four patients suffered from ductal carcinoma in situ (DCIS) after breast conserving therapy. All patients included in this investigation received a postoperative tangential breast or chest wall irradiation. They all had a low risk for a regional recurrence and received breast or chest wall irradiation alone without identi®cation of lymphatic targets.
0167-8140/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0167-814 0(01)00322-X
268
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
The patients were irradiated in supine position on a head support. A cranial arm holder was used for a stable and comfortable arm positioning. For the planning CT study, a spiral CT (PQ 2000, Picker International, St. David, PA) was used to generate the scans with a slice thickness of 5 mm covering a volume from the upper rim of the head of the humerus to 3 cm below the lower rim of the breast using CTparameters of 130 kV and 125 mA. For all patients the clinical and planning target volumes as well as the ipsilateral lung and in left sided tumours the heart were outlined using the AcQSime (Picker International, St. David, PA) software. The extension of the Clinical Target Volume (CTV) was calculated according to the visible breast tissue in the axial CT slices and to the outer breast contour. For the Planning Target Volume (PTV) a margin of 8 mm was added. The ipsilateral lung was contoured automatically using a threshold technique (the threshold for the lung tissue was 2300 HU).The data were transferred to the planning system CADPLANe (Varian-Dosetek) for 3-D treatment planning in all patients. An experienced physicist generated all plans and calculated their dose distributions. A standard technique was used with two tangentially, collimated, wedged ®elds for 6 MVphotons. The patients were randomised according to their month of birth into two planning procedures. Group A and B patients were either conventionally (n 11) or virtually (n 12) planned. Group A patients did not got any marking at the skin after planning CT scan. After contouring and physical treatment planning a conventional simulation was performed to verify the isocenter and the beams of the treatment plan (Fig. 1). The superior-inferior position of the isocenter was found using the a±p and lateral topograms of the planning CT scan, showing the position of the individual axial CT slices. The a±p and lateral isocenter positions were found according to bony landmarks like the vertebral bodies, the chest wall and the sternum. The precise isocenter de®nition was controlled using the comparison of the treatment ®elds with the beamseye-views. Group B patients remained positioned after acquisition of the CT data set. The data were immediately transferred to
the workstation VoxelQ (Picker International, St. David, PA). The isocentre was de®ned using the AcQSime software in the centre according to the cranio-caudal extension of the outer breast contour and nearly the chest wall. The position of the CT couch was changed according to the calculated superior-inferior, right-left and anterior-posterior translations and the isocentre was marked at the patients skin using a laser system [2]. Additionally the patients were marked at the chest wall inferior of the breast to increase the reproducibility of the positioning. After physical treatment planning the plan was retransferred to the workstation VoxelQ and DRR's of the irradiation ®elds were reconstructed using the AcQSime software (Fig. 1). We used a standard window with a level of 150 HU and a window of 600 HU, followed by an individual ®ne tuning of the contrast and brightness for the reconstruction of DRR's. The DRR's were used as reference images for the ®rst treatment session. The setup veri®cation at the ®rst treatment was done using an electronic portal imaging device (Varian medical systems). For the determination of the accuracy the distances between the ®eld borders and the chest wall, the inferior and anterior breast contour as well as the angle between the posterior ®eld border and the chest wall were measured. The results of both treatment planning procedures were compared. All values given here are Median values, 25% Quartile and 75% Quartile. For statistical comparison of the time expenditures for the medical staff and patients the t-test for independent samples was used. A P , 0:05 was considered statistically signi®cant. 3. Results Of 23 patients included in this investigation, 11 were conventionally (group A) and 12 virtually (group B) planned. The generation of the volumetric CT data set followed by the transfer to the workstation needed a median time period of 10.0 min. The contouring of the CTV, PTV and ipsilateral lung needed a median of 19.0 min (Fig. 1).
Fig. 1. Flow sheet of virtual and conventional CT-based treatment planning including the duration of the different steps (Median, 25% (Q1/4) and 75% (Q3/4) Quartile in minutes).
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
269
Table 1 Comparison of resulting total time expenditure for the patients and medical staff with virtual and conventional planning procedure a
Number of patients
Total duration for the Physician Technician Patient Whole planning procedure a
Conventional simulation
Virtual simulation
11
12
P-value
Median (min)
Q1/4: Q3/4 (min)
Median (min)
Q1/4; Q3/4 (min)
55.0 45.0 45.0 65.0
47.0; 61.5 40.0; 56.5 40.0; 56.5 58.0; 78.0
32.0 16.5 16.5 43.0
30.0; 37.5 13.8; 20.3 13.8; 20.3 38.3; 47.5
,0.005 ,0.001 ,0.001 ,0.001
Median, 25% (Q1/4) and 75% (Q3/4) Quartile, t-test for independent samples.
In group A, a conventional simulation was performed after CT-guided 3-D treatment planning. The median duration was 35.0 min for the patient, a physician and a technician (Fig. 1). In three patients differences up to 10 mm were observed between planned and simulated isocentre. For virtually simulated patients (group B) additionally 7.0 min were necessary after the CT-scan for the patient, a physician and a technician for the de®nition of the isocentre and patients marking (Fig. 1). A 3-D treatment plan was generated after contouring using the previously de®ned isocentre. This isocentre could be used in all virtually simulated cases without the necessity for corrections. After transmission of the treatment plan to the workstation, a physician needed a median of 9.5 min for the generation of the DRR's in virtual simulation (Fig. 1).The resulting median time expenditures for the treatment planning were signi®cantly reduced by means of virtual simulation from 45.0 to 16.5 min (P , 0:001, t-test for independent samples) and from 55.0 to 32.0 min (P , 0:005) for the technician and physician, respectively. Similarly the median durations for the patient and the whole planning procedure were reduced from 45.0 to 16.5 min (P , 0:001) and from 65.0 to 43.0 min (P , 0:001), respectively (Table 1). We measured the homogeneities of the dose distributions of the Planning Target Volume as a reference parameter for an acceptable position of the isocentre. The minimum doses were in median 88.4 and 89.8% (no signi®cant) and the maximum dose 105.2 and 106.3% (no signi®cant) with conventional and virtual simulation, respectively. A comparison of a simulation ®lm and a DRR is shown in Figs. 2 and 3. The quality of simulation ®lms and DRR's was scored into adequate, suf®cient or insuf®cient according to the visualisation of the chest wall, the outer breast contour and the central lung distance. Two physicians did scoring of each ®lm independently; the more worsening result of the two physicians was used for the comparison. The simulation ®lms were scored adequate, suf®cient or insuf®cient in 6, 5 and 0 of 11 patients, respectively, with conventional simulation. With virtual simulation, DRR's were evaluated adequate, suf®cient or insuf®cient in 11, 0 and 1 of 12 patients, respectively (no signi®cant). The crucial point
was the overexposed outer breast contour (®ve patients) with conventional simulation. The whole breast was incompletely visualised in one patient after virtual simulation, caused by a caudally insuf®cient extension of the CT scan, which clearly differed from the protocol parameters. Clips within the tumour bed could only be detected in one of ®ve patients with DRR's without contouring of them, whereas they could be detected in all of ®ve patients with conventional simulation. The omission of the conventional simulation for virtually planned patients resulted in a shorter planning-CT to ®rst treatment period, which was reduced by 31% compared with conventional treatment planning procedure. Repeated veri®cations at the simulator after the ®rst treatment setup were not necessary in any patient. An electronic portal imaging device was used for the veri®cation at the ®rst treatment fraction as described above. The rotational error (differences of the angles of the chest wall to the posterior ®eld border) with conventional simulation/DRR and portal ®lms was lower than 18 in all cases. The distances between the inferior breast contour and inferior ®eld border were ,4 mm in all cases and not corrected. An on-line correction of the patients was only necessary according to differences of more than 3 mm of the distance between the chest wall and the posterior ®eld border (central lung distance) in two of 11 patients with conventional and two of 12 patients with virtual simulation, respectively. In additionally three conventionally simulated patients corrections were necessary because of imprecise reference images due to differences between the planned and simulated isocentre. 4. Discussion The virtual simulation is a feasible tool for radiation treatment planning in breast cancer patients. Time-consuming procedures involved herein were the CT scan and the data transfer. The median duration of this treatment planning step for the patients was 17 min. This is comparable with an image acquisition time of ,10 min and duration for the patient of 20 min as described by Mah et al. [16]. Earlier publications showed longer scanning
270
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
Fig. 2. Comparison of a stimulation ®lm of a conventionally and a DRR of another virtually simulated patient. The irradiation technique consisted of two isocentric, tangential, wedged, asymmetric ®elds.
and transfer times due to former CT scanner generations and networks [19,22]. The time expenditures were signi®cantly reduced for the technician and physician with virtual simulation. Furthermore, this method made the time schedule easier for the patients compared with conventional breast treatment planning. Likewise virtual simulation has improved staff ef®ciency and resource utilisation [16,20]. Virtual simulation as a tool for radiation treatment planning can be of particular value for departments with limited staff and ®nancial resources. A distinct variation of the virtual/CT simulation procedure is to mark the whole treatment ®elds at the patients skin, not only an isocentre. Using this procedure, the contouring of the target volumes and the subsequent generation of the treatment plan is necessary immediately after data transfer. This variation requires extended immobilisation periods of the patients (30±90 min) on the CT couch, who are not able to lie such a long time motionless [8,18,20±
Fig. 3. Comparison of a stimulation ®lm of a conventionally and a DRR of another virtually simulated patient. The irradiation technique consisted of two isocentric, tangential, wedged, asymmetric ®elds.
22]. Furthermore, the CT-scanner is not available for other use. We used isocenter skin markers only, which can be calculated within a few minutes after CT scan. The AcQSime system is not capable of treatment edge marking. We investigated the homogeneities of the dose distributions within the PTV as a reference parameter for an acceptable position of the isocentre in virtually planned patients. We did not found differences between the two planning procedures concerning this parameter. The de®nition of the isocentre immediately after CT scan proved as practicable in our investigation and could be used for physical treatment planning in each patient without the need for corrections. The isocenter position proved not as critical for the treatment plan in our investigation. A homogenous dose distribution could be improved by changes of asymmetric collimators, wedges, ®eld weights and gantry and collimator angles in cases with a sub optimal isocenter position. On the other hand, in these cases the physicist could be changed the isocenter. If the true isocenter had been differed
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272
from the CT simulation isocenter, the software has been calculated the shift using anterior-posterior, right-left and superior-inferior translations [16]. In comparison with the simulation ®lms, the contrast between chest wall and the outer contour of the breast was lower in DRR's caused on the use of a standard window and an individual ®ne-tuning of the contrast and brightness levels. This leads to an increased discrimination of chest wall structures, the central lung distance and the outer breast contour. The quality of DRR's was more similar to that of electronic portal images in comparison to simulation ®lms. Clips, which were positioned in the tumour bed for the boost irradiation, could not be envisaged in DRR's [3]. In these patients, an additionally contouring of the clips was necessary, which lasts additionally 1±2 min. The resolution of the DRR's is limited by the voxel size of the CT scan and by 512 £ 512 pixels. The slice width and thickness as well as patient's movement during the CT scan procedure further degrades the DRR's [9,17]. The acquisition of the scan volume should be performed without rest [18]; we needed an interval of about 5 s between the acquisitions of the two spirals with a scanning time of approximately 35 s each. The precision of the calculated divergence in the DRR's depends on the slice thickness and was found to be less than 1.0 mm [17].Due to an inadequate generated scanning volume virtual simulation and DRR's were insuf®cient in one patient. The omission of a conventional simulator appointment after planning CT using virtual simulation seems to reduce the delays before treatment. A reduction of the postoperative treatment interval leads to better local tumour control ®gures [6]. Additionally a postoperative interval of less than 6 weeks is often a prerequisite for patients to be eligible for breast cancer trials. From this perspective the reduction of the postoperative interval due to virtual simulation can be of some value. Moreover the schedule for the patient could be done easier using virtual simulation. The initial set-up check of the patients was done using an electronic portal imaging device. Being aware of uncertainties in the judgement of on-line portal ®lms up to 5 mm [10], the set-up of the patients has been corrected according to the simulation results in terms of the central lung distance and breast contour, if any distance difference between simulation and electronic portal image had been more than 3 mm. This was necessary in two virtually and conventionally simulated patients, each. A correction was also needed in three conventionally simulated patients, in whom differences between the planned and simulated isocenter were found. The range was in the order of described simulation errors of 6.4 ^ 3.9 mm (mean and 1 SD) and ^5 mm (in 78% of the patients) in different cancers, respectively [15,23]. These errors can compromise the treatment success, if not detected immediately and can be avoided using virtual simulation only. Virtual simulation has the potential to eliminate the trial and error of conventional simulation [24], but it cannot avoid the variances between the
271
marked and treatment isocenter, which seems to be comparable between the two groups. In contrast to this, the accuracy of patients marking with CT simulation was measured in different settings and found to be between 1 and 2.2 mm [8,17,20±22]. The degree of the corrections at the ®rst irradiation was in the order as others reported for interfractional set-up variations (^4.3 mm (1 SD) and 0.1 ^ 4.8 mm (Mean ^ 1 SD)) [1,14]. The set-up error between simulation, however, and the ®rst treatment session can be larger than interfractional variations [7]. Das et al. found a reduction of the ®rst set-up error after a conventional simulation by introducing a 2nd simulation including a sub sequential correction prior to the ®rst treatment fraction [5]. This corresponds to a reduction of the incorrect simulated isocentre at the 1st simulation as well as to respiratory movements, which have a major impact on the precision of the ®rst set-up. The CT scan represents the momentary respiratory situation of the patient during the scan. It can be vary considerable from the ®rst treatment situation. On the other hand, the spiral CT-scan lasts longer than one respiratory cycle in our investigation. Consequently, the rib cage and breast contour movements were recorded in the volumetric CT data set, producing the DRR's, which can be probably reduced by means of respiratory gating and/or mask ®xations [4,13]. Butker et al. have not found respiratory movements to be a signi®cant obstacle in virtual simulation [3]. The patient's set-up with virtual simulation according to the marked isocentre and additional marks in the lower chest wall region was adequate in this investigation. A CT simulation with subsequent treatment planning leads to patients marks indicating the treatment portals on the body surface. The disadvantages were discussed above. The mark of the whole treatment ®eld with conventional simulation leads not to a reduction of the set-up errors. As other authors have shown, no necessity exists for an additional conventional simulation after virtual simulation for the patients [3,8,25]. Galvin estimated 70% of the patients qualifying for the CT simulation [8]. 5. Conclusion The virtual simulation is a feasible tool for the treatment planning of patients undergoing irradiation of the breast. Compared with the conventional simulation procedure virtual simulation is superior regarding to the precision of patients marking, the quality of the reference images and, the time expenditure for the patients and medical staff. References [1] Boehmer D, Feyer P, Harder C, et al. Veri®cation of set-up deviations in patients with breast cancer using portal imaging in clinical practice. Strahlenther Onkol 174: Suppl. 1998;II:36±39. [2] Buchali A, Dinges S, Koswig S, et al. Virtual Simulation ± ®rst
272
[3] [4]
[5] [6] [7]
[8] [9] [10] [11] [12] [13] [14]
A. Buchali et al. / Radiotherapy and Oncology 59 (2001) 267±272 clinical results in patients with prostate cancer. Strahlenther Onkol 1998;174:88±91. Butker EK, Helton DJ, Keller JW, Hughes LL, Crenshaw T, Davis LW. A totally integrated simulation technique for three-®eld breast treatment using a CT simulator. Med Phys 1996;23:1809±1814. Creutzberg CL, Althof VG, Huizenga H, Visser AG, Levendag PC. Quality assurance using portal imaging: the accuracy of patient positioning in irradiation of breast cancer. Int J Radiat Oncol Biol Phys 1993;25:529±539. Das IJ, Cheng CW, Fosmire H, Kase KR, Fitzgerald TJ. Tolerances in setup and dosimetric errors in the radiation treatment of breast cancer. Int J Radiat Oncol Biol Phys 1993;26:883±890. Dubey AK, Recht A, Come S, Shulman L, Harris J. Why and how to combine chemotherapy and radiation therapy in breast cancer patients. Recent Results Cancer Res 1998;152:247±254. Fein DA, McGee KP, Schultheiss TE, Fowble BL, Hanks GE. Intraand interfractional reproducibility of tangential breast ®elds: a prospective on-line portal imaging study. Int J Radiat Oncol Biol Phys 1996;34:733±740. Galvin JM. Is CT simulation the wave of the future? Med Phys 1993;20:1565±1657. Galvin JM, Sims C, Dominiak G, Cooper JS. The use of digitally reconstructed radiographs for three-dimensional treatment planning and CT-simulation. Int J Radiat Oncol Biol Phys 1995;31:935±942. Herman MG, Abrams RA, Mayer RR. Clinical use of on-line portal imaging for daily patient treatment veri®cation. Int J Radiat Oncol Biol Phys 1994;28:1017±1023. Hess C-F, Christ G, Jany R, Bamberg M. Dose speci®cation at the 'ICRU-reference point`: consequences for the clinical practice. Strahlenth Onkol 1993;169:660±667. ICRU-Report No. 50. Prescribing, recording and reporting photon beam therapy. 1992. Kubo HD, Hill BC. Respiration gated radiotherapy treatment: a technical study. Phys Med Biol 1996;41:83±91. Lirette A, Pouliot J, Aubin M, Larochelle M. The role of electronic portal imaging in tangential breast irradiation: a prospective study. Radiother Oncol 1995;37:241±245.
[15] Lohr F, Schramm O, Schraube P, et al. Simulation of 3D-treatment plans in head and neck tumors aided by matching of digitally reconstructed radiographs (DRR) and on-line distortion corrected simulator images. Radiother Oncol 1997;45:199±207. [16] Mah K, Danjoux CE, Manship S, Makhani N, Cardoso M, Sixel KE. Computed tomographic simulation of craniospinal ®elds in pediatric patients: improved treatment accuracy and patient comfort. Int J Radiat Oncol Biol Phys 1998;41:997±1003. [17] McGee KP, Das IJ, Sims C. Evaluation of digitally reconstructed radiographs (DRRs) used for clinical radiotherapy: A phantom study. Med Phys 1995;22:1815±1827. [18] Nishidai T, Nagata Y, Takahashi M, et al. CT simulator: a new 3-D planning and simulating system for radiotherapy: Part 1. Description of system. Int J Radiat Oncol Biol Phys 1990;18:499±504. [19] Perez CA, Purdy JA, Harms W, et al. Design of a fully integrated three-dimensional computed tomography simulator and preliminary clinical evaluation. Int J Radiat Oncol Biol Phys 1994;30:887± 897. [20] Ragan DP, Forman JD, He T, Mesina CF. Clinical results of computerized tomography-based simulation with laser patient marking. Int J Radiat Oncol Biol Phys 1996;34:691±695. [21] Ragan DP, He T, Liu X. Correction for distortion in a beam outline transfer device in radiotherapy CT-based simulation. Med Phys 1993;20:179±185. [22] Ragan DP, He T, Mesina CF, Ratanatharathorn V. CT-based simulation with laser patient marking. Med Phys 1993;20:379±380. [23] Rosenman J, Sailer SL, Sherouse GW, Chaney EL, Tepper JE. Virtual simulation: initial clinical results. Int J Radiat Oncol Biol Phys 1991;20:843±851. [24] Sherouse GW, Bourland JD, Reynolds K, McMurry HL, Mitchell TP, Chaney EL. Virtual simulation in the clinical setting: some practical considerations. Int J Radiat Oncol Biol Phys 1990;19:1059±1065. [25] Valicenti RK, Waterman FM, Corn BW, Curran WJ. A prospective, randomized study addressing the need for physical simulation following virtual simulation. Int J Radiat Oncol Biol Phys 1997;39:1131± 1135.