1. The use of 4DCT images to optimize the Internal Target Volume in Radiotherapy  Nikos Giakoumakis, Brian Winey, Joseph Killoran, Tania Lingos, Laurence Court Dana Farber Cancer Institute, Brigham and Women’s Hospital, Boston MA.
MEDICAL SCHOOL
UNIVERSITY OF PATRAS
Background:
The introduction of 4D CT imaging techniques can prove the clinician with much needed patient-specific information on the motion of lung tumors. In many clinics the target volume is currently drawn as the outer envelope of the tumor motion seen in these images – rather than using margins based on population statistics. This use of the outer-envelope has been shown to be a conservative approach.
Van Herk et. al found that respiratory motion could be accounted for using an inferior margin of 0.25 of the peak-to-peak motion, and a superior margin of 0.45 of peak-to-peak motion. That is, a total margin of 0.7 times the peak-to-peak motion.
Mutaf and Brinkmann found that respiratory motion could be accounted for using a total margin of 0.72.A – 2.5mm, where A is the peak-to-peak motion.
Results (continued):
Treatment Planning
For each patient, VMAT plans was created (RapidArc, Varian Palo Alto Ca) for each of the four ITV targets (ITV_10/10, ITV_8/10, ITV_7/10, and ITV_6/10). The exhale phase CT image (phase 50%) was used for optimization and 3D dose calculation. All optimizations and dose calculation used Eclipse 8.6 (Varian). Dose was calculated using the AAA algorithm (Varian), with heterogeneity correction turned on. For each patient, the prescription was kept the same as that used for the actual original treatment (range: 37.5 – 66Gy).
Figure 6 shows how the D95, normalized to D95 for the ITV_10/10 plan (4D dose calculations), changes with the number of phases used to create the ITV. As less phases are used to create the ITV, the normalized D95 decreases. For some patients (e.g. the patient identified with triangles), the decrease is rapid. For others, (eg squares) there is little change.
Purpose:
To investigate a possible interim solution to better utilizing 4D CT images in determining the treatment volume, or ITV. The hypothesis is that instead of using all the phases in the 4D CT image to create the outer-envelope used for treatment planning, it should be possible to exclude phases closest to inhale, where the tumor spends less time.
TARGETcoverage constraint
At least 95% of the target received the prescribed dose.
NORMAL TISSUE constraints
50% of the total lung volume should receive a maximum dose of 5 Gy (V5<50%).
30% of the total lung volume should receive a maximum dose of 20 Gy (V20<30%)
Mean Lung dose should be less than 17 Gy
maximum spine dose was 50Gy
maximum dose within 7mm of the spinal cord was 54Gy
~2.5cm
Methods and Materials:
Patient selection
Ten lung cancer patients were identified who had received 4DCT imaging as part of their treatment simulation, and who had a noticeable tumor motion. The tumor location, and extent of motion is shown in table 1.
Figure 6. Comparison of the dose to 95% of the target as a function of the number of CT phases included in the ITV. All values are from 4D dose calculations, and are normalized to the D95 for N=10 (i.e. normalized to current clinical practice). Each marker type is for a separate patient.
After the optimization and dose calculation the test plans were compared with the plans used for the original actual treatments to ensure that the plans were as realistic as possible. For each patient all the plans were normalized so that they have the same D95 (dose to 95% of the target volume).
The treatment fields were then applied to all the other phases of the 4DCT, and the dose on each CT phase calculated separately. The dose from each phase was then deformably mapped to CT_50 using MimVista’s VoxAlign Deformation Engine, and the cumulative dose from the treatment calculated. This calculation is identified below as a 4D dose calculations.
Plan comparisons
The values for D95 calculated using 3D dose calculations on the exhale CT image, and 4D dose calculations on all images then deformed to the exhale CT image, were compared. The different plans were evaluated using the dose to 95% of the GTV_50 (D95) when the doses to all phases were deformed to the exhale phase CT, as shown in Figure 4. A plan was considered to be acceptable if the reduction in D95 was less than 5% compared with that for the plan based on ITV_10/10.
Patient
AP (cm)
SI (cm)
LR (cm)
Position (lobe) 1
0.4
0.9
lower right 2
0.6
0.8
1.1
middle right 3
0.7
0.5
lower left 4
upper right 5
0.3 6 7
1.4 8 9
upper left 10
1.6
Figure 7. shows the ratio of the 4D and 3D dose calculations for each target, with this ratio normalized to the ratio for the ITV_10/10 plan (i.e. the values in table 3) As the ITV shrinks, the agreement between the 4D and 3D dose calculations gets increasingly poor. This is an attempt to illustrate the impact of variations between plans on the results. Although there are some differences in the two plots (figure 6 and 7), they are very close in nature – indicating that differences in details of the individual plans should not impact the conclusions of this work.
Table.1 Motion and position of the patient tumors. AP= Anteroposterior, SI=superoinferior, LR= left-right
Figure 1. The countouring of GTV_50 on exhale phase of the 4DCT
Figure 7. The ratio of the dose to 95% of the target calculated using 4D dose calculation to that calculated in the 3D treatment plan, as a function of the number of CT phases included in the ITV. All data are normalized to the D95 for N=10. This figure is very similar to figure 2, indicating that although there are differences in the treatment plans, these differences do not impact the conclusions of this work. Each marker type is for a separate patient.
Target delineation
A physicist redrew the GTV on the exhale phase of the 4DCT, using the clinically drawn target as a guide. This new GTV is identified as GTV_50. Example is shown in Figure 1.
The GTV_50 contour was then deformed to the other 9 phases of the 4DCT using deformable image registration software (MIMVista Cleveland OH), giving a total of 10 GTV contours, one on each phase of the 4DCT. ITV contours were then created by the Boolean sum of the individual GTV contours applied to the exhale phase CT image, as shown in Figure 2.
ITV contours are identified as ITV_N/10, where N represents the number of phases used to create the ITV. For example, the ITV that represents our current clinical practice, based on all 10 phases of the 4DCT, is identified as ITV_10/10. We also created ITV_8/10, ITV_7/10 and ITV_6/10, excluding the breathing phases closer to the inhale phase, as shown in Figure 3.
. In order to accentuate the impact of the ITV contour on the delivered doses, no additional CTV or PTV margins were used for treatment planning. That is, the ITV was considered to be the planning target volume, PTV.
Table 3 shows how the number of phases that must be included in the ITV if the D95 for GTV_50 from the 4D dose calculations is to be kept within 5% of the D95 for GTV_50 when all phases are used for the plan. The number of phases that must be included ranges from 7 to 10. There does not appear to be any link between this and the size or location of the tumor, or the extent of the motion. For nine of the patients, it is possible to reduce the number of phases used from 10 to 8, but the associated reductions in ITV are small.
Figure 2. The creation of the ITV contour as a Boolean sum of the individual GTVs
Figure4. Differences in dose distribution for 3D and 4D dose calculations for ITV_10/10
Patient
GTV_50
volume (cc)
ITV_10/10
volume
(cc)
Number of phases needed for ITV (5% criterion)
ITVopt volume (cc)
ITV reduction (cc)
ITV reduction (%)
Volume of the margin (cc)
Volume of the optimal margin (cc)
Internal Margin reduction (%) 1
138
190 10 52
0.0 2 41 59 6 49 17 18 8 55 3 37 56 48 19 11 42 4 81
105
102 24 21 12 5 38 45 7 30
429
472
461 43 32 25 88
129
123 40 35 15 64 75 73 9 86
112
106 26 20 23 93
118
Results:
The ratio of D95 for the target (GTV_50) when calculated in 4D (on all phases, and the doses then deformed to the exhale scan) to that calculated in 3D (on the exhale CT scan) for the plan where the target should be completely included in the ITV (i.e. ITV_10/10) is given in table 2. The average ratio is 0.93 ± 0.04, range:
Patient
Ratio of D95 for GTV_50 from 4D dose calculation to that from the 3D dose calculation for the plan using ITV_10/10 1
0.86 2
0.96 3
0.95 4 5
0.90 6 7 8
0.92 9
0.93 10
Figure 3. The different ITVs created using less phases closest to exhale phase
Table 3.
Conclusions:
In this work we attempted to reduce the size of the ITV by reducing the number of CT phases that are used to draw the outer envelope of the target motion
For some patients it may be possible to reduce the size of the ITV, but this is not true for all patients
As we shrink the planning target volume the accumulated deformed dose to the GTV gets worse.
Based on our data, there is no specific formulae that can be used in every case for the reduction of the internal target volume
The only way to find the true necessary treatment volume is to do 4D dose calculations.
Figure 5. shows the DVHs of the GTV_50 from the 3D and 4D dose calculations for patient 1 when different ITVs are used. It can be seen that the DVH of the GTV_50 is very similar for all 3D dose calculations, but 4D dose calculations show the DVH is increasingly degraded as less phases of the 4DCT are used to create the ITV.
Table 2. Ratio of D95 for GTV_50 from 4D dose calculation to that from the 3D dose calculation for the plan using ITV_10/10.
52nd Annual Meeting of the American Association of Physicists in Medicine in Philadelphia, Pennsylvania, July 18-22, 2010