- Email: [email protected]
In vivo dosimetry during pelvic treatment
Radiotherapy and Oncology, 25 (1992) 111-120 O 1992 Elsevier Science Publishers B.V. All rights reserved. 0167-8140/92/$05.00
111
RADION 01051
In vivo dosimetry during pelvic treatment S. H e u k e l o m , J. H . L a n s o n a n d B. J. M i j n h e e r The Netherlands Cancer Institute, Amsterdam, The Netherlands (Received 12 December 1991, accepted 22 May 1992)
Key words. Pelvic treatment; Radiotherapy; In vivo dosimetry; Dose measurements; Treatment planning
Summary High precision in vivo entrance and exit dose measurements have been performed with p-type diodes on patients during 8 MV X-ray irradiation of the pelvis, to investigate the accuracy of dose calculations in this region. Based on phantom measurements the accuracy of the p-type diode measuring system itself, i.e. the agreement with ionisation chamber dose measurements, was shown to be better than 0.7?0 while the reproducibility in the dose determination was 1.1%, 1.5% and 1.67o (1 S.D.) at the entrance point, isocentre and exit point, respectively, for the wedged lateral fields. Patient movement and the uncertainty in the diode position increased these values to 1.7 ~ , 1.5 ?o and 3.1% (1 S.D.) for dose determinations on patients. From the entrance and exit in vivo dose values the dose actually delivered to the isocentre was determined. For the anterior-posterior beams a good correspondence for most patients was observed at the entrance and exit point and at the isocentre between the in vivo and calculated dose values. For the wedged lateral beams a systematic deviation of about 3 ~o was observed. In addition to the in vivo dose measurements phantom dose measurements have been performed to quantify the accuracy of the dose calculation algorithms including the computation of the number of monitor units. These measurements also served to quantify the effects of the actual patient on the dose delivery. The measurements showed that accurate calculation of the dose requires a separation of the head and phantom scatter contribution of the output of the treatment machine. The dependence of the wedge factor on field size, depth and source-skin-distance has also to be considered for accurate dose calculations. The effect of the patient on the dose calculation is mainly related to the actual electron densities of fat and bone structures compared to water: neglecting these densities in the dose computation could yield deviations up to 8.5 % for the exit point in wedged beams. Based on these results, improvements in the dose calculation algorithms and monitor unit calculation including the use of the actual electron densities will be implemented in the treatment planning procedure.
Introduction The chain of steps in specifying the dose delivered to a patient undergoing radiotherapy includes several steps that may introduce an uncertainty in the actual dose value. T o obtain an optimum dose in relation to tumour control and damage to normal tissues, the overall uncertainty in the dose delivery to a patient has to be small. Some authors [4,7,17] proposed an accuracy requirement of about 3.5?0 (1 S.D.). In clinical practice, however, this desired level of accuracy cannot always be achieved, due to several sources of uncertainty [e.g. see 6]. An overall check of the whole dosimetry
procedure by in vivo dosimetry is therefore useful and sometimes even necessary if a high accuracy is required. Accurate in vivo dosimetry can be performed with several purposes. In clinical routine, usually the entrance dose is measured as a quality control procedure to check the actual dose delivered to dma×, i.e. the point on the central beam axis with maximum dose. Such a measurement includes a check of the patient set-up and a check of the machine performance [3,14,18]. This type o f in vivo dosimetry measurement is performed on a large group o f patients in which the entrance dose is determined at least once on each patient. An action level of 5 ?o is often applied before differences between
Address for correspondence: S. Heukelom, Radiotherapy Department, Academic Hospital Free University, De Boelelann 1117, 1081 HV, Amsterdam, The Netherlands.
112 measured and planned dose values are further investigated. A more accurate check of the actual dose delivery at the dose specification point, with an action level of for instance 3 ~o, requires a combination of entrance and exit dose measurements [e.g. 15,22]. In addition, to explain differences between measured and planned dose values, extensive phantom measurements have to be performed. In this way the influence of machine performance, the variation of patient contour and density and limitations of the dose calculation algorithms of the treatment planning system can be distinguished. Because dose calculation algorithms are generally semiempirical procedures based on beam data gathered in an extended homogeneous medium, e.g. a rectangular water tank, phantom measurements in combination with in vivo dosimetry allow the study of the influence of shape, size and composition of the patient on dose calculations. Such a study can be performed on a relatively small group of patients in which a set of careful dose determinations are performed on each patient. At the Netherlands Cancer Institute an in vivo dosimetry project was started with the aim to evaluate the accuracy of the dose calculation of existing treatment procedures and of newly designed techniques [13] which make use of new options of modern accelerators and 3-D treatment planning systems. For this purpose dose values at several points measured in individual patients were compared with phantom dose measurements and computer dose calculations. The results of these dose comparisons should be used to improve the dose calculation procedure which includes the calculation of the number of monitor units. The high sensitivity, the absence of bias voltage, the good spatial resolution and the immediate availability of the result define the diode as a good dosimeter to perform in vivo dose measurements for our study. However, several authors have demonstrated that accurate dose measurements with diodes require a number of correction factors [e.g. see 8,14,21]. Furthermore, little information is available about the accuracy and reproducibility of dose measurements using diodes under clinical conditions. Such an indication is a prerequisite if in vivo dosimetry is used to improve the dose calculation procedure. The purpose of this study was therefore firstly to quantify in clinical practice the accuracy of our technique of dose measurements based on the use of diodes which has been described elsewhere [8]. The second aim of this study is the investigation of the accuracy of the dose calculation procedure for the pelvic area by means of in vivo dosimetry in combination with phantom measurements. In this way the patient-related influence and the limitations of the dose calculation algorithms on the dose calculation
procedure can be quantified and may lead to improved procedures. Materials and methods
A schematic diagram of our investigation procedure is depicted in Fig. 1. In vivo dose measurements are combined with consecutive measurements on and in a polystyrene phantom, simulating the whole treatment as much as possible. These dose values are compared with treatment plans based on measurements in a large water tank. The three aspects of our study will be discussed separately. Patient measurements
The external irradiation technique of a number of tumours in the pelvic region consists in our clinic of a combination of an anterior-posterior (AP) and posterior-anterior (PA) open beam (two-field SSD technique) or an AP open beam combined with two opposing lateral wedged fields (three-field SAD technique). The wedges generally have isodose angles varying between 45 and 55 degrees. The three-field technique is preferred to the box technique to minimize the dose delivery to the rectum or bladder. The patients are positioned both in the supine or prone position depending on the type of tumour to be treated. Dose values have been specified according to the recommendations given in ICRU Report 29 [11]: for the two-field technique at the point positioned on the central axis of the beams in the midplane of the patient and for the threefield technique at the isocentre. Patient treatments have been performed with 8 MV X-ray beams generated by a Philips SL 75/14 or an SL 75/20 accelerator. in vivo
simulation
planning
r
patient
0
small
large
polystyrene phantom
water
tank
diode ionization chamber
Fig. 1. Schematic diagram of the investigation procedure. Dose values have been determined on the patient, in a polystyrene phantom of similar dimensions and calculated with the treatment planning system, based on extended phantom (watertank) data.
113 For each photon beam p-type diodes (type EDP-20, Scanditronix) were positioned with the PVC support on the entrance and exit surface of the patient. This type of diode consists of a semiconductor detector covered with a half-spherical build-up cap of 2.2 mm stainless steel and 2.8 mm epoxy which together is equivalent to approximately 2 cm water for 8 MV X-rays. The diodes are read out with an 8-channel electrometer developed in our institution. The readings of the diodes have been converted to dose values at the entrance or exit point [8]. The entrance point is defined in our work as the point along the central axis of the beam at a distance dmax, the depth of dose maximum, from the entrance surface. The exit point is defined as the point along the central axis of the beam positioned at a distance dma× proximal to the exit surface. Both definitions do not refer to the surface dose as used for instance in I C R U Report 24 [10]. Clinically the latter quantity is usually of less interest than the dose at dmax. The entrance and exit dose values are determined from the diode reading, Rd~ode, by the expression: O d i o d e = R d i o d e * N D * 1-I i C i
(1)
ND is the calibration factor of the diode obtained under reference irradiation conditions [8] and the Crvalues are correction factors to modify ND for non-reference circumstances. The correction factors Ci applied in this equation originate mainly from the variation of the sensitivity of the diode with dose per pulse value [20] and photon energy spectrum [8]. Details about the different correction factors and the experimental procedure to determine numerical values for these factors, can be found elsewhere [8]. It should be noted that the ND and Ci values are generally different for the entrance and exit point. For this study it was necessary to determine an additional correction factor, Cair gap, required for diodes positioned in the AP open beam at the exit side of the patient. Due to the direct contact of the patient with the treatment couch, it was unfeasible to position a diode between the back of the patient and the couch. This diode was therefore placed on the Melinex sheet, after removing a panel of the treatment couch. This method sometimes yielded a small air gap between the back of the diode and the patient. The change of the sensitivity of the diode by varying the volume of this air gap, which is defined by the exit field size and the distance between the diode and the skin of the patient, has been measured and expressed by the correction factor Ca~r gap. Before each treatment the distance between the diode and the skin of the patient was estimated.
The dose at the midline point can be deduced from the entrance and exit dose values following the method proposed by Rizzotti et al. [22]. The midline point is defined as the point on the central axis of a beam in the middle of the patient. The relation between entrance, exit and midline dose, Den t. . . . . . Oexit and Omidline, r e s p e c t i v e l y , was determined using a polystyrene phantom and a cylindrical ionisation chamber (type NE 2505/3A) and is shown in Fig. 2. For our study this relation was determined for a source-skin-distance (SSD) of 80 cm, a 15 x 15 cm field size at the isocentre and a wedge of 45 °, which are the approximate values for the lateral wedged photon beams in clinical routine. Although for large thicknesses of the patient, i.e. above about 25 cm, the relation depends on these parameters, the use of a single curve introduces only a small error, less than 0.4 ~o, in the determination OfDmidlin e. For the AP and PA photon fields the same relation has been used in the determination of Dmidlin e. In clinical practice the midline point was not always at the isocentre for the three-field technique, particularly for the AP beam. Therefore, a conversion of the midline dose to the dose at the isocentre was performed for each photon beam, using the percentage depth dose (PDD) curve corresponding to the field size at the skin of the patient. The magnitude of this conversion could range up to 5 YD. For 11 patients the entrance and exit dose of each field was determined 3-8 times during her or his treatment of 35 fractions. The entrance and exit diodes were positioned about 1 cm off the central axis of the beam in opposite direction, to avoid the "shadowing" effect of one diode by the other. For wedged beams the diodes were positioned on the axis perpendicular to the wedge profile and the central beam axis, using the cross-wire system of the photon beam. The reduction 1.00
o
(3
0.80
8 ~c uJ
0.60
0.40
3 0.20
0.OO 0.00
~
J
~
J
0.20
0.40
0.60
0.80
1.00
E x i t Dose / Entrance Dose
Fig. 2. The ratio o f the dose value at the midline a n d entrance point vs. the ratio o f the dose value at the exit a n d entrance point as determined in a h o m o g e n e o u s p h a n t o m .
114 of the total dose delivered to the target volume, caused by the attenuation of the primary photon beam by the build-up cap of the entrance diode, was at most 1.5 ~o due to the limited number of measurements with the diodes [ 18,21]. The positioning of the diodes and the conversion of the diode reading to dose values were performed by a technician or a physicist. The performance of in vivo dosimetry increased the treatment time per session with at most 10 min. This increase in time was mainly caused by the positioning of the diodes on the patient for the AP field and both lateral wedged beams which required additional rotations of the gantry. Phantom measurements An essential part of this study was the accompanying dose determination in a polystyrene phantom with dimensions similar to each patient. These measurements served several purposes. Firstly, the measurements allow the assessment of the accuracy and reproducibility of the diode measuring system under clinical conditions. For this purpose entrance and exit dose as well as the dose at the isocentre were determined with the diode measuring system in a similar way as during patient treatment. These dose values were compared with values determined with a Farmer type 0.6-cm 3 cylindrical ionisation chamber (NE 2505/3A) connected to a Keithley electrometer. Secondly, the phantom measurements served to check the accuracy of the dose calculation algorithms, including the algorithm to calculate the number of monitor units, and to determine the uncertainty in the dose prediction due to patient related effects. Both checks have been performed by comparing the phantom dose values with those predicted by dose calculations using our treatment planning system, for the phantom and patient irradiations (see next section). The phantom dose values used in these dose comparisons have been determined by means of ionisation chamber measurements as mentioned before. Finally, additional phantom measurements have been performed to determine the correction factor Cai r gap for the entrance and exit point as a function of the volume of the air gap. The method to determine this factor is similar to those applied for the other correction factors as described elsewhere [8]. During the phantom measurements the patient treatment was simulated as close as possible. The beam set-up parameters were taken identical to those during the actual patient treatment while the thickness of the phantom was equal to the patient thickness as determined during the simulator session of the treatment. In this way patient and phantom dose values could be
compared directly. These phantom measurements will be described as "simulation" measurements in our study. For each patient, simulation measurements have been repeated at least 3 times during the course of the treatment in which each dose value obtained with the ionisation chamber and diode was based on at least 3 readings. The variation in magnitude between consecutive readings was smaller than 0.2~ (1 S.D.) for the ionisation chamber and 0.3~o (1. S.D.) for the diode, respectively. The readings of the ionisation chamber have been converted to dose values by applying the NCS code of practice [ 16]. No correction for variation of stopping power ratios with field size and depth is taken into account in this code of practice. This additional uncertainty has been estimated to be smaller than 0.5% [2]. The simulation measurements were performed on a polystyrene (PS) slab phantom. The dose values determined in this phantom, Dps, have been converted to dose-to-water values, Dwater, because in all dose calculations the patient and phantom material have been assumed to be water equivalent. This conversion has been performed by applying the procedure outlined in the A A P M protocol [ 1 ]. The resulting dose correction was less than 2~o for phantom thicknesses up to 40 cm. Treatment planning Dose values determined in patients and in the phantom have been compared with calculated dose values. For the AP-PA irradiations the number of monitor units to deliver a specified dose at the midline point is calculated from measured depth dose data. The dose distribution in the pelvic region for the three-field irradiation technique is planned in our institution with a treatment planning system (Nucletron Planning System), using 2-D dose calculation algorithms. The patient contour is taken into account using a modification of the effective SSD method, which gives an improvement in the dose calculation for oblique incident beams over conventional algorithms [23]. Tissue inhomogeneities are calculated using the equivalent path length model. In routine clinical practice the dose calculation procedure is simplified: the effect of inhomogeneities such as bone structures are ignored in treatment planning. However, to explain differences observed between in vivo and calculated dose data, the dose distribution for the threefield technique has been recomputed, considering the actual electron densities in the treatment volume in the patient. These electron densities have been based on computer tomographic (CT) information, which was available for five patients. Computation of absolute dose values and monitor units is incorporated in the treatment planning system.
115 For this purpose, output factors for square field sizes and wedge factors have to be added to other input data. In the dose calculation algorithm the output factor as defined in the British Journal of Radiology Supplement 17 [5], has been adopted. Furthermore, the output factor of a rectangular field size is considered to be equal to the output factor of the equivalent square field [5]. In our clinic the wedge factor is defined as the ratio of the dose with and without a wedge in the beam for the same number of monitor units while it is only determined for one irradiation geometry: a depth of 8 cm in a water phantom, a 10 x 10 cm field size at the isocentre and an SSD equal to the source-axis-distance. The dependence of the wedge factor on field size [e.g. see 19] and phantom depth, i.e. beam hardening, is ignored in our planning system. Results
Accuracy and reproducibility of the diode measuring system The accuracy of the diode system can be defined as the ratio of the dose values determined at identical points during the simulation measurements with the diodes and the ionisation chamber. Table I summarises these results for the entrance point, isocentre and exit point for the open and wedged beams. The systematic differences between dose values determined with both measuring systems are smaller than 0.7%. The reproducibility in dose determinations by the diode system, also given in Table I, has been obtained from an analysis of the dose ratios determined for comparable dose measurements performed during different simulation sessions. The reproducibility of the diode measuring system varies between 0.5% (1 S.D.) at the entrance point for open beams to 1.6% (1 S.D.) at the exit point for wedged beams.
Determination of C, grgap The determination of C a i r gap at the entrance side of the phantom yields small values: 1.005 and 1.007 for a thickness of the air gap of 10 mm and 20 ram, respectively, both independent of field size. At the exit side of the phantom the C a i r gap values are depicted in Fig. 3. During the determination of the correction values, the variation of the diode reading due to the effect of the inverse square law, has been eliminated. For clinical situations, i.e. an equivalent square field size with a side of about 15 cm and a thickness of the air gap of about 10 mm, a correction of about 1% has been observed. In addition, the reading of the diode has to be corrected for the effect of the inverse square law which magnitude depends on the clinical situation. Dose comparison: patient dose vs. dose calculation For each patient average dose values at the entrance point, isocentre and exit point have been deduced from all in vivo dose measurements performed for a particular patient. The average dose value at each point has been compared with the expected dose value: Fig. 4 shows for each point and each patient the ratio of the average in vivo dose and calculated dose, both for open and wedged beams. For the lateral wedged beams at the entrance point the actual delivered dose is on the average 3.3 % ( + 1.8 % 1 S.D.) higher than the expected dose value. For the isocentre and exit point such systematic differences have also been observed, 2.8 % for both points, but the standard deviations are larger: 2.4~o and 4.9%, respectively. For the AP-PA beams similar results have been observed although they were less pronounced. The standard deviations belonging to the average in vivo dose ratios are also indicated in Fig. 4. From these data an average variation (1 S.D.) in the dose determination on a patient was derived for each point
m 120 air
7
gap thickness (mm) | /
TABLE I
• = 30
|
o = 23
L
• = 16
Accuracy and reproducibility (1 S.D.) of the diode measuring system, determined during phantom measurements. Position
Entrance Isocentre Exit
Open fields
~= 10
~:=
110
3
Wedged fields
Accuracy
Reproducibility
Accuracy
Reproducibility
1.002
0.5 %
1.000
1.007 1.002
0.9% 0.9°;
1.003 1.006
1.1% 1.5% 1.6%
For the accuracy, dose values have been determined relative to ionisation chamber measurements. The values are based on a total number of 181 and 139 measurements for the open and wedged fields,
1.00
10
15
25 30 side of square field at exit side (cm)
Fig. 3. The correction factor Cair gap for the exit point as a function of field size at the exit surface of the phantom for different air gap thicknesses between the surface and the back of the exit diode. The effect of the inverse square law on the reading of the diode with increasing air gap thickness, has been eliminated.
116
AP-PA
Open Beams
Lateral 1.10
1.10 1.08
A
1.oe
Entrance
1.06
Wedged Beams
13,
1.06
I"
~
~
T
-
Entrance
{
T
1.04 1.02 1.00 0. &
0.911
0.96
O.96
0.94 I~ (/)
0
o.88 0.86
1:3
0.94
@ o') 0
0.92 o.go i
i
i
i
i
i
i
i
i
i
f
l
t
l
r
l
l
l
l
l
l
r'l
l
~
1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 0 1 0 1 0 1 1 1 1
"0
1.1o
-~J ir~
1,08 1.06
0
1.04
-~
1.02
(~
1.00 O.g*g
O')
0.92
o
0.9O
(.)
}~,__!~_~.=
..... ~ .... T_~__
0.86
~
l
l
r
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 1 0 1 0 1 1 1 1
0) ---
1.10
"~
1.08
0-
1,06
i
i
i
i
i
i
i
r
I
I
7
7
8
8
9
9
10
10
11
11
Isocentre
{
1.04
{{
o.~'°°. . . . . .
~"
0.96
O)
0.94
03
0.92
0 C~
0.9o 0.811 0,86
O)
•- -
I
i 6
0
+~
l
i 6
1.1o 1~
I socentre
0.88 -~
o.92 o.9o o.8a
!
I
~--} . . . . . . . . . . . . . . . . .
I
I
I
I
I
I
r
I
I
i
I
i
6
6
7
7
8
8
9
9
I0
10
11
11
1.10
Exit
1.04 1.02 tOO
i
0.98 0.96
rT
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0.98 0.96
0,94 0.~)2
0.92
O.gO
0.9O
0.88 0.8~
0.88 r r I I I i I , i i ~ t L 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 g 9 10101111 I
I
I
I
f
I
r
I
I
0 ~
Patient Number
I
I
r
I
I
I
I
I
I
I
I
I
6
6
7
7
8
8
0
9
10
10
11
11
Patient Number
Fig. 4. The ratio of patient dose and calculated dose values, given for each point of measurement and both open AP-PA beams and both lateral wedged beams.
of dose measurement in the AP-PA open and lateral wedged beams. The results have been tabulated in Table II. As shown in this table, the average variation increases from 0.9% for the entrance point for open beams to 3.1% at the exit point for wedged beams.
Dose comparison: phantom dose vs. dose calculation Figure 5 summarises the results of the phantom dose comparison. In this figure the frequency distribution is given of the average dose measured in the phantom at the entrance point, isocentre and exit point for the open and lateral wedged beams relative to the calculated dose. For open beams, difference have been observed between measured and calculated dose values ranging from - 5 % to 5 %. For wedged beams almost all calculated dose values were systematically lower than the measured values. A maximum difference of + 10 % and + 12% was observed at the isocentre and exit point, respectively.
Discussion
Accuracy and reproducibility of the diode measuring system The good accuracy and reproducibility of the diode system (Table I) show that determination of the entrance and exit dose and the dose at the isocentre can TABLE II Average variation in the dose determination during pelvic treatment of 11 patients applying diodes. Position
Open fields
Wedged fields
Entrance Isocentre Exit
0.9 % 1.3 % 2.2 %
1.7 % 1.5 % 3.1%
The values are based on a total number of 193 and 154 measurements for the open and wedged fields, respectively.
117 Entrance
[
I Isocentre
~
to the wedge. The worst reproducibility of the diode measuring system will therefore be observed for the exit dose measurements in the lateral fields. It has to be noted that the uncertainty in dose determination with the diode system due to variation of the output of the treatment machine is eliminated in these results, due to the simultaneous dose determination with the diodes and the ionisation chamber.
Extt
25
N= 1 7 8
Opera B e a m s 2O
15
10
Determination of C, tr gap 5
>, U C
. . . .
0 0.g5
1.00
1.05
,
. . . .
1.10
,
1.15
O-
kS 25
9Vedged B e a m s
N=142
2O
15
10
5
In order to determine with high accuracy patient dose values, the additional correction factor Cai~ gap has been applied in the reading-to-dose conversion of diodes positioned in the AP field. For entrance dose determinations this factor could almost be ignored. For the exit dose determination the air gap could be about 10 mm introducing a total correction, i.e. including a correction for the inverse square law, on the diode reading up to 3~o for field sizes applied in the pelvic treatment technique. The correction factor Cair gap originates from a lack of phantom scatter, i.e. from electrons, compared with the non-air gap situation. For increasing field size this build-down effect in the region near the diode, is partially compensated by electrons coming from a larger distance from the diode.
0
0.g5
1 C0
105
1.10
1.15
>hantom Dose / Calculated Dose Fig. 5. The frequency of the ratio of the measured dose and the dose calculated by the computer planning system for phantom irradiations, given for each point of measurement in the open AP-PA beams and the lateral wedged beams. N is the number of phantom measurements.
be performed with p-type diodes almost as accurately as performed with an ionisation chamber. Accurate diode dose measurements require besides a well known dose calibration factor ND, a large number of correction factors Ci on the diode reading both fore the entrance and for the exit point [8,15,18,21]. For the exit point these correction factors are generally more pronounced and vary more strongly, e.g. with field size and depth compared with similar correction factors required for the entrance dose determination. The reproducibility of exit dose measurements is therefore somewhat less compared with entrance dose measurements. This is especially the case for the irradiation of patients of large size, e.g. up to 40 cm in the lateral direction, applying wedged photon beams with a wedge angle equal or larger than 45 °. Furthermore, for wedged beams an additional uncertainty of about 17o (1 S.D.) is introduced due to the positioning of the diode with respect
Dose comparison: patient dose vs. dose calculation
The deviations between the in vivo dose and calculated dose values as depicted in Fig. 4 can be explained by limitations in the dose calculation algorithms and by patient related effects. The limitations of the algorithms have been quantified by comparing measured dose values with calculated values for the phantom irradiations, yielding underestimations up to 12 ~o for the calculated dose at the exit point in the wedged lateral beams (see Fig. 5). Uncertainties in dose computations due to patient related effects were quantified from a comparison TABLE III Maximum deviation between the dose calculated by the treatment planning system and the dose measured in a phantom, for the lateral fields of pelvic irradiations.
Output calculation Output factor Collimator exchange effect Wedge factor (field size) Wedge factor (SSD)
-
1.6% 1.0yo 2.0% 1.0%
Calculation of the depth dose distribution PDD wedged beam Lack of backscatter Overall difference
Entrance point + 1.0%
Isocentre 1.63;
Exit point -4.1~ + 1.3%
- 4.6 %
- 7.~,/o
- 8.4%
118 of the in vivo and phantom dose values. Analysis of this part of the dose comparison showed that these uncertainties in dose calculations were mainly introduced by the non-water equivalence of the treatment volume of the patient, i.e. the presence of fat and bone structures in the patients. The influence of these inhomogeneities on the dose prediction has been estimated from dose calculations using the actual electron densities in the target volume, available for five patients. For the open beams the overall effect was negligible while for the exit points in the wedged lateral beams a reduction up to 8.5~o of the calculated dose values was observed in patient no. 7 (see Fig. 4) if the electron densities were take into account. This reduction is caused by the high density of the femoral heads which overcompensates the lower density of fat tissue relative to water. It has to be noted that the presence of inhomogeneities in the treatment field will introduce an error in the determination of the dose at the isocentre applying the curve given in Fig. 2, as this curve is only valid for homogeneous geometries. However, as the diodes were positioned in the lateral wedged beams behind both femoral heads, this error can be assumed to be negligible, due to the symmetrical distribution of almost similar inhomogeneities. For the AP, PA beams an uncertainty up to about 0.6~o in the dose value at the midline point has been estimated due to the asymmetrical position of the sacral bone in these fields. For very accurate dose determinations the exact position of the diodes with respect to these inhomogeneities has to be determined by making megavoltage images of a patient. In clinical practice the total dose at the isocentre and its deviation from the expected dose value are important and therefore have to be assessed. About half the dose at this point is given by the AP or PA beam, for which a small average deviation of 1.0~o has been observed between the actual and expected dose values. For the lateral wedged beams an average difference of 2.8~o has been observed, yielding on the average a small but systematic overestimation of 1.8~o of the actual dose value at the isocentre for the combination of the AP or PA and lateral photon beams. Comparison of the magnitude of the average variations in in vivo dose measurements (Table II) with the reproducibility in dose determination of the diode measuring system alone (Table I), shows that daily variation in the patient positioning and patient movements decrease the reproducibility of in vivo dose determinations. Especially for dose determinations in wedged beams, the positioning of the diode and patient movement will decrease the precision due to the unknown movement of the diode over the wedge profile.
Dose comparison: phantom dose vs. dose calculation The large deviations between the actual phantom dose values and the expected dose data, especially for the wedged beams (Fig. 5), show that limitations in the dose calculation algorithms introduce large differences between the computed dose values and the in vivo dose results. A quantification of these limitations has been obtained from a further investigation of the accuracy of the dose calculation algorithms, and is given in Table III. Four effects are related to limitations in the algorithm of the absolute dose calculation at dma x (output computation) whil~ two other effects are related to the calculation of the relative dose distribution. Firstly, the increase of the output of an accelerator with increasing field size originates from two effects: an increase of the fluence of the primary photon beam, i.e. due to an increase of the head scatter [ 12], and an increase in the phantom scattered dose contribution. Considering the output variation only due to variation in the phantom scattered dose contribution, an error up to -1.6~o in the calculated output was introduced for an SSD of 80 cm. Secondly, computation of the output factor for a rectangular field by the method of the equivalent square field [ 5 ], introduced an error of - 1.0 ~o due to the collimator exchange effect [ 12]. Thirdly, the wedge factor depends on the field size [e.g. see 19]. Neglecting this effect resulted in an underestimation of the actual output of the accelerator of about 1.5 ~/o for the field sizes used during pelvic irradiations. The final refinement in the absolute dose calculation algorithm concerns the increase of the wedge factor with decreasing SSD which resulted in an underestimation of the actual output for wedged beams of about 1.0~o if the SSD changes from 100 cm to 80 cm. The limitations related to the calculation of the relative dose distributions are mainly due to neglecting the depth dependence of the wedge factor, yielding an error in dose computation up to about -4.0~o for the exit points in the lateral wedged beams. In addition, the lack of backscatter at the exit points has not been considered in the treatment planning process which results in too high an estimation of the actual dose at these points with about 1.3~o.
Conclusions
In this study the accuracy and precision of in vivo dose determinations using a p-type diode measuring system have been explored for patients treated in the pelvic region. A high degree of accuracy and precision in dose determination was observed which required the determination of a large set of correction factors on the dose
119 calibration factor determined under reference irradiation conditions. The dose comparison performed in this study shows that even for the simple irradiation geometry of the pelvic region, large deviations have been observed between dose values determined at identical points in patients on a phantom and calculated by the computer planning system. These deviations were much more pronounced for wedged beams than for open beams. They originate firstly from limitations in the absolute dose calculation algorithm, i.e. in the output computation, resulting in an uncertainty in the number of monitor units. Quality control of treatment planning systems should therefore always include tests of the algorithms concerning the calculation of the absolute dose and the number of monitor units. Special attention has to be given to the head and phantom scatter component of the output of the accelerator and the wedge factor as a function of field size and SSD. Additional errors in the routine dose calculation are introduced due to limitations in the computation of the relative dose distribution, e.g. in the P D D of wedged beams. It is important to note that correction of the algorithm which calculates the number of monitor units, to improve the dose prediction for a certain technique, might deteriorate the dose computation for another technique. In our institution such a result has been observed for the treatment of breast carcinoma [9]. Adaptations of dose calculation algorithms should
therefore be accompanied by verification of the calculated dose for various important treatment techniques. Our study also demonstrated that the prediction of the actual dose delivery at the isocentre can be improved if the electron densities of fat and bone structures in the treatment volume of the patient, obtained from CT information, are taken into account. The results of this study show that improvement of the dose calculation procedure can be obtained from accurate in vivo dose measurements at the entrance and exit side of a patient. An in vivo dosimetry program should always include phantom measurements to quantify the accuracy of the measuring system and to separate errors in the dose calculation algorithms from effects related to the patient. The results of this study have guided us in investigating more complicated treatment techniques such as the treatment of breast carcinoma [9] and prostatic cancer using the simultaneous boost technique [13]. Accurate in vivo dosimetry in combination with portal imaging are indispensable tools to perform these studies.
Acknowledgements We would like to thank Ir A. A. M. Hart for useful discussion and Ing. J. Weeda for his help during some of the phantom measurements. This work was financially supported by the Netherlands Cancer Foundation (NKB Grant 87-13).
References 1 AAPM Task Group 21, American Association of Phycisists in Medicine. A protocol for the determination of absorbed dose from high energy photon and electron beams. Med. Phys. 10: 741-771, 1983. 2 Andreo, P. and Brahme, A. Stopping power datafor high-energy photon beams. Phys. Med. Biol. 31: 839-858, 1986. 3 Bascuas, J.L., Chavaudra, J.P., Vauthier, G. and Dutreix, J. Inter6t des measures "in vivo" systematiques en radioth6rapie. J. Radiol. Electrol. 58: 701-708, 1977. 4 Brahme, A., Chavaudra, J., Landberg, T., McCullough, E., Niisslin, F., Rawlinson, A., Svensson, G. and Svensson, H. Accuracy requirements and quality assurance of external beam therapy with photons and electrons. Acta Oncol. Suppl. 1, 1988. 5 Central Axis Depth Dose Data for Use in Radiotherapy. In: Br. J. Radiol. Suppl. 17, p. 145. Editors: D. K. Bewley, A. L. Bradshaw, D. Greene, J. L. Haybittle and L. F. Secretan. The British Institute of Radiology, 1983. 6 Dutreix, A. When and how can we improve precision in radiotherapy? Radiother. Oncol. 2: 275-292, 1984. 7 Goitein, M. Nonstandard deviations. Med. Phys. 10: 709-711, 1983. 8 Heukelom, S., Lanson, J. H. and Mijnheer, B.J. Comparison of entrance and exit dose measurements using ionization
9
10
11
12 13
14
15
chambers and silicon diodes. Phys. Med. Biol. 36: 47-59, 1991. Heukelom, S., Lanson, J. H., van Tienhoven, G. and Mijnheer, B.J. In vivo dosimetry during tangential breast treatment. Radiother. Oncol. 22: 269-279, 1991. ICRU, International Commission on Radiation Units and Measurements. Determination of absorbed dose in a patient irradiated by beams of X or gamma rays in radiotherapy procedures. Report 24, ICRU Publications, Washington, DC, 1976. ICRU, International Commission on Radiation Units and Measurements. Dose specification for reporting external beam therapy with photons and electrons. Report 29, ICRU Publications, Washington, DC, 1978. Kase, K. R. and Svensson, G . K . Head scatter data for several linear accelerators (4-18 MV). Med. Phys. 13: 530-532, 1986. Lebesque, J.V. and Keus, R. The simultaneous boost technique; the concept of relative normalized total dose. Radiother. Oncol. 22: 45-55, 1991. Leunens, G., van Dam, J., Dutreix, A. and van der Schueren, E. Quality assurance in radiotherapy by in vivo dosimetry. 1. Entrance dose measurements, a reliable procedure. Radiother. Oncol. 17: 141-151, 1990. Leunens, G., van Dam, J., Dutreix, A. and van der Schueren,
120
16
17
18
19
E. Quality assurance in radiotherapy by in vivo dosimetry. 2. Determination of the target absorbed dose. Radiother. Oncol. 19: 73-87, 1990. Mijnheer, B. J., Aalbers, A. H. L., Visser, A. G. and Wittk~nper, F.W. Consistency and simplicity in the determination of absorbed dose to water in high energy photon beams: A new code of practice. Radiother. Oncol. 7: 371-384, 1986. Mijnheer, B.J., Battermann, J. J. and Wambersie, A. What degree of accuracy is required and can be achieved in photon and neutron therapy? Radiother. Oncol. 8: 237-252, 1987. Nilsson, B., Ruden, B-I. and Sorcini, B. Characteristics of silicon diodes as patient dosemeters in external radiation therapy. Radiother. Oncol. 11: 279-288, 1988. Palta, J. R., Daftari, I. and Suntharalingam, N. Field size dependence of wedge factors. Med. Phys. 15: 624-626, 1988.
20 Rikner, G. and Grusell, E. Effects of radiation damage on p-type silicon detectors. Phys. Med. Biol. 28: 1261-1267, 1983. 21 Rikner, G. and Grusell, E. Patient dose measurements in photon fields by means of silicon semiconductor detectors. Med. Phys. 14: 870-873, 1987. 22 Rizzotti, A., Compri, C. and Garusi, G . F . Dose evaluation to patients irradiated by 6°Co beams, by means of direct measurement on the incident and on the exit surfaces. Radiother. Oncol. 3: 279-283, 1985. 23 Westermann, C. F. and Mijnheer, B.J. Dose correction for oblique incidence of megavoltage photon beams. In: Proc. 8th Int. Conf. on the Use of Computers in Radiation Therapy, Toronto, Canada, pp. 335-338, 1984.
111
RADION 01051
In vivo dosimetry during pelvic treatment S. H e u k e l o m , J. H . L a n s o n a n d B. J. M i j n h e e r The Netherlands Cancer Institute, Amsterdam, The Netherlands (Received 12 December 1991, accepted 22 May 1992)
Key words. Pelvic treatment; Radiotherapy; In vivo dosimetry; Dose measurements; Treatment planning
Summary High precision in vivo entrance and exit dose measurements have been performed with p-type diodes on patients during 8 MV X-ray irradiation of the pelvis, to investigate the accuracy of dose calculations in this region. Based on phantom measurements the accuracy of the p-type diode measuring system itself, i.e. the agreement with ionisation chamber dose measurements, was shown to be better than 0.7?0 while the reproducibility in the dose determination was 1.1%, 1.5% and 1.67o (1 S.D.) at the entrance point, isocentre and exit point, respectively, for the wedged lateral fields. Patient movement and the uncertainty in the diode position increased these values to 1.7 ~ , 1.5 ?o and 3.1% (1 S.D.) for dose determinations on patients. From the entrance and exit in vivo dose values the dose actually delivered to the isocentre was determined. For the anterior-posterior beams a good correspondence for most patients was observed at the entrance and exit point and at the isocentre between the in vivo and calculated dose values. For the wedged lateral beams a systematic deviation of about 3 ~o was observed. In addition to the in vivo dose measurements phantom dose measurements have been performed to quantify the accuracy of the dose calculation algorithms including the computation of the number of monitor units. These measurements also served to quantify the effects of the actual patient on the dose delivery. The measurements showed that accurate calculation of the dose requires a separation of the head and phantom scatter contribution of the output of the treatment machine. The dependence of the wedge factor on field size, depth and source-skin-distance has also to be considered for accurate dose calculations. The effect of the patient on the dose calculation is mainly related to the actual electron densities of fat and bone structures compared to water: neglecting these densities in the dose computation could yield deviations up to 8.5 % for the exit point in wedged beams. Based on these results, improvements in the dose calculation algorithms and monitor unit calculation including the use of the actual electron densities will be implemented in the treatment planning procedure.
Introduction The chain of steps in specifying the dose delivered to a patient undergoing radiotherapy includes several steps that may introduce an uncertainty in the actual dose value. T o obtain an optimum dose in relation to tumour control and damage to normal tissues, the overall uncertainty in the dose delivery to a patient has to be small. Some authors [4,7,17] proposed an accuracy requirement of about 3.5?0 (1 S.D.). In clinical practice, however, this desired level of accuracy cannot always be achieved, due to several sources of uncertainty [e.g. see 6]. An overall check of the whole dosimetry
procedure by in vivo dosimetry is therefore useful and sometimes even necessary if a high accuracy is required. Accurate in vivo dosimetry can be performed with several purposes. In clinical routine, usually the entrance dose is measured as a quality control procedure to check the actual dose delivered to dma×, i.e. the point on the central beam axis with maximum dose. Such a measurement includes a check of the patient set-up and a check of the machine performance [3,14,18]. This type o f in vivo dosimetry measurement is performed on a large group o f patients in which the entrance dose is determined at least once on each patient. An action level of 5 ?o is often applied before differences between
Address for correspondence: S. Heukelom, Radiotherapy Department, Academic Hospital Free University, De Boelelann 1117, 1081 HV, Amsterdam, The Netherlands.
112 measured and planned dose values are further investigated. A more accurate check of the actual dose delivery at the dose specification point, with an action level of for instance 3 ~o, requires a combination of entrance and exit dose measurements [e.g. 15,22]. In addition, to explain differences between measured and planned dose values, extensive phantom measurements have to be performed. In this way the influence of machine performance, the variation of patient contour and density and limitations of the dose calculation algorithms of the treatment planning system can be distinguished. Because dose calculation algorithms are generally semiempirical procedures based on beam data gathered in an extended homogeneous medium, e.g. a rectangular water tank, phantom measurements in combination with in vivo dosimetry allow the study of the influence of shape, size and composition of the patient on dose calculations. Such a study can be performed on a relatively small group of patients in which a set of careful dose determinations are performed on each patient. At the Netherlands Cancer Institute an in vivo dosimetry project was started with the aim to evaluate the accuracy of the dose calculation of existing treatment procedures and of newly designed techniques [13] which make use of new options of modern accelerators and 3-D treatment planning systems. For this purpose dose values at several points measured in individual patients were compared with phantom dose measurements and computer dose calculations. The results of these dose comparisons should be used to improve the dose calculation procedure which includes the calculation of the number of monitor units. The high sensitivity, the absence of bias voltage, the good spatial resolution and the immediate availability of the result define the diode as a good dosimeter to perform in vivo dose measurements for our study. However, several authors have demonstrated that accurate dose measurements with diodes require a number of correction factors [e.g. see 8,14,21]. Furthermore, little information is available about the accuracy and reproducibility of dose measurements using diodes under clinical conditions. Such an indication is a prerequisite if in vivo dosimetry is used to improve the dose calculation procedure. The purpose of this study was therefore firstly to quantify in clinical practice the accuracy of our technique of dose measurements based on the use of diodes which has been described elsewhere [8]. The second aim of this study is the investigation of the accuracy of the dose calculation procedure for the pelvic area by means of in vivo dosimetry in combination with phantom measurements. In this way the patient-related influence and the limitations of the dose calculation algorithms on the dose calculation
procedure can be quantified and may lead to improved procedures. Materials and methods
A schematic diagram of our investigation procedure is depicted in Fig. 1. In vivo dose measurements are combined with consecutive measurements on and in a polystyrene phantom, simulating the whole treatment as much as possible. These dose values are compared with treatment plans based on measurements in a large water tank. The three aspects of our study will be discussed separately. Patient measurements
The external irradiation technique of a number of tumours in the pelvic region consists in our clinic of a combination of an anterior-posterior (AP) and posterior-anterior (PA) open beam (two-field SSD technique) or an AP open beam combined with two opposing lateral wedged fields (three-field SAD technique). The wedges generally have isodose angles varying between 45 and 55 degrees. The three-field technique is preferred to the box technique to minimize the dose delivery to the rectum or bladder. The patients are positioned both in the supine or prone position depending on the type of tumour to be treated. Dose values have been specified according to the recommendations given in ICRU Report 29 [11]: for the two-field technique at the point positioned on the central axis of the beams in the midplane of the patient and for the threefield technique at the isocentre. Patient treatments have been performed with 8 MV X-ray beams generated by a Philips SL 75/14 or an SL 75/20 accelerator. in vivo
simulation
planning
r
patient
0
small
large
polystyrene phantom
water
tank
diode ionization chamber
Fig. 1. Schematic diagram of the investigation procedure. Dose values have been determined on the patient, in a polystyrene phantom of similar dimensions and calculated with the treatment planning system, based on extended phantom (watertank) data.
113 For each photon beam p-type diodes (type EDP-20, Scanditronix) were positioned with the PVC support on the entrance and exit surface of the patient. This type of diode consists of a semiconductor detector covered with a half-spherical build-up cap of 2.2 mm stainless steel and 2.8 mm epoxy which together is equivalent to approximately 2 cm water for 8 MV X-rays. The diodes are read out with an 8-channel electrometer developed in our institution. The readings of the diodes have been converted to dose values at the entrance or exit point [8]. The entrance point is defined in our work as the point along the central axis of the beam at a distance dmax, the depth of dose maximum, from the entrance surface. The exit point is defined as the point along the central axis of the beam positioned at a distance dma× proximal to the exit surface. Both definitions do not refer to the surface dose as used for instance in I C R U Report 24 [10]. Clinically the latter quantity is usually of less interest than the dose at dmax. The entrance and exit dose values are determined from the diode reading, Rd~ode, by the expression: O d i o d e = R d i o d e * N D * 1-I i C i
(1)
ND is the calibration factor of the diode obtained under reference irradiation conditions [8] and the Crvalues are correction factors to modify ND for non-reference circumstances. The correction factors Ci applied in this equation originate mainly from the variation of the sensitivity of the diode with dose per pulse value [20] and photon energy spectrum [8]. Details about the different correction factors and the experimental procedure to determine numerical values for these factors, can be found elsewhere [8]. It should be noted that the ND and Ci values are generally different for the entrance and exit point. For this study it was necessary to determine an additional correction factor, Cair gap, required for diodes positioned in the AP open beam at the exit side of the patient. Due to the direct contact of the patient with the treatment couch, it was unfeasible to position a diode between the back of the patient and the couch. This diode was therefore placed on the Melinex sheet, after removing a panel of the treatment couch. This method sometimes yielded a small air gap between the back of the diode and the patient. The change of the sensitivity of the diode by varying the volume of this air gap, which is defined by the exit field size and the distance between the diode and the skin of the patient, has been measured and expressed by the correction factor Ca~r gap. Before each treatment the distance between the diode and the skin of the patient was estimated.
The dose at the midline point can be deduced from the entrance and exit dose values following the method proposed by Rizzotti et al. [22]. The midline point is defined as the point on the central axis of a beam in the middle of the patient. The relation between entrance, exit and midline dose, Den t. . . . . . Oexit and Omidline, r e s p e c t i v e l y , was determined using a polystyrene phantom and a cylindrical ionisation chamber (type NE 2505/3A) and is shown in Fig. 2. For our study this relation was determined for a source-skin-distance (SSD) of 80 cm, a 15 x 15 cm field size at the isocentre and a wedge of 45 °, which are the approximate values for the lateral wedged photon beams in clinical routine. Although for large thicknesses of the patient, i.e. above about 25 cm, the relation depends on these parameters, the use of a single curve introduces only a small error, less than 0.4 ~o, in the determination OfDmidlin e. For the AP and PA photon fields the same relation has been used in the determination of Dmidlin e. In clinical practice the midline point was not always at the isocentre for the three-field technique, particularly for the AP beam. Therefore, a conversion of the midline dose to the dose at the isocentre was performed for each photon beam, using the percentage depth dose (PDD) curve corresponding to the field size at the skin of the patient. The magnitude of this conversion could range up to 5 YD. For 11 patients the entrance and exit dose of each field was determined 3-8 times during her or his treatment of 35 fractions. The entrance and exit diodes were positioned about 1 cm off the central axis of the beam in opposite direction, to avoid the "shadowing" effect of one diode by the other. For wedged beams the diodes were positioned on the axis perpendicular to the wedge profile and the central beam axis, using the cross-wire system of the photon beam. The reduction 1.00
o
(3
0.80
8 ~c uJ
0.60
0.40
3 0.20
0.OO 0.00
~
J
~
J
0.20
0.40
0.60
0.80
1.00
E x i t Dose / Entrance Dose
Fig. 2. The ratio o f the dose value at the midline a n d entrance point vs. the ratio o f the dose value at the exit a n d entrance point as determined in a h o m o g e n e o u s p h a n t o m .
114 of the total dose delivered to the target volume, caused by the attenuation of the primary photon beam by the build-up cap of the entrance diode, was at most 1.5 ~o due to the limited number of measurements with the diodes [ 18,21]. The positioning of the diodes and the conversion of the diode reading to dose values were performed by a technician or a physicist. The performance of in vivo dosimetry increased the treatment time per session with at most 10 min. This increase in time was mainly caused by the positioning of the diodes on the patient for the AP field and both lateral wedged beams which required additional rotations of the gantry. Phantom measurements An essential part of this study was the accompanying dose determination in a polystyrene phantom with dimensions similar to each patient. These measurements served several purposes. Firstly, the measurements allow the assessment of the accuracy and reproducibility of the diode measuring system under clinical conditions. For this purpose entrance and exit dose as well as the dose at the isocentre were determined with the diode measuring system in a similar way as during patient treatment. These dose values were compared with values determined with a Farmer type 0.6-cm 3 cylindrical ionisation chamber (NE 2505/3A) connected to a Keithley electrometer. Secondly, the phantom measurements served to check the accuracy of the dose calculation algorithms, including the algorithm to calculate the number of monitor units, and to determine the uncertainty in the dose prediction due to patient related effects. Both checks have been performed by comparing the phantom dose values with those predicted by dose calculations using our treatment planning system, for the phantom and patient irradiations (see next section). The phantom dose values used in these dose comparisons have been determined by means of ionisation chamber measurements as mentioned before. Finally, additional phantom measurements have been performed to determine the correction factor Cai r gap for the entrance and exit point as a function of the volume of the air gap. The method to determine this factor is similar to those applied for the other correction factors as described elsewhere [8]. During the phantom measurements the patient treatment was simulated as close as possible. The beam set-up parameters were taken identical to those during the actual patient treatment while the thickness of the phantom was equal to the patient thickness as determined during the simulator session of the treatment. In this way patient and phantom dose values could be
compared directly. These phantom measurements will be described as "simulation" measurements in our study. For each patient, simulation measurements have been repeated at least 3 times during the course of the treatment in which each dose value obtained with the ionisation chamber and diode was based on at least 3 readings. The variation in magnitude between consecutive readings was smaller than 0.2~ (1 S.D.) for the ionisation chamber and 0.3~o (1. S.D.) for the diode, respectively. The readings of the ionisation chamber have been converted to dose values by applying the NCS code of practice [ 16]. No correction for variation of stopping power ratios with field size and depth is taken into account in this code of practice. This additional uncertainty has been estimated to be smaller than 0.5% [2]. The simulation measurements were performed on a polystyrene (PS) slab phantom. The dose values determined in this phantom, Dps, have been converted to dose-to-water values, Dwater, because in all dose calculations the patient and phantom material have been assumed to be water equivalent. This conversion has been performed by applying the procedure outlined in the A A P M protocol [ 1 ]. The resulting dose correction was less than 2~o for phantom thicknesses up to 40 cm. Treatment planning Dose values determined in patients and in the phantom have been compared with calculated dose values. For the AP-PA irradiations the number of monitor units to deliver a specified dose at the midline point is calculated from measured depth dose data. The dose distribution in the pelvic region for the three-field irradiation technique is planned in our institution with a treatment planning system (Nucletron Planning System), using 2-D dose calculation algorithms. The patient contour is taken into account using a modification of the effective SSD method, which gives an improvement in the dose calculation for oblique incident beams over conventional algorithms [23]. Tissue inhomogeneities are calculated using the equivalent path length model. In routine clinical practice the dose calculation procedure is simplified: the effect of inhomogeneities such as bone structures are ignored in treatment planning. However, to explain differences observed between in vivo and calculated dose data, the dose distribution for the threefield technique has been recomputed, considering the actual electron densities in the treatment volume in the patient. These electron densities have been based on computer tomographic (CT) information, which was available for five patients. Computation of absolute dose values and monitor units is incorporated in the treatment planning system.
115 For this purpose, output factors for square field sizes and wedge factors have to be added to other input data. In the dose calculation algorithm the output factor as defined in the British Journal of Radiology Supplement 17 [5], has been adopted. Furthermore, the output factor of a rectangular field size is considered to be equal to the output factor of the equivalent square field [5]. In our clinic the wedge factor is defined as the ratio of the dose with and without a wedge in the beam for the same number of monitor units while it is only determined for one irradiation geometry: a depth of 8 cm in a water phantom, a 10 x 10 cm field size at the isocentre and an SSD equal to the source-axis-distance. The dependence of the wedge factor on field size [e.g. see 19] and phantom depth, i.e. beam hardening, is ignored in our planning system. Results
Accuracy and reproducibility of the diode measuring system The accuracy of the diode system can be defined as the ratio of the dose values determined at identical points during the simulation measurements with the diodes and the ionisation chamber. Table I summarises these results for the entrance point, isocentre and exit point for the open and wedged beams. The systematic differences between dose values determined with both measuring systems are smaller than 0.7%. The reproducibility in dose determinations by the diode system, also given in Table I, has been obtained from an analysis of the dose ratios determined for comparable dose measurements performed during different simulation sessions. The reproducibility of the diode measuring system varies between 0.5% (1 S.D.) at the entrance point for open beams to 1.6% (1 S.D.) at the exit point for wedged beams.
Determination of C, grgap The determination of C a i r gap at the entrance side of the phantom yields small values: 1.005 and 1.007 for a thickness of the air gap of 10 mm and 20 ram, respectively, both independent of field size. At the exit side of the phantom the C a i r gap values are depicted in Fig. 3. During the determination of the correction values, the variation of the diode reading due to the effect of the inverse square law, has been eliminated. For clinical situations, i.e. an equivalent square field size with a side of about 15 cm and a thickness of the air gap of about 10 mm, a correction of about 1% has been observed. In addition, the reading of the diode has to be corrected for the effect of the inverse square law which magnitude depends on the clinical situation. Dose comparison: patient dose vs. dose calculation For each patient average dose values at the entrance point, isocentre and exit point have been deduced from all in vivo dose measurements performed for a particular patient. The average dose value at each point has been compared with the expected dose value: Fig. 4 shows for each point and each patient the ratio of the average in vivo dose and calculated dose, both for open and wedged beams. For the lateral wedged beams at the entrance point the actual delivered dose is on the average 3.3 % ( + 1.8 % 1 S.D.) higher than the expected dose value. For the isocentre and exit point such systematic differences have also been observed, 2.8 % for both points, but the standard deviations are larger: 2.4~o and 4.9%, respectively. For the AP-PA beams similar results have been observed although they were less pronounced. The standard deviations belonging to the average in vivo dose ratios are also indicated in Fig. 4. From these data an average variation (1 S.D.) in the dose determination on a patient was derived for each point
m 120 air
7
gap thickness (mm) | /
TABLE I
• = 30
|
o = 23
L
• = 16
Accuracy and reproducibility (1 S.D.) of the diode measuring system, determined during phantom measurements. Position
Entrance Isocentre Exit
Open fields
~= 10
~:=
110
3
Wedged fields
Accuracy
Reproducibility
Accuracy
Reproducibility
1.002
0.5 %
1.000
1.007 1.002
0.9% 0.9°;
1.003 1.006
1.1% 1.5% 1.6%
For the accuracy, dose values have been determined relative to ionisation chamber measurements. The values are based on a total number of 181 and 139 measurements for the open and wedged fields,
1.00
10
15
25 30 side of square field at exit side (cm)
Fig. 3. The correction factor Cair gap for the exit point as a function of field size at the exit surface of the phantom for different air gap thicknesses between the surface and the back of the exit diode. The effect of the inverse square law on the reading of the diode with increasing air gap thickness, has been eliminated.
116
AP-PA
Open Beams
Lateral 1.10
1.10 1.08
A
1.oe
Entrance
1.06
Wedged Beams
13,
1.06
I"
~
~
T
-
Entrance
{
T
1.04 1.02 1.00 0. &
0.911
0.96
O.96
0.94 I~ (/)
0
o.88 0.86
1:3
0.94
@ o') 0
0.92 o.go i
i
i
i
i
i
i
i
i
i
f
l
t
l
r
l
l
l
l
l
l
r'l
l
~
1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 0 1 0 1 0 1 1 1 1
"0
1.1o
-~J ir~
1,08 1.06
0
1.04
-~
1.02
(~
1.00 O.g*g
O')
0.92
o
0.9O
(.)
}~,__!~_~.=
..... ~ .... T_~__
0.86
~
l
l
r
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
l
1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 9 9 1 0 1 0 1 1 1 1
0) ---
1.10
"~
1.08
0-
1,06
i
i
i
i
i
i
i
r
I
I
7
7
8
8
9
9
10
10
11
11
Isocentre
{
1.04
{{
o.~'°°. . . . . .
~"
0.96
O)
0.94
03
0.92
0 C~
0.9o 0.811 0,86
O)
•- -
I
i 6
0
+~
l
i 6
1.1o 1~
I socentre
0.88 -~
o.92 o.9o o.8a
!
I
~--} . . . . . . . . . . . . . . . . .
I
I
I
I
I
I
r
I
I
i
I
i
6
6
7
7
8
8
9
9
I0
10
11
11
1.10
Exit
1.04 1.02 tOO
i
0.98 0.96
rT
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0.98 0.96
0,94 0.~)2
0.92
O.gO
0.9O
0.88 0.8~
0.88 r r I I I i I , i i ~ t L 1 1 2 2 3 3 4 4 5 5 6 6 7 7 8 8 g 9 10101111 I
I
I
I
f
I
r
I
I
0 ~
Patient Number
I
I
r
I
I
I
I
I
I
I
I
I
6
6
7
7
8
8
0
9
10
10
11
11
Patient Number
Fig. 4. The ratio of patient dose and calculated dose values, given for each point of measurement and both open AP-PA beams and both lateral wedged beams.
of dose measurement in the AP-PA open and lateral wedged beams. The results have been tabulated in Table II. As shown in this table, the average variation increases from 0.9% for the entrance point for open beams to 3.1% at the exit point for wedged beams.
Dose comparison: phantom dose vs. dose calculation Figure 5 summarises the results of the phantom dose comparison. In this figure the frequency distribution is given of the average dose measured in the phantom at the entrance point, isocentre and exit point for the open and lateral wedged beams relative to the calculated dose. For open beams, difference have been observed between measured and calculated dose values ranging from - 5 % to 5 %. For wedged beams almost all calculated dose values were systematically lower than the measured values. A maximum difference of + 10 % and + 12% was observed at the isocentre and exit point, respectively.
Discussion
Accuracy and reproducibility of the diode measuring system The good accuracy and reproducibility of the diode system (Table I) show that determination of the entrance and exit dose and the dose at the isocentre can TABLE II Average variation in the dose determination during pelvic treatment of 11 patients applying diodes. Position
Open fields
Wedged fields
Entrance Isocentre Exit
0.9 % 1.3 % 2.2 %
1.7 % 1.5 % 3.1%
The values are based on a total number of 193 and 154 measurements for the open and wedged fields, respectively.
117 Entrance
[
I Isocentre
~
to the wedge. The worst reproducibility of the diode measuring system will therefore be observed for the exit dose measurements in the lateral fields. It has to be noted that the uncertainty in dose determination with the diode system due to variation of the output of the treatment machine is eliminated in these results, due to the simultaneous dose determination with the diodes and the ionisation chamber.
Extt
25
N= 1 7 8
Opera B e a m s 2O
15
10
Determination of C, tr gap 5
>, U C
. . . .
0 0.g5
1.00
1.05
,
. . . .
1.10
,
1.15
O-
kS 25
9Vedged B e a m s
N=142
2O
15
10
5
In order to determine with high accuracy patient dose values, the additional correction factor Cai~ gap has been applied in the reading-to-dose conversion of diodes positioned in the AP field. For entrance dose determinations this factor could almost be ignored. For the exit dose determination the air gap could be about 10 mm introducing a total correction, i.e. including a correction for the inverse square law, on the diode reading up to 3~o for field sizes applied in the pelvic treatment technique. The correction factor Cair gap originates from a lack of phantom scatter, i.e. from electrons, compared with the non-air gap situation. For increasing field size this build-down effect in the region near the diode, is partially compensated by electrons coming from a larger distance from the diode.
0
0.g5
1 C0
105
1.10
1.15
>hantom Dose / Calculated Dose Fig. 5. The frequency of the ratio of the measured dose and the dose calculated by the computer planning system for phantom irradiations, given for each point of measurement in the open AP-PA beams and the lateral wedged beams. N is the number of phantom measurements.
be performed with p-type diodes almost as accurately as performed with an ionisation chamber. Accurate diode dose measurements require besides a well known dose calibration factor ND, a large number of correction factors Ci on the diode reading both fore the entrance and for the exit point [8,15,18,21]. For the exit point these correction factors are generally more pronounced and vary more strongly, e.g. with field size and depth compared with similar correction factors required for the entrance dose determination. The reproducibility of exit dose measurements is therefore somewhat less compared with entrance dose measurements. This is especially the case for the irradiation of patients of large size, e.g. up to 40 cm in the lateral direction, applying wedged photon beams with a wedge angle equal or larger than 45 °. Furthermore, for wedged beams an additional uncertainty of about 17o (1 S.D.) is introduced due to the positioning of the diode with respect
Dose comparison: patient dose vs. dose calculation
The deviations between the in vivo dose and calculated dose values as depicted in Fig. 4 can be explained by limitations in the dose calculation algorithms and by patient related effects. The limitations of the algorithms have been quantified by comparing measured dose values with calculated values for the phantom irradiations, yielding underestimations up to 12 ~o for the calculated dose at the exit point in the wedged lateral beams (see Fig. 5). Uncertainties in dose computations due to patient related effects were quantified from a comparison TABLE III Maximum deviation between the dose calculated by the treatment planning system and the dose measured in a phantom, for the lateral fields of pelvic irradiations.
Output calculation Output factor Collimator exchange effect Wedge factor (field size) Wedge factor (SSD)
-
1.6% 1.0yo 2.0% 1.0%
Calculation of the depth dose distribution PDD wedged beam Lack of backscatter Overall difference
Entrance point + 1.0%
Isocentre 1.63;
Exit point -4.1~ + 1.3%
- 4.6 %
- 7.~,/o
- 8.4%
118 of the in vivo and phantom dose values. Analysis of this part of the dose comparison showed that these uncertainties in dose calculations were mainly introduced by the non-water equivalence of the treatment volume of the patient, i.e. the presence of fat and bone structures in the patients. The influence of these inhomogeneities on the dose prediction has been estimated from dose calculations using the actual electron densities in the target volume, available for five patients. For the open beams the overall effect was negligible while for the exit points in the wedged lateral beams a reduction up to 8.5~o of the calculated dose values was observed in patient no. 7 (see Fig. 4) if the electron densities were take into account. This reduction is caused by the high density of the femoral heads which overcompensates the lower density of fat tissue relative to water. It has to be noted that the presence of inhomogeneities in the treatment field will introduce an error in the determination of the dose at the isocentre applying the curve given in Fig. 2, as this curve is only valid for homogeneous geometries. However, as the diodes were positioned in the lateral wedged beams behind both femoral heads, this error can be assumed to be negligible, due to the symmetrical distribution of almost similar inhomogeneities. For the AP, PA beams an uncertainty up to about 0.6~o in the dose value at the midline point has been estimated due to the asymmetrical position of the sacral bone in these fields. For very accurate dose determinations the exact position of the diodes with respect to these inhomogeneities has to be determined by making megavoltage images of a patient. In clinical practice the total dose at the isocentre and its deviation from the expected dose value are important and therefore have to be assessed. About half the dose at this point is given by the AP or PA beam, for which a small average deviation of 1.0~o has been observed between the actual and expected dose values. For the lateral wedged beams an average difference of 2.8~o has been observed, yielding on the average a small but systematic overestimation of 1.8~o of the actual dose value at the isocentre for the combination of the AP or PA and lateral photon beams. Comparison of the magnitude of the average variations in in vivo dose measurements (Table II) with the reproducibility in dose determination of the diode measuring system alone (Table I), shows that daily variation in the patient positioning and patient movements decrease the reproducibility of in vivo dose determinations. Especially for dose determinations in wedged beams, the positioning of the diode and patient movement will decrease the precision due to the unknown movement of the diode over the wedge profile.
Dose comparison: phantom dose vs. dose calculation The large deviations between the actual phantom dose values and the expected dose data, especially for the wedged beams (Fig. 5), show that limitations in the dose calculation algorithms introduce large differences between the computed dose values and the in vivo dose results. A quantification of these limitations has been obtained from a further investigation of the accuracy of the dose calculation algorithms, and is given in Table III. Four effects are related to limitations in the algorithm of the absolute dose calculation at dma x (output computation) whil~ two other effects are related to the calculation of the relative dose distribution. Firstly, the increase of the output of an accelerator with increasing field size originates from two effects: an increase of the fluence of the primary photon beam, i.e. due to an increase of the head scatter [ 12], and an increase in the phantom scattered dose contribution. Considering the output variation only due to variation in the phantom scattered dose contribution, an error up to -1.6~o in the calculated output was introduced for an SSD of 80 cm. Secondly, computation of the output factor for a rectangular field by the method of the equivalent square field [ 5 ], introduced an error of - 1.0 ~o due to the collimator exchange effect [ 12]. Thirdly, the wedge factor depends on the field size [e.g. see 19]. Neglecting this effect resulted in an underestimation of the actual output of the accelerator of about 1.5 ~/o for the field sizes used during pelvic irradiations. The final refinement in the absolute dose calculation algorithm concerns the increase of the wedge factor with decreasing SSD which resulted in an underestimation of the actual output for wedged beams of about 1.0~o if the SSD changes from 100 cm to 80 cm. The limitations related to the calculation of the relative dose distributions are mainly due to neglecting the depth dependence of the wedge factor, yielding an error in dose computation up to about -4.0~o for the exit points in the lateral wedged beams. In addition, the lack of backscatter at the exit points has not been considered in the treatment planning process which results in too high an estimation of the actual dose at these points with about 1.3~o.
Conclusions
In this study the accuracy and precision of in vivo dose determinations using a p-type diode measuring system have been explored for patients treated in the pelvic region. A high degree of accuracy and precision in dose determination was observed which required the determination of a large set of correction factors on the dose
119 calibration factor determined under reference irradiation conditions. The dose comparison performed in this study shows that even for the simple irradiation geometry of the pelvic region, large deviations have been observed between dose values determined at identical points in patients on a phantom and calculated by the computer planning system. These deviations were much more pronounced for wedged beams than for open beams. They originate firstly from limitations in the absolute dose calculation algorithm, i.e. in the output computation, resulting in an uncertainty in the number of monitor units. Quality control of treatment planning systems should therefore always include tests of the algorithms concerning the calculation of the absolute dose and the number of monitor units. Special attention has to be given to the head and phantom scatter component of the output of the accelerator and the wedge factor as a function of field size and SSD. Additional errors in the routine dose calculation are introduced due to limitations in the computation of the relative dose distribution, e.g. in the P D D of wedged beams. It is important to note that correction of the algorithm which calculates the number of monitor units, to improve the dose prediction for a certain technique, might deteriorate the dose computation for another technique. In our institution such a result has been observed for the treatment of breast carcinoma [9]. Adaptations of dose calculation algorithms should
therefore be accompanied by verification of the calculated dose for various important treatment techniques. Our study also demonstrated that the prediction of the actual dose delivery at the isocentre can be improved if the electron densities of fat and bone structures in the treatment volume of the patient, obtained from CT information, are taken into account. The results of this study show that improvement of the dose calculation procedure can be obtained from accurate in vivo dose measurements at the entrance and exit side of a patient. An in vivo dosimetry program should always include phantom measurements to quantify the accuracy of the measuring system and to separate errors in the dose calculation algorithms from effects related to the patient. The results of this study have guided us in investigating more complicated treatment techniques such as the treatment of breast carcinoma [9] and prostatic cancer using the simultaneous boost technique [13]. Accurate in vivo dosimetry in combination with portal imaging are indispensable tools to perform these studies.
Acknowledgements We would like to thank Ir A. A. M. Hart for useful discussion and Ing. J. Weeda for his help during some of the phantom measurements. This work was financially supported by the Netherlands Cancer Foundation (NKB Grant 87-13).
References 1 AAPM Task Group 21, American Association of Phycisists in Medicine. A protocol for the determination of absorbed dose from high energy photon and electron beams. Med. Phys. 10: 741-771, 1983. 2 Andreo, P. and Brahme, A. Stopping power datafor high-energy photon beams. Phys. Med. Biol. 31: 839-858, 1986. 3 Bascuas, J.L., Chavaudra, J.P., Vauthier, G. and Dutreix, J. Inter6t des measures "in vivo" systematiques en radioth6rapie. J. Radiol. Electrol. 58: 701-708, 1977. 4 Brahme, A., Chavaudra, J., Landberg, T., McCullough, E., Niisslin, F., Rawlinson, A., Svensson, G. and Svensson, H. Accuracy requirements and quality assurance of external beam therapy with photons and electrons. Acta Oncol. Suppl. 1, 1988. 5 Central Axis Depth Dose Data for Use in Radiotherapy. In: Br. J. Radiol. Suppl. 17, p. 145. Editors: D. K. Bewley, A. L. Bradshaw, D. Greene, J. L. Haybittle and L. F. Secretan. The British Institute of Radiology, 1983. 6 Dutreix, A. When and how can we improve precision in radiotherapy? Radiother. Oncol. 2: 275-292, 1984. 7 Goitein, M. Nonstandard deviations. Med. Phys. 10: 709-711, 1983. 8 Heukelom, S., Lanson, J. H. and Mijnheer, B.J. Comparison of entrance and exit dose measurements using ionization
9
10
11
12 13
14
15
chambers and silicon diodes. Phys. Med. Biol. 36: 47-59, 1991. Heukelom, S., Lanson, J. H., van Tienhoven, G. and Mijnheer, B.J. In vivo dosimetry during tangential breast treatment. Radiother. Oncol. 22: 269-279, 1991. ICRU, International Commission on Radiation Units and Measurements. Determination of absorbed dose in a patient irradiated by beams of X or gamma rays in radiotherapy procedures. Report 24, ICRU Publications, Washington, DC, 1976. ICRU, International Commission on Radiation Units and Measurements. Dose specification for reporting external beam therapy with photons and electrons. Report 29, ICRU Publications, Washington, DC, 1978. Kase, K. R. and Svensson, G . K . Head scatter data for several linear accelerators (4-18 MV). Med. Phys. 13: 530-532, 1986. Lebesque, J.V. and Keus, R. The simultaneous boost technique; the concept of relative normalized total dose. Radiother. Oncol. 22: 45-55, 1991. Leunens, G., van Dam, J., Dutreix, A. and van der Schueren, E. Quality assurance in radiotherapy by in vivo dosimetry. 1. Entrance dose measurements, a reliable procedure. Radiother. Oncol. 17: 141-151, 1990. Leunens, G., van Dam, J., Dutreix, A. and van der Schueren,
120
16
17
18
19
E. Quality assurance in radiotherapy by in vivo dosimetry. 2. Determination of the target absorbed dose. Radiother. Oncol. 19: 73-87, 1990. Mijnheer, B. J., Aalbers, A. H. L., Visser, A. G. and Wittk~nper, F.W. Consistency and simplicity in the determination of absorbed dose to water in high energy photon beams: A new code of practice. Radiother. Oncol. 7: 371-384, 1986. Mijnheer, B.J., Battermann, J. J. and Wambersie, A. What degree of accuracy is required and can be achieved in photon and neutron therapy? Radiother. Oncol. 8: 237-252, 1987. Nilsson, B., Ruden, B-I. and Sorcini, B. Characteristics of silicon diodes as patient dosemeters in external radiation therapy. Radiother. Oncol. 11: 279-288, 1988. Palta, J. R., Daftari, I. and Suntharalingam, N. Field size dependence of wedge factors. Med. Phys. 15: 624-626, 1988.
20 Rikner, G. and Grusell, E. Effects of radiation damage on p-type silicon detectors. Phys. Med. Biol. 28: 1261-1267, 1983. 21 Rikner, G. and Grusell, E. Patient dose measurements in photon fields by means of silicon semiconductor detectors. Med. Phys. 14: 870-873, 1987. 22 Rizzotti, A., Compri, C. and Garusi, G . F . Dose evaluation to patients irradiated by 6°Co beams, by means of direct measurement on the incident and on the exit surfaces. Radiother. Oncol. 3: 279-283, 1985. 23 Westermann, C. F. and Mijnheer, B.J. Dose correction for oblique incidence of megavoltage photon beams. In: Proc. 8th Int. Conf. on the Use of Computers in Radiation Therapy, Toronto, Canada, pp. 335-338, 1984.