Uncertainties in CT-based radiation therapy treatment planning associated with patient breathing

Uncertainties in CT-based radiation therapy treatment planning associated with patient breathing

Int. J. Radiation Biol. Phys.. Vol. 36. No. I, pp. 167-174. 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0360...

2MB Sizes 1 Downloads 48 Views

Int. J. Radiation

Biol.

Phys.. Vol. 36. No. I, pp. 167-174. 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0360.3016/96 $15.00 + .OO

PII: SO360-3016(96)00275-l

ELSEVIER

l

Oncology

Technical Innovations and Notes UNCERTAINTIES IN CT-BASED PLANNING ASSOCIATED

RADIATION THERAPY TREATMENT WITH PATIENT BREATHING

JAMES M. BALTER, PH.D., RANDALL K. TEN HAKEN, PH.D., THEODORE S. LAWRENCE, M.D., PH.D., KWOK L. LAM, PH.D. AND JOHN M. ROBERTSON, M.D. Department of Radiation Oncology, University of Michigan, Ann Arbor, MI Purpose: To evaluate uncertainties associated with treatment-planning computed tomography (CT) data obtained with the patient breathing freely. Methods and Materials: Patients with thoracic or abdominal tumors underwent a standard treatment-planning CT study while breathing quietly and freely, followed by CT scans while holding their breath at normal inhalation and normal exhalation. Identical treatment plans on all three CT data sets for each patient pointed out differences in: (a) radiation path lengths; (b) positions of the organs; (c) physical volumes of the lung, liver, and kidneys; (d) the interpretation of plan evaluation tools such as dose-volume histograms and normal tissue complication probability (NTCP) models; and (e) how well the planning CT data set represented the average of the inhalation and exhalation studies. Results: Inhalation and exhalation data differ in terms of radiation path length (nearly one quarter of the cases had path-length differences >l cm), although the free breathing and average path lengths do not exhibit large differences (O-9 mm). Liver and kidney movements averaged 2 cm, whereas differences between the free breathing and average positions averaged 0.6 cm. The physical volume of the liver between the free breathing and static studies varied by as much as 12%. The NTCP calculations on exhale and inhale studies varied from 3 to 43% for doses that resulted in a 15% NTCP on the free-breathing studies. Conclusion: Free-breathing CT studies may improperly estimate the position and volume of critical structures, and thus may mislead evaluation of plans based on such volume dependent criteria such as dose-volume histograms and NTCP calculations.

INTRODUCTION

resents an important possible source of patient model uncertainties. Although now of undisputed benefit for treatment of many tumors with radiation, a number of factors can compromise the integrity of the CT scan for use in radiation therapy treatment planning (10). For example, the limited slice thickness and separation typical in treatment-planning CT studies may lead to partial volume averaging of structures. This could result in errors in determining the size and position of such structures, and may improperly represent sharp interfaces. In addition, interfaces of media of strongly dissimilar densities can lead to reconstruction artifacts which confound the determination of structure boundaries as well as electron density in slices containing these artifacts.

Recent advances in radiation therapy treatment planning and dose delivery now allow the practical calculation and delivery of dosedistributions that tightly conform to target volumes (16). Prediction of the dose delivered by a complex beam arrangement relies on an accurate description of the patient. Errors in the model of the patient used for treatment planning may lead to erroneous estimations of the position and dosimetric involvement of critical structures in or near the high-dose region, as well as radiation path lengths to the target. These effects may hinder objective comparison and evaluation of treatment plans, thus reducing the use of conformal therapy. The computed tomography (CT) data set used for treatment planning rep-

Presentedin part at the 36th Annual Meeting of the American Society of TherapeuticRadiology and Oncology, 2-6 October 1994,San Francisco,CA. Reprintrequeststo: RandallK. Ten Haken,Ph.D., Department of RadiationOncology, University of Michigan Medical Center, UH-B2C490-Box 0010. 1500 E. Medical Center Drive, Ann Arbor, MI 48109-0010.

authorsthank Lon Marsh andPaulArcherfor their valuableeffort in analysisof the treatmentplanning informationfor this study. They alsoacknowledgeLeslieQuint, M.D., for assistance in study design.This work was supported by NC1 Grant POl-CA59827. Acceptedfor publication20 May 1996.

Acknowledgements-The

167

168

I. J. Radiation Oncology l Biology l Physics

The CT studiesof abdominal and thoracic structuresmay also suffer becauseof the effects of the patient breathing. Henkelman and Mah reported on the importance of breathing in radiation therapy for the thorax (4). Typically, in many institutions patients breathe freely during treatment-planning CT studies, under the assumption that this will average out the effects of breathing. Because of the short sampling time of CT scanners(<2 s/slice) and the longer period of ventilation, free breathing may not significantly blur image resolution, but it may introduce significant geometric uncertainties into the CT data. For example (Fig. l), if a structure moves superiorly or inferiorly between slice acquisitions, the sequenceof imagesmay become disordered, with inferiorly indexed (relative to the table) slicessampling superiorportions of the structure (and vice versa). This can introduce nonphysical features into the CT data set such as a portion of the diaphragm extending into lung tissue (Fig. 2a), or a jagged appearanceto surfacesgenerated from the scans(Fig. 2b). The position of critical structures with respect to the treatment isocenter greatly affects the selection of beam arrangements. Although a given treatment portal may appear to miss a critical structure on the planning CT study, ventilation may actually move critical structures into and out of the radiation field. Similarly, as a patient breathes, the amount of densetissue in the beam path may change. thus altering the radiation path length to the center of the tumor. Calculations of beam weightings and final dose distributions use these path-length data. Finally, the previously mentioned artifacts may significantly affect estimations of the volumes of critical structures. As dosevolume histograms (DVHs) and plan evaluation indices such as normal tissue complication probability (NTCP) often usethe relative volume of an organ irradiated to high doses, errors in measurement of the total organ volume may lead to erroneous plan evaluation conclusions. This study explores the impact of patient breathing on the CT data set used in treatment planning. Investigations concern the effect of patient ventilation on organ position, organ volume, radiation path length, and predicted outcome of treatment. METHODS

AND MATERIALS

Nine patients participated in this study as an adjunct to their normal treatment-planning CT process. Six patients had abdominal tumors and three had lung tumors. Abdominal scansencompassedthe entire liver and both kidneys; thoracic scansincluded all lung tissue. Each treatment-planning CT study was composedof the acquisitions of three data sets.During the first (treatmentplanning data set), patients breathed quietly and freely. For the second and third acquisitions, patients practiced breathing normally and then held their breaths at levels of normal inhalation and exhalation, respectively. The useof a spiral CT scanner permitted continuous acquisition of CT slices at a rate of 1 slice/s. Breaking the breath hold

Volume 36, Number I, 1996

Fig. 1. Artifacts in CT samplingdue to breathing.The inferiorsuperiormovementof the object during breathingleadsto improper sampling of the structure on CT.

data sets down into sequential subsetsof 10 slices each helped ensure that patients could comfortably hold their breaths during the scan. Sequential subsetshad overlapping slices to ensure that patients inhaled or exhaled to the samelevel. The slice separation ranged from 7 to 10 mm. Maintenance of patient position (using low-density foam cradles) throughout the acquisitions of the three CT data sets(at a single CT scanning session)helped to minimize uncertainties due to patient setup. Semiautomatic and manual segmentation routines helped to define organ volumes using standard three-dimensional (3D) treatment-planning methods (3, 22). A boundary-tracking routine based on the large differences in density between lung and normal tissue automatically delineated lung surface contours on sequential transaxial scans.Experienced dosimetrists interactively defined surface contours for liver and kidney volumes. Definition of all relevant volumes at a single planning sessionhelped reduce possibleerrors due to ambiguities in image display and anatomic interfaces. Treatment planning involved the generation of 3D dose distributions for beam arrangements designed within our treatment planning system (2) using the 3D representation of the CT data described above (from the planning CT data set only) and beam’s eye view tools (13). Extensive efforts helped minimize the volume of liver and/or lung tissue exposed to high doses of radiation in accordance with ongoing dose escalation studies using conformal therapy (9, 12). This often required the use of noncoplanar beam arrangements (22). Sequential generation of all three scansat the sameCT session,together with instructions to patients not to move between scans,led to the expectation of little or no change in patient position among the scans.Observations of patient anatomy on CT slices in the axial plane and CT slices reconstructed in the orthogonal coronal, and sagittal planes helped to verify this assumption. In particular, surfaces generated for the spine, external anatomy, and su-

Uncertainties in CT data due to breathingl

J. M. BALTER

et a/.

169

(a)

(b) Fig. 2. Effects of CT sampling without suspended ventilation. (a) The coronal slice on the right (reconstructed along the line through the axial slice on the left) shows a nonphysical jagged extension of the diaphragm (near the arrow) resulting from scan acquisitions at different parts of the ventilatory cycle. (b) Lung surfaces derived from contours generated in the planning CT data set exhibit a jagged appearance due to scan acquisitions at different parts of the ventilator-y cycle.

perior portion of the lungs from the normal planning data set intersected CT cuts generated through the inhale and exhale CT data sets at the correct positions. Verification of data set alignment allowed the direct transfer of beam arrangements from the planning CT study to the other two suspended ventilation CT data sets for purposes of dose calculation and plan evaluation. Examinations of differences among the radiologic path lengths (sum of the products of physical path length and electron density relative to water for each tissue segment along the ray from the source to the treatment isocenter) and tissue-phantom ratios (TPRs) computed for each beam

(5) in each CT data set helped point out possible dosimetric differences due to patient breathing. Comparison of the average radiologic path length between inhalation and exhalation to that measured from the planning CT scan helped to point out the validity of doses computed on the planning scan. The ratios of TPRs determined for the different radiologic path lengths for the treatment beams indicated the magnitude of the resulting dosimetric changes. Comparisons of the locations of the centers of mass of the liver and kidneys between the two breath-hold data sets indicated movement of these structures under normal

I. J. Radiation Oncology 0 Biology 0 Physics

Inhale

Exhale

Fig. 5. Example of organ motion. The right kidney moves into and out of the treatment field as the patient breathes. The uniform inferior-superior expansion of the target for ventilation increases me volume of the high-dose region in the region inferior to the target, into which the tumor does not move.

Fig. 3. Differences in radiation path length due to breathing. The planned anterior-inferior oblique beam shows a path-length difference of 1.5 cm between the inhale (outer external contour line) and exhale (inner contour line and gray-scale image) CT scans.

ventilation.

structures (calculated as the midpoint of the line that connected the centers of the organ at inhale and exhale) to the position of the center of the structure on the planning CT scanindicated how well the planning scan represented the average position of each structure. Comparisons of the volumes (cm3) of the lung, liver, and kidneys computed from the breath-hold CT studies indicated both real and apparent changesin organ volume with patient ventilation, Comparisons to the volume determined from the planning CT data set indicated how well the planning study representedthe average organ volume. The NTCP calculations for the liver and lung from each of the three CT data sets for each patient used the model

Patients

b) Tboracic

Patients

I2 I



I-;. I5 s Ill 1 Path change1

2tJ (mm)

RESULTS

of the average positions of these

Comparisons

a) Abdominal

by Lyman (11) together with the effective volume DVH reduction scheme (7) and model parameters determined from previous studies(8, 12). After adjusting the isocenter dosesto generate NTCP values of 15% for liver or lung from the treatment-planning CT data set, the results of NTCP calculations using these samedosesin the breathhold data sets pointed out differences in these biologic predictions due to patient breathing.

25



5

IO I5

1 Path changel

(mm)

Fig. 4. Magnitude of path-length differences between inhale and exhale CT scans of patients treated for (a) abdominal and (b) thoracic tumors.

Radiologic

path lengths

A beam entering the patient from the anterior-inferior (AP) direction for treatment of a liver tumor helps to illustrate (Fig. 3) a change in radiation path length. The density-weighted path length to the isocenter via this beam along the central axis changesfrom 8.7 to 7.2 cm between the inhale and exhale static CT scans (inhale and exhale data setsused the sameisocenter coordinates). The frequency distributions of radiologic path-length differences between inhalation and exhalation studies for patients with both abdominal and thoracic tumors (Fig. 4) indicates the observation of generally small changes, although 24% of planned beams showed differences in path length exceeding 1 cm. The average of inhalation and exhalation path lengths compared favorably to the path length from the planning CT data set; for abdominal and thoracic sites combined, the magnitude of the differences averaged -0.5 mm with a standard deviation of 4 mm. Path-length changescan lead to improper beam weightings. Lower-energy photon beams for targets at a depth greater than the depth of maximum doseexhibit this effect to a greater extent. Typical beamsfor liver patients in this study involved beam energies of 15 MV, field sizes of lo- 12 cm, and depths of lo- 15 cm. Within this range of fields, a l-cm difference in path length would lead to a change of 2-3% in TPR. A 5-mm change (more typical of the difference in path length between the average of the static scans and the planning CT) results in a 1- 1.5% change in TPR. Although the range of depths and field sizes for lung tumors varies more than for liver tumors in the patients studied, similar TPR changesresult.

171

Uncertainties in CT data due to breathing 0 J. M. BALTER rf cd.

Table 2. Conservation of organ volume between inhale (i). exhale (e), and free-breathing states (70)

Table 1. Changes in organ position betweeninhalation(i) and exhalation (e), and between the averageand

free-breathingstates Average (i - e) (cm) Liver Right kidney Left kidney

1.7 1.8 1.8

a(Average

- Plan)

(cm) 0.5 0.6 0.6

Orgun position The beam’s eye view display of an AP field derived from the planning CT of a patient with a liver tumor (Fig. 5) helps to illustrate changes in organ position due to patient ventilation. For this case, the right kidney appearsto be sparedon the inhale CT, but moves into the path of the beam as the patient exhales. In addition, the block margin added for patient breathing (inferior and superior margins determined from observation of patient breathing under fluoroscopy at the treatment simulator) does ensure that the clinical target volume remains in the field. However, because in this case the planning CT scan more closely approximated the inhale breath-hold CT data set, the margin added for breathing in the inferior portion of the field unnecessarily treats normal tissue. For the patients studied (Table 1), abdominal structures such as the kidneys and the liver appeared to move an average of 1.5-2 cm between inhalation and exhalation. In some cases, the volume of liver and/or kidney in the radiation field changed noticeably with ventilation. Organ volume Observation of the liver volume determined for a sample liver patient (Fig. 6) helps to illustrate the effects of

Planning

Liver Right kidney Left kidney

Ii ~ el

dli - cl/e)

/ Average - Plan 1

4 4 3-

5 5 2

12 7 8

patient ventilation on organ volumes obtained from free breathing CT scans.The liver representedby the planning CT study exhibits considerable irregularity introduced by breathing (e.g., note Figs. 1 and 2). The liver volumes determined from the breath-hold CT data sets (1838 and 1872 cm3 from the inhalation and exhalation scans, respectively) greatly exceed the volume (1488 cm3) obtained from the planning CT data set. In this casethe freebreathing CT scan apparently under samples both the superior and inferior portions of the liver. Overall, the free-breathing scansfor the population of patients studied both under sampled and over sampled liver volumes. In general, for the patients studied here, the breath-hold data setsindicate (Table 2) good conservation of the physical liver and kidney volumes between different statesof breathing. However, the difference between the average volume of these organs and the volume measured from the planning CT can be considerable (especially for the liver), as indicated above. The 12% average difference in liver volumes between static and free-breathing CT studies can be accounted for as the addition or removal of l2 CT slices of the superior portion of the liver, consistent with the sampling problems described in Fig. 1. The change in volume between the exhale and inhale CT volumes averaged 945 cm3 for the lung patients studied. Excluding one patient with a 16Xcm3 volume

Inhale

Exhale

Fig. 6. Differences in organ volume determination due to patient breathing. The planning (free-breathing) surface exhibits a more jagged appearance as well as a smaller overall volume than the static (breath-hold) surfaces.

liver liver

172

I. J. Radiation Oncology

l

Biology

l

Physics

the time course of breathing and the actual volume of tissue involved in the high-dose region over the course of plan delivery. Changes in the position and density of elements of the lung volume with breathing do not make an analysis straightforward. Here, the three patients studied showed more agreement between the planning CT study and the exhale study than with the average of the inhale and exhale studies; however, low patient numbers preclude further analysis.

100 Planning CT

--mm=

Inhale CT

0 10 20 30 40 50 60 70 80 90 100

Dose (%) (4 -

Planning Scan

11~~~~~~~ Exhale

mm===

0

20

40

60

Scan

InhaleScan

80

100

Dose (%) 04 Fig. 7. (a) Dose-volume histograms for the lung based on ning CT data (solid line), inhale static CT data (dashed and exhale static CT data (dotted line). (b) Dose-volume tograms for the liver based on planning CT data (solid inhale static CT data (dashed line), and exhale static CT (dotted line).

Volume 36, Number 1, 1996

planline), hisline), data

change, the average difference becomes589 cm3. The difference in volume between the free-breathing and inhale lung volumes ranged from 25 to 46% of the lung volume from the free-breathing CT study. The static exhale lung volumes more closely matched the free-breathing volumes, with changes ranging from 0 to 5%. By definition, lung volumes change during patient breathing, making the description of an average lung volume complex. An appropriate representation of the lung for plan evaluation purposesshould take into account considerations such as

Dose evaluation The dose-volume histograms from the three CT data setsof samplepatients (Fig. 7) help illustrate the effect of organ volume and position changes on treatment plan evaluation. A planning CT which representsthe average status of the organ under consideration should lead to a DVH which falls between those generated from the inhalation and exhalation data (Fig. 7a). However, errors in the physical volume of an organ derived from a freebreathing CT data set (Fig. 6) can lead to cases(Fig. 7b) where the DVH for the planning CT appearsto be worse than either the inhalation or exhalation case. Because of both changes in organ position and organ volume, computed NTCP values change relative to the state of patient breathing used to obtain the input data (Table 3). The liver NTCPs do not tend to increase or decreasesystematically as a function of inhalation owing to the varying location of tumors in the liver between patients. However, Patient 5 had a smaller liver on the planning CT than on either the inhale or exhale CT because of unfortunate sampling during the planning scan, as discussedabove, and thus had reduced NTCPs for both inhale and exhale states. That is, the irradiated liver volume remained approximately the same for all three studies, but the unirradiated volume increased for the larger overall liver volumes on the static scansleading to smaller effective volumes with the DVH reduction scheme used. and hence lower NTCP values. As expected, the lung patients all exhibited a reduction in NTCP based on the inhale study (again, smaller effective volumes due to increased overall lung volumes at inhale).

Table 3. Normal tissue complication probabilities (NTCPs) for inhale and exhalestudiesbasedon dosesthat generate15% NTCP on the planning study (%) Patient Liver Liver Liver Liver Liver Liver Lung Lung Lung

1 2 3 4 5 6 1 2 3

Inhale NTCP

Exhale NTCP

9.5 43.7 19.2 29.9 3.1 15.5 12.9 2.7 6.6

15.6 4.4 4.5 14.2 7.0 12.3 16.6 17.8 25.4

Uncertainties

in CT data due to breathing

DISCUSSION The differences between the inhale and exhale static CT scans indicate that the changes associated with normal ventilation may have a significant impact on treatment planning data. These effects bring into question the validity of free-breathing CT data as an accurate model of the patient for treatment planning. Suramo et al. (19) reported craniocaudal movements of the liver and kidneys up to 5.5 cm during maximum ventilation and 2.5 cm during normal ventilation. Schwartz et ul. (18) reported movements of the kidneys up to 4 cm during deep ventilation. In a magnetic resonance imaging (MRI) study, Moerland et al. (14) described displacements of the left and right kidneys up to 2.4 and 3.5 mm, respectively, under normal respiration conditions, with increases up to 6.6-8.6 cm under forced respiration. Korin et al. (6) performed a quantitative study to measure the relative longitudinal (S/I) and transverse (R/L and A/P) displacements of the upper abdominal organs of 15 volunteers during breathing using an MRI line scan technique. Their results showed primarily translational motion of organs, predominantly along the longitudinal direction (average ratio of S/l movement to R/L or A/P movement = 6.0 and 5.0. respectively, for normal breathing). The average amplitude of the S/I motion (1.3 cm) increased (to 3.9 cm) with deep breathing, with little abdominal organ dilation. Davies et nl. (1) also observed motion of the liver with quiet ventilation of 0.7-2.8 cm, again primarily in the S/I direction. Our results support the conclusions of these other reports. In their study. Henkelman and Mah observed (4) changes in radiologic path length on the order of 5% for the lung during normal respiration. Willett ef ul. (25) studied changes in mediastinal width to a thoracic diameter of 3- 11% when quiet breathing changed to deep respiration; this lead, on average, to 8% more lung volume protected at deep inspiration for their mantle treatments. More recently, Ross et al. ( 17) studied tumor motion in 20 patients with intrathoracic neoplasms using ultrafast (tine) CT. They observed tumor movement as a consequence of both cardiac and respiratory activity, with the greatest movement (up to 2 cm; average near 1 cm) for lesions located adjacent to the heart or aorta or near the diaphragm. Our results support the conclusions of these other reports. Measurements made here to determine the difference between the average of inhale and exhale CT data and free-breathing CT data in terms of radiation path length, organ position, and organ volume assumed that the inhale and exhale data should be equally weighted (that is, patients breathe uniformly between inhale and exhale, and the breath-hold inhale and exhale scans obtained appropriately represent these two fractions of the ventilatory cycle). The actual time course of breathing, however, may vary from patient to patient. A strong bias in a breathing cycle (e.g.. the patient spending a significant portion of

0 J. M.

BALTER

er 01.

173

time at or near exhale) may skew our estimation of average away from the time-weighted average. Although weighted averages may affect the changes in radiation path lengths and organ positions stated, they would not significantly affect the analyses of changes in apparent organ shapes and physical volumes due to the state of patient breathing at the time of CT data acquisition, some of the more significant effects observed here. Although instructed to inhale and exhale normally, some patients in the current study (e.g., Lung Patient 2 showed a 1658 cm’ change in volume between inhale and exhale) may be more representative of forced inhalation. This inability to monitor patient breathing in a stringent fashion may slightly affect the quantitative aspects of the current study but should not distract from the overall conclusion. Future studies will use external indices such as spirometry to indicate the level of ventilation before scanning. Observations of organ position changes with patient breathing such as those presented here should lead to further consideration of normal tissue structure movement in treatment planning for the thorax and abdomen, as has been considered for movement of the bladder, rectum. and prostate in the pelvis (21). Generally, treatment plans in the abdomen and thorax incorporate margins to cover the tumor and avoid high doses to critical structures based on a static CT data set. Although these expansions may ensure tumor volume coverage, such expansions may result (as demonstrated here) in significant involvement of critical structures that may move into the radiation field during ventilation, or unnecessarily irradiate large portions of normal tissue included for expected tumor movement that does not occur. As indicated here, the large variance in NTCP resulting from differences in organ location and volume may seriously affect dose escalation protocols, as well as biologic modeling of dose response. Modeling of biologic responses to radiation of normal structures in the abdomen and thorax may be compromised using current models owing to existing errors in the planning CT data used for correlation with patient outcome. Although the conservation of liver volume during breathing allows the computation of average dose distributions, including the effects of organ motion due to breathing (20), greater difficulty arises in evaluation of lung patients using current NTCP models. Not only does the volume of the lung change during ventilation, but the location and density of local anatomy also change in a nonrigid fashion. As some dose-escalation studies base their design on dose-volume analysis (9, 23-24). errors in the dose-volume data used could affect both the study design (when based on retrospective data) as well as analysis of the results. Clearly, more work in this area should be considered. A more appropriate representation of structures in the abdomen and thorax may be studied using dynamic models which, for example, consolidate static data obtained at known breathing states with other information about the

174

1. J. RadiationOncology 0 Biology 0 Physics

breathing cycle (e.g., the percentage of time spent in each breathing state). The use of dynamic models could serve to validate the use of treatment strategies such as ventilatory gating (15). Prior to the development of dynamic models, it may be beneficial to determine the state of breathing (e.g., exhale) the patient spends the most time

Volume 36, Number 1, 1996 in reproducibly, and to acquire a static CT with the patient at this state of ventilation. The combination of dynamic models of ventilation in the lung with an understanding of local lung density over the course of ventilation may improve the future understanding of radiation response of lung tissue.

REFERENCES 1. Davies, S. C.; Hill, A. L.: Holmes, R. B.; Halliwell. M.: Jackson, P. C. Ultrasound quantitation of respiratory organ motion in the upper abdomen. Br. J. Radio]. 67: 1096-I 102; 1994. 2. Fraass, B. A.; McShan, D. L. 3-D treatment planning. I. Overview of a clinical planning system. In: Bruinvis, I. A. D., ed. The use of computers in radiation therapy. Amsterdam: Elsevier; 1987:273-276. 3. Fraass, B. A.; McShan. D. L. Three-dimensional photon beam treatment planning. In: Smith, A. R., ed. Radiation therapy physics. New York: Springer-Verlag; 1994:43-93. 4. Henkelman, R. M.; Mah, K. How important is breathing in radiation therapy of the thorax? lnt. J. Radiat. Oncol. Biol. Phys. 8:2005-2010; 1982. 5. Khan, F. M. The physics of radiation therapy. Baltimore: Williams and Wilkins; 1994. 6. Korin, H. W.; Ehman, R. L.; Riederer, S. J.; Felmlee, J. P.: Grimm, R. C. Respiratory kinematics of the upper abdominal organs: A quantitative study. Magn. Reson. Med. 23:172178; 1992. 7. Kutcher, G. J.; Burman, C. Calculation of complication probability factors for nonuniform normal tissue irradiation: The effective volume method. lnt. J. Radiat. Oncol. Biol. Phys. 16:1623-1630: 1989. 8. Lawrence, T. S.; Ten Haken, R. K.; Kessler, M. L.; Robertson, J. M.: Lyman, J. T.; Lavigne, M. L.; Brown, M. B.; DuRoss, D. J.; Andrews, J. C.; Ensminger, W. D.; Lichter, A. S. The use of 3-D dose volume analysis to predict radiation hepatitis. lnt. J. Radiat. Oncol. Biol. Phys. 23:781-788; 1992. 9. Lawrence, T. S.; Tesser, R. J.; Ten Haken, R. K. An application of dose volume histograms to treatment of intrahepatic malignancies with radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 19:1041-1047: 1990. 10. Ling, C. C.: Rogers, C. C.; Morton, R. J., eds. Computed tomography in radiation therapy. New York: Raven Press: 1983. 11. Lyman, J. T. Complicationprobability as assessed from dose-volumehistograms.Radiat.Res. 104:SI3-S19; 1985. 12. Martel, M. K.; Ten Haken,R. K.; Hazuka, M. B.; Turrisi, A. T.; Fraass,B. A.; Lichter, A. S. Dose-volumehistogram and3-D treatmentplanningevaluationof patientswith pneumonitis.lnt. J. Radiat.Oncol. Biol. Phys.28:575-581: 1994. 13. McShan,D. L.; Fraass,B. A.: Lichter, A. S. Full integration of the beam’seyeview conceptinto computerizedtreatment planning.lnt. J. Radiat. Oncol. Biol. Phys. 18:1485-1494; 1990. 14. Moerland. M. A.; van den Bergh, A. C. M.; Bhagwandien, R.: Janssen,W. M.; Bakker, C. J. G.: Lagendijk, J. J. W.;

Battermann. J. J. The influence of respiration induced motion of the kidneys on the accuracy of radiotherapy treatment planning, a magnetic resonance imaging study. Radiother. Oncol. 30: 150-l 54; 1994. 1.5. Ohara, K.; Okumura, T.; Akisada, M.; lnada, T.; Mori, T.; Yokota, H.; Calaguas. M. J. B. Irradiation synchronized with respiration gate. lnt. J. Radiat. Oncol. Biol. Phys. 17:853857; 1989. 16. Photon Treatment Planning Collaborative Working Group. State-of-the-art of external photon beam radiation treatment planning. lnt. J. Radiat. Oncol. Biol. Phys. 21:9-23; 1991. 17. Ross. C. S.; Hussey, D. H.; Pennington, E. D.: Stanford, W.; Doornbos, J. F. Analysis of movement of intrathoracic neoplasms using ultra&t computerized tomography. lnt. J. Radiat. Oncol. Biol. Phys. 18:671-677; 1990. 18. Schwartz. L. H.; Richaud, J.; Buffat, L.; Touboul, E.; Schlienger, M. Kidney mobility during respiration. Radiother. Oncol. 32:84-86; 1994. 19. Suramo, I.; Paivansalo, M.: Myllyla. V. Cranio-caudal movements of the liver. pancreas and kidneys in respiration. Acta. Radio]. Diagn. 25: 129-131; 1984. 20. Ten Haken, R. K.; Balter, J. M.; Marsh, L. H.; Robertson, J. M.: Lawrence. T. S. Potential benefits of gating conformal irradiation of liver tumors to the ventilatory cycle. lnt. J. Radiat. Oncol. Biol. Phys. 32(Suppl I ): 187; 1995. 21. Ten Haken. R. K.; Forman, J. D.: Heimburger, D. K.; Gerhardsson, A.; McShan, D. L.; Perez-Tamayo, C.; Schoeppel, S. L.; Lichter. A. S. Treatment planning issues related to prostatemovementin responseto differential filling of the rectum and bladder. lnt. J. Radiat. Oncol. Biol. Phys. 20:1317-1324:1991. 22. Ten Haken,R. K.; Lawrence,T. S.; McShan,D. L.; Tesser, R. J.: Fraass.B. A.; Lichter, A. S. Technicalconsiderations in the useof 3-D beamarrangements in the abdomen.Radiother.Oncol. 22:19-28; 1991. 23. Ten Haken, R. K.; Martel, M. K.; Kessler,M. L.: Hazuka. M. B.; Lawrence,T. S.; Robertson,J. M.; Turrisi, A. T.: Lichter. A. S. Useof Veff and iso-NTCPin the implementation of doseescalationprotocols. Int. J. Radiat. Oncol. Biol. Phys.27:689-695: 1993. 24. Viggars. D. A.; Shalev, S.; Stewart.M.: Hahn.R. The objective evaluation of alternative treatment plans Ill: The quantitative analysisof dosevolume histograms.Int. J. Radiat. Oncol. Biol. Phys. 23:419-427; 1992. 25. Willett. C. G.; Linggood.R. M.; Stracher,M. A.; Goitein, M. A.; Doppke, K.: Kushner,D. C.; Morris, T.: Pardy, J.; Carrol, R. The effect of respiratorycycle on mediastinaland lung dimensionsin Hodgkin’s disease.Cancer 60:12321237; 1987.