Four-dimensional radiotherapy planning for DMLC-based respiratory motion trackingPaul J. Keall, Sarang Joshi, S. Sastry Vedam, Jeffrey V. Siebers, Vijaykumar R. Kini, and Radhe Mohan Citation: Medical Physics 32, 942 (2005); doi: 10.1118/1.1879152 View online: http://dx.doi.org/10.1118/1.1879152 View Table of Contents: http://scitation.aip.org/content/aapm/journal/medphys/32/4?ver=pdfcov Published by the American Association of Physicists in Medicine Articles you may be interested in Lung sparing and dose escalation in a robust-inspired IMRT planning method for lung radiotherapy that accountsfor intrafraction motion Med. Phys. 40, 061705 (2013); 10.1118/1.4805101 Systematic evaluation of four-dimensional hybrid depth scanning for carbon-ion lung therapy Med. Phys. 40, 031720 (2013); 10.1118/1.4792295 Four-dimensional dose evaluation using deformable image registration in radiotherapy for liver cancer Med. Phys. 40, 011706 (2013); 10.1118/1.4769427 The radiobiological P + index for pretreatment plan assessment with emphasis on four-dimensional radiotherapymodalities Med. Phys. 39, 6420 (2012); 10.1118/1.4754653 A simplified method of four-dimensional dose accumulation using the mean patient density representation Med. Phys. 35, 5269 (2008); 10.1118/1.3002304

Four-dimensional radiotherapy planning for DMLC-based respiratorymotion tracking

Paul J. Kealla!

Department of Radiation Oncology, Virginia Commonwealth University, Richmond, Virginia 23298

Sarang JoshiDepartment of Radiation Oncology, University of North Carolina, Chapel Hill, North Carolina 27514

S. Sastry Vedam, Jeffrey V. Siebers, and Vijaykumar R. KiniDepartment of Radiation Oncology, Virginia Commonwealth University, Richmond, Virginia 23298

Radhe MohanDepartment of Radiation Physics, The University of Texas M. D. Anderson Cancer Center,Houston, Texas 77030

sReceived 9 July 2004; revised 2 February 2005; accepted for publication 3 February 2005;published 16 March 2005d

Four-dimensionals4Dd radiotherapy is the explicit inclusion of the temporal changes in anatomyduring the imaging, planning, and delivery of radiotherapy. Temporal anatomic changes can occurfor many reasons, though the focus of the current investigation is respiration motion for lungtumors. The aim of this study was to develop 4D radiotherapy treatment-planning methodology forDMLC-based respiratory motion tracking. A 4D computed tomographysCTd scan consisting of aseries of eight 3D CT image sets acquired at different respiratory phases was used for treatmentplanning. Deformable image registration was performed to map each CT set from the peak-inhalerespiration phase to the CT image sets corresponding to subsequent respiration phases. Deformableregistration allows the contours defined on the peak-inhale CT to be automatically transferred to theother respiratory phase CT image sets. Treatment planning was simultaneously performed on eachof the eight 3D image sets via automated scripts in which the MLC-defined beam aperture conformsto the PTVswhich in this case equaled the GTV due to CT scan length limitationsd plus a penum-bral margin at each respiratory phase. The dose distribution from each respiratory phase CT imageset was mapped back to the peak-inhale CT image set for analysis. The treatment intent of 4Dplanning is that the radiation beam defined by the DMLC tracks the respiration-induced targetmotion based on a feedback loop including the respiration signal to a real-time MLC controller.Deformation with respiration was observed for the lung tumor and normal tissues. This deformationwas verified by examining the mapping of high contrast objects, such as the lungs and cord,between image sets. For the test case, dosimetric reductions for the cord, heart, and lungs werefound for 4D planning compared with 3D planning. 4D radiotherapy planning for DMLC-basedrespiratory motion tracking is feasible and may offer tumor dose escalation and/or a reduction intreatment-related complications. However, 4D planning requires new planning tools, such as de-formable registration and automated treatment planning on multiple CT image sets. ©2005American Association of Physicists in Medicine. fDOI: 10.1118/1.1879152g

Key words: 4D radiotherapy, deformable image registration, treatment planning

I. INTRODUCTION

The advent of four-dimensionals4Dd thoracic CT scans thatcreate separate CT images at discrete phases of the respira-tory cycle1–9 introduces the issue of how best to use thisinformationstypically over 1000 slices for a full 4D CT scanof the thoraxd for radiation therapy planning.

The use of multiple sequential CT data sets to improveradiotherapy’s ability to account for interfraction motion,termed “adaptive” radiotherapy, has been discussed by Yanetal.,10–14University of Wisconsin reports,15–19and others.20,21

In this approach, positioning and dosimetric informationfrom CT scans acquired during treatment is used to vary

patient positioning and/or the treatment beams for subse-quent treatments. Such strategies allow improved target cov-erage and/or normal tissue dose reduction.

4D radiotherapy planning to account for respiratory mo-tion applies concepts of adaptive radiotherapy to a 4D CTdata set, given that the radiation treatment beam will trackthe respiration-induced anatomical changes during treatment.In 4D radiotherapy planning, different treatment plans aregenerated for different respiratory phases with the constraintsof the treatment device’s ability to deliver these plans. Thisdefinition is consistent with that of the consensus of the AS-TRO 2003 “Time—the 4th dimension in radiotherapy” panelthat 4D radiotherapy is the “explicit inclusion of the tempo-ral changes of anatomy during the imaging, planning and

942 942Med. Phys. 32 „4…, April 2005 0094-2405/2005/32 „4…/942/10/$22.50 © 2005 Am. Assoc. Phys. Med.

delivery of radiotherapy.”22 Note that the definition of 4Dradiotherapy planning defined here differs from that dis-cussed by Shiratoet al.;23 of the breath-hold and free-breathing CT scans acquired in their study, only one wasused for treatment planning. Though the tumor tracking of animplanted fluoroscopic marker via x-ray is real-time, thetreatment delivery itself is gated.24–31

The aim of this research was to develop 4D radiotherapyplanning methodology to account for respiratory motion us-ing DMLC-based tumor tracking and provide a proof-of-principle example of 4D radiotherapy planning.

II. METHODS AND MATERIALS

A. 4D CT image

The 4D CT patient image acquired and used for this studyis described in detail in the publication by Vedamet al.2 Thepatient was a 50-year-old male with stage IIB non-small celllung cancer located in the right hilum. The 4D CT consistedof eight discrete respiratory phases: peak inhale, early ex-hale, mid exhale, late exhale, peak exhale, early inhale, midinhale, and late inhale. Due to both tube heat and reconstruc-tion time limitations, only a 10.5 cm region of the patient’sanatomy was scanned, covering the gross tumor volumesGTVd. On the peak-inhale CT image, the following anatomywas defined: GTV, lungs, heart, and the spinal cord. ThePINNACLE treatment-planning systemsPhilips Medical Sys-tems, Milpitas, CAd was used. The choice of using the peak-inhale CT image for the reference is somewhat arbitrary,however it is advisable to use either peak inhale or peakexhale as these phases have the fewest motion artifacts on4D CT. Because the scan only covered a 10.5 cm range, toensure that the planning target volumesPTVd did not extendoutside the CT volume, the PTV was assumed to equal theclinical target volumesCTVd, which was assumed to beequal to the GTV. Though this assumption is unrealistic forclinical radiotherapy, to achieve the aims of this study, theassumption was deemed acceptable.

B. Deformable image registration

Large deformation diffeomorphic image registration32–37

was used to map with a one-to-one correspondence all pointsin the CT image from one respiratory phasespeak inhaledwith corresponding points on the other respiratory phases,accommodating for the anatomic deformation caused by res-piration. LetV be the coordinate space of the peak-inhale CTimage. A time index transformation,hsx,td :Vt→V, map-ping the coordinate space of each of the respiratory phases ofthe 4D CT is estimated. The transformation is constrained tobe one to one while accommodating large deformations byexpressing it as a solution to a viscous-fluid, ordinary differ-ential equation:

dhsx,tddt

= vshsx,td,td,

wherevsh,td is the velocity field. It is assumed that the CTnumber of the tissue undergoing deformation remains con-

stant within the respiratory cycle and tissue deformation canbe tracked by directly correlating the CT images at differentphases of respiration using a minimum mean squared errormetric: Fshd=eVuIAshsx,tdd− IBsxdu2 dx, the difference be-tween the intensities in the transformed imageIA and imageIB. Should the approximation of constant tissue density provenot to be valid, adding a termfsIAd to IAshsx,tdd in Fshdwould improve the algorithm.fsIAd should be inverse Jaco-bian of the transformationh as the Jacobian measures thelocal volume change, and density is inversely proportional tovolume. The inclusion of this term significantly increases thecomplexity of the minimization problem, and is the subjectof ongoing research. Note also that as the anatomy includedin the CT data sets obtained differ near the boundaries, someinterpolation of the deformation fields near the edges of thedata sets is performed. The accuracy of this process has yetto be sufficiently quantified.

The transformationh is estimated in a manner that mini-mizes the metricDshd while at the same time constrainingthe transformation to satisfy the laws of continuum mechan-ics, derived from viscous-fluid modeling using the Navier–Stokes equations on the velocity fields. Having computed thetransformationh that maps the peak-inhale CT onto each ofthe respiratory phases of the 4D CT, the segmentation foreach of the respiratory phases is accomplished by applyingthe transformation to the contourssi.e., physician-drawn con-toursd of the peak-inhale CT. The transformationh thusquantifies the organ motion and the fine-featured organ shapechanges that occur during the breathing cycle. Verification ofdeformation registration on contours was performed by vi-sual inspection of those structures with sharp anatomicalcontrast boundaries, i.e., the external skin contour, the lungs,and the spinal cord. These contours represent structures thatmove little with respirationse.g., the cord in supine positiondand structures that move significantly with respirationse.g.,the lungsd. Quantification of the difference between manuallydefined anatomical points and those automatically deter-mined using the deformable image registration was per-formed. Four anatomic points were chosen: a bifurcation in aleft lower lobe vessel, the lateral aspect of a right lower lobebronchiole just proximal to a bifurcation, the anterior-mostaspect of the vertebral body at the level of the carina, and theright nipple. The first two points represent structures in thelung likely to move with respiration, the third point is acontrol, and the fourth point represents chest wall motion.These points were manually contoured on each respiratoryphase CT set. Taking the points defined on the end-inspiration CT image set, the points were automatically de-termined on subsequent respiratory phase CT image sets us-ing the transformations from the deformable imageregistrations.

In a separate study using this algorithm,38 the accuracy ofthe deformation process was quantified by comparing it tomanual segmentation for inhale and exhale CT scans of sixlung cancer patients. The results yielded a median error of0.7 mm and a 95% confidence error of 3.1 mm, with theerrors largest in the superior-inferior direction. This is ex-

943 Keall et al. : 4D radiotherapy planning 943

Medical Physics, Vol. 32, No. 4, April 2005

pected for two reasons: the CT resolution is significantlypoorer in this directionsthe 95% confidence interval error isof the order of the CT scan resolutiond, and this is also thepredominant direction of tumor motion.

Scripts were written to allow information associated withone CT image set, such as anatomic contours and dose dis-tributions, to be transferred to other respiratory phases of the4D CT via the deformation field. These scripts consist ofprograms that extract data from the treatment-planning sys-tem, such as contours or dose distributions associated withone CT image set, operate on this data via the transforma-tions described in the following, and input the transformedcontours/dose distributions back into the treatment-planningsystem. The interactions with the treatment-planning systemused the PinnComm scripting language, which is availablewith PINNACLE. Thus, the contours defined on the peak-inhale CT image were transferred to the other respiratoryphase of the 4D CT by passing through the deformable im-age registration operator, as shown in Fig. 1. Due to limita-tions of the contour storage structure of the treatment-planning system for 4D planning applications, thex and ysin-sliced points were deformed correctly, however, an aver-agez deformationsthe average of the individualz deforma-tion of constituent in-slice contour pointsd was used. Thislimitation will be eliminated in future versions ofPINNACLE.39

After treatment planning was performed for each respira-tory phasessee section below for specific planning detailsd,the dose distribution for each CT phase was transformedback to the peak-inhale CT image set for evaluation of theentire treatment plan accounting for all respiratory phases,where the combined or 4D dose distribution,D4D, is givenby

D4D = oiPhA,. . .,Hj

wiDshiAsxddi , s1d

wherewi is the fractional weight of the dose distributionDi

for each of thehA, . . . ,Hj CT phases. For this study, equalweighting was used for each phase, as approximately equalnumbers of images were assigned for each phase.2 However,in general thewi values will not be equal. Further informa-tion on the use of deformable image registration for dose

accumulation can be found in Refs. 33 and 40–52.

C. Treatment planning

A six-field coplanar nonopposed conformal plan was de-veloped on the peak-inhale phase, with the beam angles pre-dominantly in the anterior directionsangles 20°, 140°, 180°,220°, 300°, and 340°d. 6 MV energy was used for all beams.No wedges were used for the fields. The multileaf collimatorwas aligned in the superior–inferior direction, which hasbeen noted to be the predominant direction of motion forlung tumors.53–57 A 0.8 cm margin was added outside thePTV to account for penumbral effects, and the multileaf col-limator was used to define the field boundary. The beamweights were manually selected to give a reasonably uniformPTV dose with acceptable critical structure doses.

Automated scripts were written using the PinnCommscripting language and external programs to transfer thetreatment plan for the peak-inhale CT for each of the otherrespiratory phases. The autoblocking utility of the treatment-planning system allowed conformality of the field shape tothe PTV in each respiratory phase, as shown in Fig. 2 for twophases of the posterior beam. To create the final combined4D plan, the dose for each constituent respiratory phase wassummed via deformable image registration. For this ap-proach the beams weights used for the manually plannedphase were the same for each automatically planned respira-tory phase. It would be possiblesthough not investigatedhered to optimize the beam weights for the 4D plan whichmay further improve the plan quality.

To create the 3D plan, the PTV for each of the respiratoryphases was superimposed, i.e., the 3D PTV is the union ofeach 4D PTV phase. The treatment plan for the peak-inhalephase was the same as that used for the 4D plan, except thatthe fields conformed to the 3D PTV as shown in Fig. 2. The3D vs 4D comparison performed here is a “best-case” com-parison for 3D in that no extra margins were added to ac-count for the artifacts during imaging,1,2,27,58–64and addi-tional internal margins were required to ensure that the CTVwas adequately dosed during delivery in which respiratorymotion occurred. It should also be noted that a clinical 4Dtreatment plan should have additional geometric margins toaccount for the errors in the deformable image registrationprocess and also the ability of the DMLC to track the movingtarget.65–67 The 3D and 4D treatment plans are compared inSec. III.

FIG. 1. A schematic showing how contours defined on one phase, in thiscase inhale, are automatically transferred to the exhale image through thedeformation operator.

FIG. 2. The beam view of the posterior beam, with the aperture set to a0.8 cm margin surroundingsad the 3D planning target volumesPTVd, sbd thepeak-inhale 4D PTV, andscd the peak-exhale 4D PTV. Note the obviousdeformation of the PTV at the different respiration phases.

944 Keall et al. : 4D radiotherapy planning 944

Medical Physics, Vol. 32, No. 4, April 2005

III. RESULTS

A. Deformable image registration

The comparison of the deformable image registration al-gorithm results compared with points manually defined oneach respiratory phase are shown in Fig. 3. The differencesbetween the manual and automatically defined points werewithin one CT slice thicknesss0.4 cmd for all points studied.It is interesting to note that the displacements predicted bydeformable registration showed smaller variations fromphase to phase than the manually defined points.

The 4D PTVs in axial, coronal, and sagittal views at fourphases of the respiratory cycle are shown in Fig. 4. Thevolumes encompassed by the skin, heart, PTV, and lungs ateach respiratory phase are given in Fig. 5. Note that thenormal tissue values shown in Fig. 5 are limited to the vol-ume within the CT scans10.5 cm in lengthd, which did notencompass the entire thorax. The PTV volumes are fairlyconstant apart from the early inspiration phase value that isapproximately 7% smaller than the other volumes. It is un-clear if this low value is an artifact of the 4D CT itself, thedeformable image registration algorithm, or simply due tovariations introduced by the 4 mm CT slice thickness. Thelung volumes change as expected with respiration. The vol-ume encompassed by the skins stays nearly constantslessthan 35 cc change with respirationd for the CT scan length,though we would expect for a full thoracic CT scan that thevolume would change by the volume of air inhaledsaccount-ing for pressure differencesd. The heart changes by less than5% from the mean volume, and is also influenced by cardiacmotion. The cord changes by less than 0.5 cc from the meanvolume.

The centroid of the 4D PTV for each respiratory phase isshown in Fig. 6, giving evidence of cyclic motion as ob-served in the fluoroscopic implanted marker study of Sep-penwooldeet al.53 Note that the minimum and maximum

superior–inferior positions of the PTV centroid do not coin-cide with the peak inhale and peak exhale positions as deter-mined from the respiratory signal. Thus there is evidence ofa phase shift between the internal motion and the respiratorysignal. This phase shift itself does not degrade the 4D plan,however, variations in the phase shift, or more generallyvariations in the correlation between the internal motion andrespiratory signal, would negatively impact the accuracy ofthe 4D delivery.

B. 4D treatment plan

The dose distribution for each CT phase and the dosedistribution transformed back to the peak-inhale CT imageset for evaluation of the entire treatment plan compiled forall respiratory phasessthe 4D pland is shown in Fig. 7. Thedose distributions in each phase are fairly similar due to therelatively s,10%d small change in overall volume of thePTV for each respiratory phase, and the adaptation of thetreatment plan to these positional and volumetric PTVchanges.

Dose–volume histogramssDVHsd for each constituentbreathing phase and for the combined dose distributionsfound by summing the constituent dose distributions via de-formable operatorsd are given in Fig. 8 for the PTV, lungs,cord, and heart. The DVHs for the PTV at all phases and thecombineds4Dd DVH are all very similarsnote the expandedx-axis scaled. The lung DVHs are closely bunched, with thecombined 4D DVH closer to the end-inhale DVH due to thefact that DVHs typically use normalized rather than absolutevolumes, and that the entire lung was not included in the CTscan. The 4D DVHs for the cord and heart appear to be nearthe middle of the constituent-phase DVHs. The variation incord and heart DVHs for the constituent phases is due to thechange in beam aperture with respiratory phase and, hence,the fraction of the organ intersecting the beam passingthrough these structures, as the PTV deforms with respira-tory phase. Note that the 4D cord DVH seems to exceed all3D DVHs around 10 Gy. This is possible due to the dosesummation to obtain the 4D dose, in that the 4D DVH willshow the maximum of the minimum dose in any phase. Tak-

FIG. 3. Manualsclosed symbolsd and automaticallysopen symbolsd deter-mined displacements of four anatomical points from the same point definedin the end inspiration respiratory phase CT image set. Lt lung vessel=bifurcation in a left lower lobe vessel. Rt bronchiole=the lateral aspect ofa right lower lobe bronchiole just proximal to a bifurcation. Ant vertebra=the anterior-most aspect of the vertebral body at the level of the carina. Rtnipple=the right nipple. The error bars of 0.4 cmscorresponding to one CTslice thicknessd is shown on the left lung vessel curve. Each anatomical siteis offset 0.5 cm for clarity.

FIG. 4. The PTV contoured on the computed tomographysCTd anatomy inaxial view sad–sdd, coronal viewsed–shd, and sagittal viewsid–sld. The fourrespiration phases shown are peak inhalesad, sed, andsid; mid exhalesbd, sfd,and sjd; peak exhalescd, sgd, andskd; and mid inhalesdd, shd, andsld. Notethat the anatomy was explicitly contoured on the peak-inhale images, andthe PTVs on the other breathing phases were automatically created based onthe deformable registration transformations.

945 Keall et al. : 4D radiotherapy planning 945

Medical Physics, Vol. 32, No. 4, April 2005

ing a simple example of a two voxel structure with doses of0 and 1, then the minimum dose in any voxel will be 0, butthe minimum dose in the summed/averaged voxel can begreater than 0. For example, phase A has doses 0 and 1,phase B has doses of 1 and 0, then the 4D dose would be 0.5and 0.5 in the two voxels, respectively.

C. 3D vs 4D

A comparison of the 3D PTV and peak-inhale PTV isgiven in Fig. 9. As expected, the 3D PTV is larger than thepeak-inhale-phase PTV. DVHs for the 3D and 4D treatmentplans are seen Fig. 10. The isodose distributions appear to berelatively similar for the central axis slice shown. However,the DVHs show dosimetric reductions for the lungs, heart,and cord for the 4D plan. Though the comparisons here showsomewhat modest gains for 4D vs 3D planning, as expanded

upon in Sec. IV, the current comparisons do not include ar-tifacts during conventional CT imaging, which are a knownsource of error.

IV. DISCUSSION

A 4D treatment-planning process for DMLC-based respi-ratory motion tracking has been developed and applied to anexisting 4D CT image. Dosimetric advantages were foundfor the 4D treatment plan compared with the 3D plan fornontarget structures. Though no general conclusions can bedrawn from the single proof-of-principle example case dem-onstrated here, clearly, if the respiratory motion is explicitlyaccounted for during the imaging, planning, and delivery ofradiotherapy, some dosimetric improvements are expected.Applying the 4D planning method to a larger patient cohortis required to quantify the potential gains for 4D radio-

FIG. 5. The volumes of the entire computed tomographysCTd anatomy in each of the eight respiratory phases:sad PTV, sbd lungs fminus the gross tumorvolume sGTVdg, scd heart,sdd cord, andsed skin. Note that the entire volumes of organssbd–sed were not included in the 4D CT scan due to acquisitionlimitations.

946 Keall et al. : 4D radiotherapy planning 946

Medical Physics, Vol. 32, No. 4, April 2005

therapy, both compared with conventional 3D treatments andalso respiratory gated and breath-hold treatments.

Without automated tools, an order-of-magnitude increasein workload for 4D treatment planning over 3D planning isrequired due to the order-of-magnitude increase in input dataand tasks such as contouring multiple data sets, creatingtreatment plans on these data sets, and analyzing the dosedistribution for each of these images. Such an increase isneither desirable nor feasible. The tools that can reduce this

workload to essentially that of a 3D treatment plan are de-formable image registration and automated planning. De-formable image registration tools for radiotherapy purposesare relatively new and untested. As with any process, geo-metric uncertainties in the registration method will need tobe quantified and taken into account during the planning pro-cess. Development and testing of these algorithms on mul-tiple 4D CT sets is required for these tools, which will be-come an integral part of treatment-planning software as thefield moves toward image-guided adaptive and 4D radio-therapy. Indeed, combining dose distributions without toolsthat map corresponding anatomical points in different datasets is problematic.

FIG. 6. The position of the centroids of the 4D planning target volumesPTVd for each of the eight respiratory phases.

FIG. 7. The process of combining the constituent dose distributions fromeach breathing phase onto the composite dose distribution to create thecombined or 4D plan. The 20, 60, 66, and 70 Gy isodose lines are shown.Note the similarity in dose distribution for each respiratory phase.

FIG. 8. Dose–volume histograms for each breathing phasesthin solid linesd and the combined distributionsthick dashed lined for the sad planning targetvolume sPTVd, sbd lungs,scd cord, andsdd heart.

947 Keall et al. : 4D radiotherapy planning 947

Medical Physics, Vol. 32, No. 4, April 2005

Planning for 4D radiotherapy must account for the geo-metric uncertainties not accounted for by the respiratorytracking. Such uncertainties include setup error and interfrac-tion motion as well as the 4D-specific uncertainties such asartifacts in 4D CT scans,9,68 deformable image registrationaccuracy,38 variations in respiratory signal and internalanatomy motion,69–72 motion prediction errors,73–75 and aquantification of the DMLC positional and temporal accu-racy as it is being operated in a mode that it was not initiallydesigned for. Obviously, for 4D radiotherapy to be a viableoption the sum of these geometric uncertainties should besignificantly less than the sum of the geometric uncertaintiesfor 3D radiotherapy.

There are both similarities and differences between 4Dradiotherapy that accounts for respiratory motion and adap-tive radiotherapy that accounts for interfractiondisplacements.10–21 Both 4D and adaptive radiotherapy planon multiple CT images, require dose calculation on multipleCT images, and require deformable registration and methodsto obtain integral dose distributions and DVHs. However, 4Dradiotherapy is delivering radiation to a moving target and,

thus, is affected by beam-delivery system constraints, such asMLC velocity and acceleration limitations, that do not affectadaptive radiotherapy. For IMRT optimization, 4D radio-therapy allows optimization on several data sets simulta-neously and within the leaf motion constraints, whereasadaptive radiotherapy is restricted to optimization on currentdata sets with knowledge of previous data sets. With thesimilarities and commons tools required for 4D and adaptiveradiotherapy, these techniques are complementary rather thancontradictory. The 4D planning method described in this ar-ticle is sensitive to variations in the respiratory surrogateused and the target anatomy motion. Adaptive radiotherapystrategies, which for 4D radiotherapy would involve reac-quiring the 4D image of the patient and the respiration sig-nal, and hence reacquiring the correlation, would reduce thissensitivity sand also provide valuable information as to themagnitude of this effectd.

Using 4D radiotherapy to explicitly account for thechange in target shape with respiration by varying the MLCpositions logically leads to the concept of the 4D PTV, wherethe PTV for a given respiratory phase is fixed in space, perthe ICRU definition;76,77 however, the PTV changes withtime. The 4D PTV concept is analogous in adaptive radio-therapy, where the PTV changes with time due to interfrac-tional changes in the GTV position and shape.

Apart from applying the 4D radiotherapy planningmethod to a significant number of patient cases, the nextlogical step is to develop a 4D IMRT planningmethodology,65–67 though due to beam-delivery system con-straints, such as leaf acceleration and velocity limitations, theleaf sequencing for a 4D IMRT plan is nontrivial. The deliv-ery aspects of 4D treatment plans66,67,73,74,78–81including res-piration prediction to account for the system response timehave not been addressed here and are the subject of ongoinginvestigations.

Reproducibility in respiratory patterns during the imagingand delivery of 4D radiotherapy is an important issue requir-ing further investigation. Irreproducible respiration patternsmay preclude some patients as candidates for 4D radio-therapy, whilst breathing training techniques82 may be re-quired for the majority of lung cancer patients in order tomaintain breathing regularity over a protracted course oftherapy. Since extracranial stereotactic radiotherapysESRdtechniques83–90have a much shorter duration of therapy, thiscohort of patients may be most suitable for initial clinicaltesting of 4D radiotherapy. This reduction in overall treat-ment time means that the respiration reproducibility require-ments are only for a week or so as opposed to six or moreweeks of standard therapy. Another advantage of applying4D radiotherapy to ESR patients is that the lesions are typi-cally smaller than those treated with conventional therapy;thus, the ratio of respiratory-induced motion to the tumorvolume is higher and the fractional improvement with 4Dradiotherapy over standard delivery will be higher. Also,with the condensed course of therapy for ESR patients, typi-cally more resources per fraction are utilized, and, thus, po-tentially resource-intensive strategies, such as 4D radio-therapy, can be employed.

FIG. 9. The 4D PTV for the peak-inhale phasesdark shadingd, and the 3DPTV created by combining the PTVs from all respiratory phasesslight shad-ingd. The axialsad, coronalsbd, and sagittalscd views are shown. Note thetotal volumes for the peak-inhale PTV is 197 cm3, and the 3D PTV237 cm3.

948 Keall et al. : 4D radiotherapy planning 948

Medical Physics, Vol. 32, No. 4, April 2005

Though not explicitly part of this manuscript, the deliveryof 4D radiotherapy using DMLC-based respiratory motiontracking should be mentioned here. Based on respiratoryphysiology literature,91–97and our institutional study collect-ing 330 respiratory motion traces from lung cancerpatients,98 cycle-to-cycle and day-to-day variations in respi-ration occur. Thus a feedback loopssee for example Fig. 8 inRef. 99d acquiring the patient’s respiratory signal and feedingthis information to a four-dimensional controller which com-municates in real-time with the MLC will be required. Vary-ing breathing patterns during delivery, compared with thosemeasured during the 4D CT imaging session used for treat-ment planning, mean that the planned and delivered dosesmay differ. For example, if the time spent within each breath-ing phase during treatment delivery,wd, differs from thatduring planning where fractionwi was used for the final dosecalculation, the combined 4D dose will change following Eq.s1d. However, using Eq.s1d with wd means that the actualdose delivered to the moving tumor and critical structures foreach treatment fraction can be calculated. A problem appearsif the respiration pattern limits during delivery exceed thoseobtained during the 4D CT session used for planning. Shouldsuch a situation occur, either the treatment should be pauseduntil the respiration pattern returns within the limits knownfrom planningsthe prudent approachd or until a reasonableextrapolation of the tumor position and shape based on therespiratory signal can be made.

V. CONCLUSION

4D radiotherapy planning for DMLC-based respiratorymotion tracking has been shown to be feasible and may offertumor dose escalation and/or a reduction in treatment-relatedcomplications. In principle, 4D planning requires little morehuman interaction time than a 3D plan, however, new plan-ning tools such as deformable registration and automatedtreatment planning on multiple CT image sets are required.

ACKNOWLEDGMENTS

The authors wish to acknowledge the support of NIH/NCIR01 CA93626. Devon Murphy carefully reviewed and sig-nificantly improved the clarity of this manuscript. Dr. An-drew Lauve assisted with contour delineation. Rohini Georgeassisted with collection of background material. Dr. SteveWebb and Catherine Coolens from the Royal Marsden Hos-pital provided useful feedback.

adElectronic mail: pjkeall@vcu.edu1E. C. Ford, G. S. Mageras, E. Yorke, and C. C. Ling, “Respiration-correlated spiral CT: A method of measuring respiratory-induced ana-tomic motion for radiation treatment planning,” Med. Phys.30, 88–97s2003d.

2S. S. Vedam, P. J. Keall, V. R. Kini, H. Mostafavi, H. P. Shukla, and R.Mohan, “Acquiring a four-dimensional computed tomography dataset us-ing an external respiratory signal,” Phys. Med. Biol.48, 45–62s2003d.

3D. A. Low, M. Nystrom, E. Kalinin, P. Parikh, J. F. Dempsey, J. D.Bradley, S. Mutic, S. H. Wahab, T. Islam, G. Christensen, D. G. Politte,and B. R. Whiting, “A method for the reconstruction of four-dimensionalsynchronized CT scans acquired during free breathing,” Med. Phys.30,1254–1263s2003d.

4E. Rietzel, K. Doppke, T. Pan, N. Choi, C. Willett, and G. Chen, “4Dcomputer tomography for radiation therapysabstractd,” Med. Phys. 30,1365–1366s2003d.

5K. Taguchi, “Temporal resolution and the evaluation of candidate algo-rithms for four-dimensional CT,” Med. Phys.30, 640–650s2003d.

6J. Sonke, P. Remeijer, and M. van Herk, “Respiration-correlated conebeam CT: Obtaining a four-dimensional data setsabstractd,” Med. Phys.30, 1415s2003d.

7E. Rietzel, G. T. Chen, K. P. Doppke, T. Pan, N. C. Choi, and C. G.Willett, “4D computed tomography for treatment planning,” Int. J. Radiat.Oncol., Biol., Phys.57, S232–S233s2003d.

8G. S. Mageras, “Respiration correlated CT techniques for gated treatmentof lung cancer,” Radiother. Oncol.64, 75 s2002d.

9P. J. Keall, G. Starkschall, H. Shukla, K. M. Forster, V. Ortiz, C. W.Stevens, S. S. Vedam, R. George, T. Guerrero, and R. Mohan, “Acquiring4D thoracic CT scans using a multislice helical method,” Phys. Med.Biol. 49, 2053–2067s2004d.

10D. Yan, D. Lockman, D. Brabbins, L. Tyburski, and A. Martinez, “Anoff-line strategy for constructing a patient-specific planning target volumein adaptive treatment process for prostate cancer,” Int. J. Radiat. Oncol.,Biol., Phys. 48, 289–302s2000d.

11D. Yan, F. Vicini, J. Wong, and A. Martinez, “Adaptive radiationtherapy,” Phys. Med. Biol.42, 123–132s1997d.

12D. Yan, J. Wong, F. Vicini, J. Michalski, C. Pan, A. Frazier, E. Horwitz,and A. Martinez, “Adaptive modification of treatment planning to mini-mize the deleterious effects of treatment setup errors,” Int. J. Radiat.Oncol., Biol., Phys.38, 197–206s1997d.

13D. Yan, E. Ziaja, D. Jaffray, J. Wong, D. Brabbins, F. Vicini, and A.Martinez, “The use of adaptive radiation therapy to reduce setup error: Aprospective clinical study,” Int. J. Radiat. Oncol., Biol., Phys.41, 715–720 s1998d.

14A. A. Martinez, D. Yan, D. Lockman, D. Brabbins, K. Kota, M. Sharpe,D. A. Jaffray, F. Vicini, and J. Wong, “Improvement in dose escalationusing the process of adaptive radiotherapy combined with three-dimensional conformal or intensity-modulated beams for prostate can-cer,” Int. J. Radiat. Oncol., Biol., Phys.50, 1226–1234s2001d.

15J. M. Kapatoes, G. H. Olivera, J. P. Balog, H. Keller, P. J. Reckwerdt, andT. R. Mackie, “On the accuracy and effectiveness of dose reconstructionfor tomotherapy,” Phys. Med. Biol.46, 943–966s2001d.

16H. Keller, M. A. Ritter, and T. R. Mackie, “Optimal stochastic correctionstrategies for rigid-body target motion,” Int. J. Radiat. Oncol., Biol., Phys.55, 261–270s2003d.

17T. R. Mackie, J. Balog, K. Ruchala, D. Shepard, S. Aldridge, E. Fitchard,P. Reckwerdt, G. Olivera, T. McNutt, and M. Mehta, “Tomotherapy,”Semin Radiat. Oncol.9, 108–117s1999d.

18J. S. Welsh, R. R. Patel, M. A. Ritter, P. M. Harari, T. R. Mackie, and M.P. Mehta, “Helical tomotherapy: An innovative technology and approachto radiation therapy,” Technol. Cancer Res. Treat.1, 311–316s2002d.

FIG. 10. Dose–volume histogram comparison of the 3Dssolid linesd treat-ment plan and the 4Dsdashed linesd plan for the PTV, lungs, cord, and heart.

949 Keall et al. : 4D radiotherapy planning 949

Medical Physics, Vol. 32, No. 4, April 2005

19C. Wu, R. Jeraj, G. H. Olivera, and T. R. Mackie, “Re-optimization inadaptive radiotherapy,” Phys. Med. Biol.47, 3181–3195s2002d.

20M. Birkner, D. Yan, M. Alber, J. Liang, and F. Nusslin, “Adapting inverseplanning to patient and organ geometrical variation: Algorithm and imple-mentation,” Med. Phys.30, 2822–2831s2003d.

21J. Lof, B. K. Lind, and A. Brahme, “An adaptive control algorithm foroptimization of intensity modulated radiotherapy considering uncertain-ties in beam profiles, patient set-up and internal organ motion,” Phys.Med. Biol. 43, 1605–1628s1998d.

22P. J. Keall, G. T. Y. Chen, S. Joshi, T. R. Mackie, and C. W. Stevens,“Time-The fourth dimension in radiotherapysASTRO Panel Discus-siond,” Int. J. Radiat. Oncol., Biol., Phys.57, S8–9s2003d.

23H. Shirato, S. Shimizu, K. Kitamura, T. Nishioka, K. Kagei, S. Hash-imoto, H. Aoyama, T. Kunieda, N. Shinohara, H. Dosaka-Akita, and K.Miyasaka, “Four-dimensional treatment planning and fluoroscopic real-time tumor tracking radiotherapy for moving tumor,” Int. J. Radiat. On-col., Biol., Phys.48, 435–432s2000d.

24T. Harada, H. Shirato, S. Ogura, S. Oizumi, K. Yamazaki, S. Shimizu, R.Onimaru, K. Miyasaka, M. Nishimura, and H. Dosaka-Akita, “Real-timetumor-tracking radiation therapy for lung carcinoma by the aid of inser-tion of a gold marker using bronchofiberscopy,” Cancer95, 1720–1727s2002d.

25K. Kitamura, H. Shirato, S. Shimizu, N. Shinohara, T. Harabayashi, T.Shimizu, Y. Kodama, H. Endo, R. Onimaru, S. Nishioka, H. Aoyama, K.Tsuchiya, and K. Miyasaka, “Registration accuracy and possible migra-tion of internal fiducial gold marker implanted in prostate and livertreated with real-time tumor-tracking radiation therapysRTRTd,” Radio-ther. Oncol. 62, 275–281s2002d.

26K. Kitamura, H. Shirato, N. Shinohara, T. Harabayashi, R. Onimaru, K.Fujita, S. Shimizu, K. Nonomura, T. Koyanagi, and K. Miyasaka, “Re-duction in acute morbidity using hypofractionated intensity-modulatedradiation therapy assisted with a fluoroscopic real-time tumor-trackingsystem for prostate cancer: Preliminary results of a phase I/II study,”Cancer J.9, 268–276s2003d.

27S. Shimizu, H. Shirato, S. Ogura, H. Akita-Dosaka, K. Kitamura, T. Nish-ioka, K. Kagei, M. Nishimura, and K. Miyasaka, “Detection of lung tu-mor movement in real-time tumor-tracking radiotherapy,” Int. J. Radiat.Oncol., Biol., Phys.51, 304–310s2001d.

28S. Shimizu, H. Shirato, B. Xo, K. Kagei, T. Nishioka, S. Hashimoto, K.Tsuchiya, H. Aoyama, and K. Miyasaka, “Three-dimensional movementof a liver tumor detected by high-speed magnetic resonance imaging,”Radiother. Oncol.50, 367–370s1999d.

29H. Shirato, S. Shimizu, T. Shimizu, T. Nishioka, and K. Miyasaka, “Real-time tumor-tracking radiotherapy,” Lancet353, 1331–1332s1999d.

30H. Shirato, S. Shimizu, T. Kunieda, K. Kitamura, M. van Herk, K. Kagei,T. Nishioka, S. Hashimoto, K. Fujita, H. Aoyama, K. Tsuchiya, K. Kudo,and K. Miyasaka, “Physical aspects of a real-time tumor-tracking systemfor gated radiotherapy,” Int. J. Radiat. Oncol., Biol., Phys.48, 1187–1195s2000d.

31H. Shirato, T. Harada, T. Harabayashi, K. Hida, H. Endo, K. Kitamura, R.Onimaru, K. Yamazaki, N. Kurauchi, T. Shimizu, N. Shinohara, M. Mat-sushita, H. Dosaka-Akita, and K. Miyasaka, “Feasibility of insertion/implantation of 2.0-mm-diameter gold internal fiducial markers for pre-cise setup and real-time tumor tracking in radiotherapy,” Int. J. Radiat.Oncol., Biol., Phys.56, 240–247s2003d.

32S. Joshi, S. Pizer, P. T. Fletcher, P. Yushkevich, A. Thall, and J. S. Mar-ron, “Multiscale deformable model segmentation and statistical shapeanalysis using medial descriptions,” IEEE Trans. Med. Imaging21, 538–550 s2002d.

33G. E. Christensen, B. Carlson, K. S. Chao, P. Yin, P. W. Grigsby, K.Nguyen, J. F. Dempsey, F. A. Lerma, K. T. Bae, M. W. Vannier, and J. F.Williamson, “Image-based dose planning of intracavitary brachytherapy:Registration of serial-imaging studies using deformable anatomic tem-plates,” Int. J. Radiat. Oncol., Biol., Phys.51, 227–243s2001d.

34G. E. Christensen, S. C. Joshi, and M. I. Miller, “Volumetric transforma-tion of brain anatomy,” IEEE Trans. Med. Imaging16, 864–877s1997d.

35M. Miller, A. Banerjee, G. Christensen, S. Joshi, N. Khaneja, U.Grenander, and L. Matejic, “Statistical methods in computationalanatomy,” Stat. Methods Med. Res.6, 267–299s1997d.

36G. E. Christensen, R. D. Rabbitt, and M. I. Miller, “Deformable templatesusing large deformation kinematics,” IEEE Trans. Image Process.5,1435–1477s1996d.

37S. Joshi and M. I. Miller, “Landmark matching via large deformationdiffeomorphisms,” IEEE Trans. Image Process.9, 1357–1370s1997d.

38G. S. Mageras, S. Joshi, B. Davis, A. Pevsner, A. Hertanto, E. Yorke, K.Rosenzweig, Y. Erdi, S. Nehmeh, J. Humm, S. M. Larson, and C. C. Ling,“Evaluation of an automated deformable matching method for quantify-ing lung tumor motion in respiration-correlated CT images,” inInterna-tional Conference on the Use of Computers in RadiotherapysJeong,Seoul, Korea, 2004d, pp. 364–367.

39V. Pekar, T. R. McNutt, and K. Michael, “Automated model-based organdelineation for radiation therapy planning in the prostate region,” Int. J.Radiat. Oncol., Biol., Phys.57, S206s2003d.

40K. K. Brock, S. J. Hollister, L. A. Dawson, and J. M. Balter, “Technicalnote: Creating a four-dimensional model of the liver using finite elementanalysis,” Med. Phys.29, 1403–1405s2002d.

41K. K. Brock, D. L. McShan, and J. M. Balter, “A comparison ofcomputer-controlled versus manual online patient setup adjustment,” J.Appl. Clin. Med. Phys.3, 241–247s2002d.

42K. K. Brock, D. L. McShan, R. K. Ten Haken, S. J. Hollister, L. A.Dawson, and J. M. Balter, “Inclusion of organ deformation in dose cal-culations,” Med. Phys.30, 290–295s2003d.

43K. M. Brock, J. M. Balter, L. A. Dawson, M. L. Kessler, and C. R.Meyer, “Automated generation of a four-dimensional model of the liverusing warping and mutual information,” Med. Phys.30, 1128–1133s2003d.

44T. Guerrero, G. Zhang, T. Huang, and K. Lin, “Intrathoracic tumour mo-tion estimation from CT imaging using the 3D optical flow method,”Phys. Med. Biol.49, 4147–4161s2004d.

45P. J. Keall, J. V. Siebers, S. Joshi, and R. Mohan, “Monte Carlo as afour-dimensional radiotherapy treatment-planning tool to account for res-piratory motion,” Phys. Med. Biol.49, 3639–3648s2004d.

46J. Liang and D. Yan, “Reducing uncertainties in volumetric image baseddeformable organ registration,” Med. Phys.30, 2116–2122s2003d.

47W. Lu, M. L. Chen, G. H. Olivera, K. J. Ruchala, and T. R. Mackie, “Fastfree-form deformable registration via calculus of variations,” Phys. Med.Biol. 49, 3067–3087s2004d.

48M. Rosu, L. A. Dawson, J. M. Balter, D. L. McShan, T. S. Lawrence, andR. K. Ten Haken, “Alterations in normal liver doses due to organ mo-tion,” Int. J. Radiat. Oncol., Biol., Phys.57, 1472–1479s2003d.

49C. Wu, R. Jeraj, W. Lu, and T. R. Mackie, “Fast treatment plan modifi-cation with an over-relaxed Cimmino algorithm,” Med. Phys.31, 191–200 s2004d.

50D. Yan, D. A. Jaffray, and J. W. Wong, “A model to accumulate fraction-ated dose in a deforming organ,” Int. J. Radiat. Oncol., Biol., Phys.44,665–675s1999d.

51D. Yan and D. Lockman, “Organ/patient geometric variation in externalbeam radiotherapy and its effects,” Med. Phys.28, 593–602s2001d.

52B. Schaly, J. A. Kempe, G. S. Bauman, J. J. Battista, and J. Van Dyk,“Tracking the dose distribution in radiation therapy by accounting forvariable anatomy,” Phys. Med. Biol.49, 791–805s2004d.

53Y. Seppenwoolde, H. Shirato, K. Kitamura, S. Shimizu, M. van Herk, J.V. Lebesque, and K. Miyasaka, “Precise and real-time measurement of3D tumor motion in lung due to breathing and heartbeat, measured duringradiotherapy,” Int. J. Radiat. Oncol., Biol., Phys.53, 822–834s2002d.

54K. E. Sixel, M. Ruschin, R. Tirona, and P. C. Cheung, “Digital fluoros-copy to quantify lung tumor motion: Potential for patient-specific plan-ning target volumes,” Int. J. Radiat. Oncol., Biol., Phys.57, 717–723s2003d.

55I. S. Grills, D. Yan, A. A. Martinez, F. A. Vicini, J. W. Wong, and L. L.Kestin, “Potential for reduced toxicity and dose escalation in the treat-ment of inoperable non-small-cell lung cancer: A comparison of intensity-modulated radiation therapysIMRTd, 3D conformal radiation, and elec-tive nodal irradiation,” Int. J. Radiat. Oncol., Biol., Phys.57, 875–890s2003d.

56C. S. Ross, D. H. Hussey, E. C. Pennington, W. Stanford, and J. F.Doornbos, “Analysis of movement of intrathoracic neoplasms using ul-trafast computerized tomography,” Int. J. Radiat. Oncol., Biol., Phys.18,671–677s1990d.

57H. W. Korin, R. L. Ehman, S. J. Riederer, J. P. Felmlee, and R. C. Grimm,“Respiratory kinematics of the upper abdominal organs: A quantitativestudy,” Magn. Reson. Med.23, 172–178s1992d.

58P. J. Keall, V. R. Kini, S. S. Vedam, and R. Mohan, “Potential radio-therapy improvements with respiratory gating,” Australas. Phys. Eng. Sci.Med. 25, 1–6 s2002d.

950 Keall et al. : 4D radiotherapy planning 950

Medical Physics, Vol. 32, No. 4, April 2005

59S. Shimizu, H. Shirato, K. Kagei, T. Nishioka, X. Bo, H. Dosaka-Akita, S.Hashimoto, H. Aoyama, K. Tsuchiya, and K. Miyasaka, “Impact of res-piratory movement on the computed tomographic images of small lungtumors in three-dimensionals3Dd radiotherapy,” Int. J. Radiat. Oncol.,Biol., Phys. 46, 1127–1133s2000d.

60R. D. Tarver, D. J. Conces, and J. D. Godwin, “Motion artifacts on CTsimulate bronchiectasis,” Am. J. Roentgenol.151, 1117–1119s1988d.

61L. A. Shepp, S. K. Hilal, and R. A. Schulz, “The tuning fork artifact incomputerized tomography,” Comput. Graph. Image Process.10, 246–255 s1979d.

62C. J. Ritchie, J. Hseih, M. F. Gard, J. D. Godwin, Y. Kim, and C. R.Crawford, “Predictive respiratory gating: A new method to reduce motionartifacts on CT scans,” Radiology190, 847–852s1994d.

63C. J. Ritchie, J. D. Godwin, C. R. Crawford, W. Stanford, H. Anno, and Y.Kim, “Minimum scan speeds for suppression of motion artifacts in CT,”Radiology 185, 37–42s1992d.

64J. R. Mayo, N. L. Müller, and R. M. Henkelman, “The double-fissuresign: A motion artifact on thin-section CT scans,” Radiology165, 580–581 s1987d.

65L. Papiez, “The Leaf Sweep algorithm for an immobile and moving targetas an optimal control problem in radiotherapy delivery,” Math. Comput.Modell. 37, 735–745s2003d.

66T. Neicu, H. Shirato, Y. Seppenwoolde, and S. B. Jiang, “Synchronizedmoving aperture radiation therapysSMARTd: Average tumour trajectoryfor lung patients,” Phys. Med. Biol.48, 587–598s2003d.

67P. J. Keall, V. Kini, S. S. Vedam, and R. Mohan, “Motion adaptive x-raytherapy: A feasibility study,” Phys. Med. Biol.46, 1–10s2001d.

68T. Pan, T. Y. Lee, E. Rietzel, and G. T. Chen, “4D-CT imaging of avolume influenced by respiratory motion on multi-slice CT,” Med. Phys.31, 333–340s2004d.

69J. M. Balter, L. A. Dawson, S. Kazanjian, C. McGinn, K. K. Brock, T.Lawrence, and R. Ten Haken, “Determination of ventilatory liver move-ment via radiographic evaluation of diaphragm position,” Int. J. Radiat.Oncol., Biol., Phys.51, 267–270s2001d.

70S. S. Vedam, V. R. Kini, P. J. Keall, V. Ramakrishnan, H. Mostafavi, andR. Mohan, “Quantifying the predictability of diaphragm motion duringrespiration with a noninvasive external marker,” Med. Phys.30, 505–513s2003d.

71J. D. Hoisak, K. E. Sixel, R. Tirona, P. C. Cheung, and J. P. Pignol,“Correlation of lung tumor motion with external surrogate indicators ofrespiration,” Int. J. Radiat. Oncol., Biol., Phys.60, 1298–1306s2004d.

72Y. Tsunashima, T. Sakae, Y. Shioyama, K. Kagei, T. Terunuma, A.Nohtomi, and Y. Akine, “Correlation between the respiratory waveformmeasured using a respiratory sensor and 3D tumor motion in gated radio-therapy,” Int. J. Radiat. Oncol., Biol., Phys.60, 951–958s2004d.

73M. J. Murphy, J. Jalden, and M. Isaksson, “Adaptive filtering to predictlung tumor breathing motion during image-guided radiation therapy,”presented at the 16th International Congress on Computer-assisted Radi-ology and SurgerysCARSd, 2002.

74G. C. Sharp, S. B. Jiang, S. Shimizu, and H. Shirato, “Prediction ofrespiratory tumour motion for real-time image-guided radiotherapy,”Phys. Med. Biol.49, 425–440s2004d.

75S. S. Vedam, P. J. Keall, A. Docef, D. A. Todor, V. R. Kini, and R.Mohan, “Predicting respiratory motion for four-dimensional radio-therapy,” Med. Phys.31, 2274–2283s2004d.

76ICRU, ICRU Report 50. Prescribing, Recording and Reporting PhotonBeam TherapysInternational Commission on Radiation Units and Mea-surements, Bethesda, MD, 1993d.

77ICRU, ICRU Report 62. Prescribing, Recording and Reporting PhotonBeam Therapy (Supplement to ICRU Report 50)sInternational Commis-sion on Radiation Units and Measurements, Bethesda, MD, 1999d.

78C. Ozhasoglu and M. J. Murphy, “Issues in respiratory motion compen-sation during external-beam radiotherapy,” Int. J. Radiat. Oncol., Biol.,Phys. 52, 1389–1399s2002d.

79C. Ozhasoglu, M. J. Murphy, G. Glosser, M. Bodduluri, A. Schweikard,

K. Forster, D. P. Martin, and J. R. Adler, “Real-time tracking of the tumorvolume in precision radiotherapy and body radiosurgery-A novel ap-proach to compensate for respiratory motion,” presented at the ComputerAssisted Radiology and Surgery, San Francisco, 2000.

80A. Schweikard, G. Glosser, M. Bodduluri, M. J. Murphy, and J. R. Adler,“Robotic motion compensation for respiratory movement during radiosur-gery,” Comput. Aided Surg.5, 263–277s2000d.

81M. J. Murphy, “Tracking moving organs in real time,” Semin Radiat.Oncol. 14, 91–100s2004d.

82V. R. Kini, S. S. Vedam, P. J. Keall, S. Patil, C. Chen, and R. Mohan,“Patient training in respiratory-gated radiotherapy,” Med. Dosim28,7–11 s2003d.

83H. Blomgren, I. Lax, I. Naslund, and R. Svanstrom, “Stereotactic highdose fraction radiation therapy of extracranial tumors using an accelera-tor. Clinical experience of the first thirty-one patients,” Acta Oncol.34,861–870s1995d.

84R. M. Cardinale, Q. Wu, S. H. Benedict, B. D. Kavanagh, E. Bump, andR. Mohan, “Determining the optimal block margin on the planning targetvolume for extracranial stereotactic radiotherapy,” Int. J. Radiat. Oncol.,Biol., Phys. 45, 515–520s1999d.

85B. D. Kavanagh, R. D. Timmerman, S. H. Benedict, Q. Wu, T. E.Schefter, K. Stuhr, S. McCourt, F. Newman, R. M. Cardinale, and L. F.Gaspar, “How should we describe the radioblologic effect of extracranialstereotactic radlosurgery: Equivalent uniform dose or tumor control prob-ability?,” Med. Phys.30, 321–324s2003d.

86T. Schefter, L. E. Gaspar, B. Kavanagh, N. Ceronsky, A. Feiner, and K.Stuhr, “Hypofractionated extracranial stereotactic radiotherapy for livertumors,” Int. J. Radiat. Oncol., Biol., Phys.57, S282s2003d.

87R. Timmerman, L. Papiez, and M. Suntharalingam, “Extracranial stereo-tactic radiation delivery: Expansion of technology beyond the brain,”Technol. Cancer Res. Treat.2, 153–160s2003d.

88M. Uematsu, A. Shioda, K. Tahara, T. Fukui, F. Yamamoto, G. Tsumatori,Y. Ozeki, T. Aoki, M. Watanabe, and S. Kusano, “Focal, high dose, andfractionated modified stereotactic radiation therapy for lung carcinomapatients: A preliminary experience,” Cancer82, 1062–1070s1998d.

89R. Timmerman, L. Papiez, R. McGarry, L. Likes, C. DesRosiers, S. Frost,and M. Williams, “Extracranial stereotactic radioablation: Results of aphase I study in medically inoperable stage I non-small cell lung cancer,”Chest 124, 1946–1955s2003d.

90R. D. Timmerman, L. Papiez, R. McGarry, L. Likes, C. DesRosiers, S.Frost, and M. Williams, “Extracranial stereotactic radioablation: Resultsof a phase I study in stage I non-small cell lung cancer,” Int. J. Radiat.Oncol., Biol., Phys.57, S280–S281s2003d.

91G. Benchetrit, “Breathing pattern in humans: Diversity and individuality,”Respir. Physiol.122, 123–129s2000d.

92E. N. Bruce, “Temporal variations in the pattern of breathing,” J. Appl.Physiol. 80, 1079–1087s1996d.

93P. Dejours,RespirationsOxford University Press, New York, 1966d, p. 52.94I. P. Priban, “An analysis of some short-term patterns of breathing in man

at rest,” J. Physiol.sLondond 166, 425–434s1963d.95S. A. Shea and A. Guz, “Personnalite ventilatoire—An overview,” Respir.

Physiol. 87, 275–291s1992d.96M. J. Tobin, T. S. Chadha, G. Jenouri, S. J. Birch, H. B. Gazeroglu, and

M. A. Sackner, “Breathing patterns. 2. Diseased subjects,” Chest84,286–294s1983d.

97M. J. Tobin, T. S. Chadha, G. Jenouri, S. J. Birch, H. B. Gazeroglu, andM. A. Sackner, “Breathing patterns. 1. Normal subjects,” Chest84, 202–205 s1983d.

98R. George, P. J. Keall, T. Chung, S. Vedam, and R. Mohan, “Does breath-ing training reduce residual motion for respiratory-gated radiotherapy?”in The Use of Computers in Radiation Therapy, edited by Ahn Yi, Choiand HasJeong, Seoul, Korea, 2004d, pp. 437–441.

99P. Keall, “4-dimensional computed tomography imaging and treatmentplanning,” Semin Radiat. Oncol.14, 81–90s2004d.

951 Keall et al. : 4D radiotherapy planning 951

Medical Physics, Vol. 32, No. 4, April 2005