`
`US 2{)[)40(}71.261/\1
`
`(19) United States
`(12) Patent Application PllbIICflti0Il
`Earl et al.
`
`(10) Pub. No.: US 2004/0071261 A1
`(43) Pub. Date:
`Apr. 15, 2004
`
`(54) N()Vl€I. MI'l'l'H()l) FOR THE PLANNING
`AND DELIVERY OF RADLATION THERAPY
`
`Related U,S_ Applicmigm Data
`
`(75)
`
`Inventors: Matt A. Earl, Columbia, MI) (US);
`l)m'icl M. Shepard. Severn. MD (US);
`xinsheng Yu. Clarksvillc. MD (US)
`
`(__'(}r[u_=,_sp(31'[[_1c1'1(_;9, Address;
`SUGHRUE MION, PLLC
`2l(l'll Pennsylvania Avenue, NW
`Washington, DC 20037-3213 (US)
`(73) Aasigncc: UNIVERSITY OF MARYLAND AT
`BALTIMORE
`
`(21) Appl, No;
`
`10,808,090
`
`(22
`
`Filed:
`
`Dec. 3, 2002
`
`(60)
`
`Provisional application No. fi(J,-338,1 18, filed on Dec.
`3,2uo1.
`
`Publication Classification
`
`II'!L CL? ..................................................... .. Afi1N 5”."
`(5 l)
`(52) U.S. Cl.
`.............................................................. .. 378165
`
`(57)
`
`ABSTRACT
`
`A new optimization mclhnd fur gcncraling trcalmcnl plans
`for raclialion oncology is described and claimed. This new
`melhnd xvurks for inicrisily modulated radiation lhcrapy
`(IMRT),
`intcnsily modulated arc therapy (IMAT). and
`hybrid IMRT.
`
`60—"fi“‘3
`
`{MAT
`
`fixed field and Hybrid
`
`ea‘ “um er.“ ran
`
`Salectnumhero deliul:
`
`
`
`F
`
`I
`Fixed field
`
`:::::::.::=.:::.=.':;::.-:.::=' -'
`:Ll‘culaI:ehI:l1e dose cont!-ibulian
`
`61
`
`'72
`
`63
`
`'7
`
`ang 2
`
`
`
`62“
`
`66
`
`am-imae ma
`‘flies ofdiscreh afigle _
`Assign Initial aperture
`
`shine for each beam
`
`
`di.‘ 5
`d
`I
`am: ‘:5 rglufianiulah
`
`cbjactive func lion
`
`
`
`Dgtide upon
`clmical
`objectives
`
`filler an aperhlne shape
`nr weight based on sorne
`selection prncedure
`
`
`
`Does the change
`satisfy constraints
`defimed in step 54
`
`—.._-
`
`\\.
`
`67
`
`No
`
`63
`
`
`
`Define geomeh-'Ic
`conslraints fnrIleiiuuy
`made and Ilnac
`
`Yes
`
`65
`
`I
`
`ND
`
`54
`
`Is uptimizalion
`
`finisln.-.I.I?
`
`71
`
`69
`
`Calculate new cl-use
`
`Accept or nniacl:
`resulling from change
`and caltulaha nbiecliva
`dlangn based on
`
`function based on new
`optimization mcslhod
`dose
`
`
`
`
`Yes
`
`
`
`0001
`0001
`
`Varian (Ex. 1009)
`Varian (EX. 1009)
`IPR of U.S. Pat. No. 7,961,843
`IPR of U.S. Pat. No. 7,961,843
`
`
`
`Patent Application Publication Apr. 15, 2004 Sheet 1 of 3
`
`US 2004/0071261 A1
`
`50 —-'F\,
`
`Fmedhmdandkwbnd
`
`
`1 Fixed field
`Divide each field inlzu a grid of
`Apprwmtimale art into
`
`discrete pencil beams and
`
`calculaua line due contribution
`series of discrete angles
`frum each
`
`
`66
`Jsslgn Initial aperlutu
`
`shape for each beani
`
`dlretflon and calculate
`Define qeorne l.ri|'.
`drain and value of
`constraints for tlalivufir
`Yes
`
`mod I: and lira ac
`objective function
`
`62b
`
`
`
`
`
`71
`
`
`
`ls optimizaiicm
`
`finished?
`
`
`
`Mizr an i|;|¢I'I‘.l.IlE shape
`on-weight based on some
`seiccflon pl'¢H:EdI.I'E
`
`67
`
`
`
`
`Calculate new dose
`resuihing frnrn dance
`and calculate obiecfivn
`function based on new
`close
`
`
`
`Accept or I-elect
`
`change based on
`
`opflmlz-nliun rnelhod
`
`ND
`
`63
`
`Does this change
`
`satisfy conslnlnls
`defined In ship I54
`
`
`Fig. l §!1IO'I‘E CORRECTION BEING SENT BY FAX!
`
`Egg
`
`Figure 2
`
`0002
`0002
`
`
`
`g H
`Select I'II.IlI‘IIIer of delivery
`
`angies and number of
`ape:-mrel fin-um each angle
`
`52.11
`
`Select nurnher and rang .
`of arts
`
`
`
`
`
`Patent Application Publication Apr. 15, 2004 Sheet 2 of 3
`
`US 2004,/(M71261 Al
`
`F_i2u£§
`
`0003
`0003
`
`
`
`US 2004/0071261 A1
`
`T53$
`
` PatentApplicationPublicationApr.15,2004Sheet3of3
`
`0004
`0004
`
`
`
`US 2004/0071261 A1
`
`Apr. 15, 2004
`
`NOVEL METHOD FOR THE PLANNING AND
`I)F.LIVI".RY OF RADIATION THERAPY
`
`adverse complications can arise in the patient being treated
`because of irradiation of normal structures.
`
`[0001] This is a non-provisional application claiming
`domestic priority from Provisional Application No. 60H338,
`118, filed on Dec. 3, 2001.
`
`[0002] A computer listing of a program according to an
`exemplary embodiment of the invention is submitted here-
`with in a CD—ROM as an Appendix to this application. The
`contents of the Cl)-ROM are incorporated by reference. The
`computer program is subject to copyright protection.
`
`[0003] This invention was made with the support of the
`U.S. government under Grant Number R29CA6607S
`awarded by NIH. The U.S. government has certain rights in
`this invention.
`
`BACKGROUND on THE INVENTION
`
`[0004]
`
`1. Field of the Invention
`
`invention relates to a computerized
`[0005] The present
`method [or the planning and delivery of radiation therapy. In
`particular, it is a computerivied method that determines the
`optimal radiation treatment plan for a patient using specified
`clinical objectives.
`
`[0006]
`
`2. Description of Related Art
`
`[0007] Radiation therapy, in general, is the use of ionizing
`radiation for the treatment ofdiseasc. The most common use
`is in the treatment of cancer. The goal of radiation therapy
`for cancer is to destroy any diseased cells while minimizing
`the damage to healthy tissue. One device for delivering the
`radiation to a patient is with a linear accelerator, a machine
`that generates a high—energy beam of radiation that can be
`controlled and directed onto specified locations. Linear
`accelerators are sometimes equipped with a multi-leaf col-
`limator (MLC), a device that shapes each individual beam of
`radiation.
`
`Prior art treatment planning for conventional can-
`[0008]
`cer radiation treatment is often performed with the aid of
`thrce—dimensional patient images acquired using a computed
`tomography (CT) scanner. Using the three-dimensional
`patient images, the radiation oncologist pinpoints the loca-
`tion of the tumor and any surrounding sensitive structures.
`Using the information provided by the radiation oncologist,
`a treatment planner devises the configuration of radiation
`beams that will deliver the desired radiation dose to the
`patient. The parameters that need to be determined by the
`treatment planner include the beam energies, beam orienta-
`tions, and field shapes. (Levitt ct. al., "Technological Basis
`for Radiation Therapy: Clinical Applications", 3'”
`Lip-
`pincotl, William & Wilkins (1999)) Using a trial-and—ertor
`approach, the treatment planner determines an acceptable
`configuration of the various parameters that meets the clini-
`cal goals
`specified by the radiation oncologist. This
`approach is called “forward-planning" because a human
`being determines the parameters that produce the best treat-
`ment plan. (Levitt, et. al.)
`
`Prior art treatment planning uses a “forward—plan—
`[0009]
`ning” technique for conventional cancer radiation treatment
`by shaping the radiation field. However, shaping the radia-
`tion [ield alone restricts one’s ability to shape the volume of
`the high radiation dose to conform to the tumor. As a result,
`
`[0010] A recent development in radiation therapy is inten-
`sity—moclulated radiotherapy (IMRT) in which the intensity
`of the radiation delivered is modulated within each lield
`delivered.
`(Webb, “The Physics of Conformal Radio-
`therapy”, Institute ol‘ Physics Publishing, Bristol (1997))
`The purpose of IMRT is to sculpt the radiation dose distri-
`bution so that it maximives the radiation dose to the tumor
`while maintaining the radiation dose to normal structures
`within some pre-specified tolerance.
`(Webb)
`In IMRT,
`highly confonnal dose distributions can be achieved through
`the delivery of optimized non—uniform radiation beam inten-
`sities from each beam angle. Successful delivery of IMRT
`can allow for an escalation of the tumor dose and may
`enhance local tumor control. The dosimetric advantages of
`IMRT can also be used to provide a reduced probability of
`normal tissue complications.
`
`[0011] Because of the complexity of the treatment plans
`for IMRT, an automated system is required to determine the
`intensity maps that produce the optimal
`radiation dose
`distribution. In contrast
`to prior art "forward planning"
`techniques,
`this approach is
`terrned “inverse-planning"
`because the automated system determines the parameters
`that produce the optimal radiation treatment plan. {Webb}
`
`[0012] Currently available IMRT delivery techniques
`include fixed field beam delivery (IMRT) and intensity
`modulated arc therapy (IMAT). When radiation is delivered
`with fixed beam angles, a series of beam shapes are deliv-
`erect at each beam angle either dynamically, where the leaves
`of the MLC move during irradiation, or in a step—and—shoot
`fashion, where the radiation is paused during the movement
`of MLC leaves. (Convery and Roseribloom (1992), Bortfeld
`et al (1994), Yu, Symons et al [1995);Boyer A.L., and Yu
`C.)(.; (1999);) In contrast, IMAT uses multiple overlapping
`arcs of radiation in order to produce intensity modulation.
`(Yu, c.x. (1995); Yu et al (2002))
`
`[0013] The complexity of IMRT and IMAI‘ is such that
`treatment plans cannot be produced through a manual trial
`and error approach. Instead, one must employ an automated
`treatment planning system. Furthermore, current automated
`planning tools are not capable of producing optimized plans
`for IMAT.
`
`[0014] Current inverse -planning algorithms for IMRT use
`a two—stcp approach (Boyer and Yu 1999}. In the first step,
`the portal that defines the radiation beam's eye view (BEV)
`for each radiation beam angle is divided into a set number
`of finite-sized pencil beams. The radiation dose for each of
`these pencil beams is then calculated and the corresponding
`beam intensities are subsequently optimized subject to pre-
`speciiied treatment goals. The second step uses the radiation
`intensity maps from each beam angle and translates the
`radiation intensity maps into a set of deliverable aperture
`shapes. During the optimization of the radiation intensity
`maps,
`the delivery constraints imposed by the design of
`various components of the linear accelerator are not taken
`into account
`resulting in treatment plans that are often
`complex and ineflicient to deliver.
`
`[0015] The two step approach used by current inverse-
`planning algorithms is unable to generate treatment plans for
`IMAT. With IMAT.
`the radiation is delivered while the
`
`0005
`0005
`
`
`
`US 2004/0071261 A1
`
`Apr. 15, 2004
`
`gantry rotates continuously. Current inverse—planning algo-
`rithms fail to take the gantry‘s continuous movement into
`account. One feature of IMAT treatment plans is that the
`aperture shapes for adjacent angles within an arc must not
`ditfer significantly. This constraint exists because there are
`limitations on the speed at which the leaves ofthe multileaf
`collimator can travel. This constraint makes it dillicult to
`translate the radiation intensity maps into a set ofdeliverable
`¢':ll'CS.
`
`[0016] This invention is an inverse—planning method that
`does not
`require the current
`two-step approach used for
`IMRT treatment planning. This invention allows for the
`planning for either IMRT, IMAF, or a new type of intensity-
`modulated radiotherapy which comprises a combination of
`IMRT and IMAT. This combination of IMRT and IMAT
`represents a hybrid approach to IMRT. Ilybrid IMRT pro-
`vides the ability to incorporate into each treatment plan the
`dosimetric advantages of both IMRT and IMAT. For
`example.
`the rotational nature of IMAT can be used to
`dissipate the deposition of radiation dose to normal tissue
`while the fixed field capabilities of IMRT allow for a high
`degree of modulation from any particular beam angle.
`
`SUMMARY OF THE INVENTION
`
`It is the purpose ofthis invention to enable a single
`[0017]
`planning system to plan for different modes of IMRT deliv-
`ery and to simplify the planning and delivery of IMRT.
`Instead ofoptimizing the intensity distributions of the beams
`and then converting them to deliverable MLC field shapes,
`this invention directly optimizes the shapes and the corre-
`sponding weights of the apertures. The combination of these
`optimally weighted apertures at every beam angle creates
`highly modulated beam intensity distributions for achieving
`the clinical objectives of the treatment plan. In the process
`of optimizing the field shapes, all delivery constraints are
`considered. For instance, fixed-field delivery would have
`constraints imposed by the MLC. Rotational delivery would
`have additional constraints imposed by the speed of the
`gantry rotation and speed of the MLC leaves.
`
`For tixed—lield delivery, the user specifies the nu m-
`[0018]
`ber of beams and their angles, the beam energies, and the
`nu mher of apertures per beam angle. For rotational delivery,
`the user specifies the number and range of the arcs. The
`goals of the treatment plan are determined and then quan-
`tilicd with an objective function, which can be of dose-
`volumc based. biological, or of other forms.
`
`For each delivery angle, the maximum extent of the
`[0019]
`beam aperture is determined based on the beam’s eye view
`of the target with sutficient margins. This beam is then
`divided into a grid of small bcamlcts called pencil beams.
`The dose distribution from each of these pencil beams is
`calculated using any conventional dose calculation method
`and stored on an appropriate medium, such as a hard drive.
`At the start of the optimization, all apertures in the same
`beam direction are set to the same shape as the maximum
`extent ofthe beam. These apertures are then optimized by an
`optimization algorithm. The optimization process generally
`involves modifying the shape or weight of the apertures,
`determining if the modification violates the delivery con-
`straints, and, finally, accepting and rejecting such modifica-
`tions based on the rules of the optimization. For each
`modification. a new dose distribution computed based upon
`
`the modified aperture shapes or weights. While simulated
`annealing lends itself well to the optimization method, other
`optimization techniques could also be used.
`
`[0020] The output of the algorithm is a set of deliverable
`apertures and their weights, which can be transferred to the
`control system of a linear accelerator and delivered to a
`patient. Because of the feature of pre-specification of the
`number of angles and apertures, the user controls the corn-
`plcxity of the treatment plan. Because the invention can
`incorporate the delivery constraints for each particular linac
`and MLC, it can be used in conjunction with any commer-
`cially available linear accelerator.
`
`BRIISF Dl.iSCl{Il"I'IOl\| OI‘ TI-IE DRAWINGS
`
`[0021] FIG. 1 shows the How chart for direct aperture
`optimization.
`
`[0022] FIG. 2 illustrates three aperture shapes determined
`using direct aperture optimization.
`
`[0023] FIG. 3 illustrates the intensity map for three aper-
`ture shapes determined using direct aperture optimization;
`
`[0024] FIG. 4 illustrates an apparatus according to an
`embodiment of the invention.
`
`DETAILED DESCRIPTION OF THE
`INVENTION
`
`[0025] Referring to FIG. 4, a linear accelerator (linac) 1
`which is a device capable of controlled delivery of radiation
`to a patient in need of radiation therapy. The radiation exits
`through the end of the treatment head which is mounted on
`the gantry (not shown}. In some linacs, the treatment head is
`equipped with a mu lti—leaf collimator (MLC) which shapes
`thc radiation field. Alinac has a control unit in a housing. A
`linac has a gantry which can rotate about a horizontal axis
`II of rotation around the patient who is lying on the bed. A
`linac emits a beam of radiation which is aimed at the patient.
`The beam of radiation can be photons, electrons, or any
`other type of radiation used for therapy.
`
`[0026] During treatment, the radiation beam is directed on
`a part of the treatment area on the patient. The gantry can
`rotate about a horizontal axis of rotation; thus allowing for
`a change in the angle of treatment.
`
`[0027] A MI..C has multiple thin leafs which can be made
`of tungsten alloy or other heavy materials stacked in two
`opposing hanks MLCI, MLC2. For one MIX.‘ the leaves are
`usually identical in width, range of travel, and restrictions in
`relation to the other leaves in the same bank or opposing
`banks. MLC leaf restrictions can be characterized as static
`constraints and dynamic constraints. Static constraints can
`include, but are not
`limited to,
`the maximum distance
`between the most forward position and the most backward
`position of any leaf in one bank and the minimum distance
`between opposing leaves in opposing banks. However, it is
`understood that different MI.C"s can have widths ranging
`from 2 mm to 12 mm, range of travel ranging from 1 cm to
`over 32 cm, and different restrictions. Dynamic constraints
`include, but not
`limited to,
`thc speed of leat‘ travel,
`the
`acceleration and deceleration. These static and dynamic
`geometric constraints determine the kind of aperture shapes
`that a particular MLC can form.
`
`0006
`0006
`
`
`
`US 2004/0071261 A1
`
`Apr. 15, 2004
`
`DJ
`
`[0028] Within a linac and in addition to the MLC, a beam
`shielding device SI.D is provided in the path of radiation
`beam to supplement the MLC in shaping the radiation fields.
`The beam shielding device includes a plurality of opposing
`plates. In one embodiment, additional pairs of plates are
`arranged perpendicular to the opposing plates. The opposing
`plates move with respect to the plate axis by a drive unit to
`change the size ofthe irradiated field. The drive unit includes
`an electric motor which is coupled to the opposing plates and
`which is controlled by a motor controller. Position sensors
`are also coupled to the opposing plates, respectively for
`sensing their positions. The plate arrangement may alterna-
`tively include a multi-leaf collimator (MLC) having many
`radiation blocking leaves.
`
`In an MLC, there are opposing banks of leaves.
`[0029]
`Each opposing leafis attached to a drive unit. The drive units
`drive the leaves,
`in and out of the treatment field,
`thus
`creating the desired lield shape. The MLC leaves. are
`relatively narrow, and cast a shadow of about 0.5 cm to 1.0
`cm onto the treatment area. The position of the leaves of the
`MLC defines the aperture shape for a treatment.
`
`[0030] The intensity of a beam refers to the amount of
`radiation that accumulates at
`a specific location of the
`treatment portal defined by the linac.
`
`[0031] A longer radiation exposure time for a specific
`location in the treatment portal corresponds to a higher
`radiation intensity. If the MLC opening is fixed during the
`entire duration of treatment, all locations in the treatment
`portal would receive approximately the same amount of
`radiation, and there would be no intensity modulation. A
`modulated intensity radiation field occurs when the MLC
`opening changes such that ditferent locations of the treat-
`ment portal are exposed for diflerent durations.
`
`[0032] The motor controller is part of the Linac Control
`System (LCS) that also contains a dosimetry system. The
`dosimetry system measures the output ol’ the radiation beam
`with a measuring chamber MC and reports to the Linac
`Control System {I..CS) the amount of radiation being deliv-
`ered at any given time. The LCS coordinates radiation
`delivery and MLC leaf movement in order to achieve the
`desired intensity patterns. The LCS controls execution of the
`prescription generated by the present invention and trans-
`ferred to the linac control system from the treatment plan-
`ning system. During delivery, the MLC leaves move in order
`to achieve the desired treatment.
`
`[0033] During treatment planning, a user is allowed to set
`the mode of treatment including IMRT or IMAT or a hybrid
`thereof, and to provide other treatment parameters such as
`the orientations of beams, ranges of arcs,
`the number of
`apertures pet‘ beam angle andfor the number of arcs. Using
`the invention described herein, the planning system auto-
`matically optimizes the shape and weightings of the aper-
`tures to best meet the objectives of the treatment The end
`product of the treatment planning process is a treatment plan
`that meets the dosimetric requirements specified by the
`physician. Once a treatment plan is approved by the physi-
`cian, the treatment planning system will generate a prescrip-
`tion, which specifies the proper coordination between radia-
`tion delivery and MLC leaf movements. The prescription.
`therefore, translates the treatment plan into the computer
`language understood by the Linac Control System (LCS)
`and programs the linac for the treatment delivery. The
`
`treatments can be entered
`prescription of conventional
`manually using a keyboard or other input type of device. For
`IMRT delivery, because of the complexity of the prescrip-
`tion, prescriptions are normally entered via digital media,
`such as a diskette or CD, or a network link, or any other
`input type of device. At a given time du ring the delivery of
`radiation to a patient, the LCS is receiving infonnation on
`close delivery from the dose control unit. The LCS also
`receives information in real-time from the MLC.‘ position
`sensors. The LCS compares the dose delivery information
`from both the MLC.‘ controller and the dosirrtetry system
`controller with the prescription. Depending on the result of
`the comparison,
`the LCS may respond in a variety of
`manners. For example, the LCS may send a signal to the
`beam triggering system to pause the radiation so that the
`MLC can advance to the proper location.
`
`[0034] The present invention covers the method of plan-
`ning and delivery of the radiation treatment plan for IMRT,
`IMAT, and hybrid IMRT. The treatment planning procedure
`is perforrned on a
`treatment planning system which is
`distinct from the LCS, so that the treatment planning system
`can generate IMRT treatment plans for all commercially
`available linacs and MLC’s. Prior art IMRT planning inven-
`tions can only plan for lixed-field IMRT delivery but not
`IMAT or hybrid IMRT (US. Pat. No. 6,240,161 (Siochi);
`U.S. Pat. No. 6,260,005 (Yang, et al.)) and there is no
`distinct separation between the treatment planning system
`and the LCS.
`
`[0035] Direct aperture optimization (DAO) which is
`described herein optimizes the position of the MLC leaves.
`thus optimizing the aperture shapes, and optimizes each
`aperture shape’s corresponding intensity based on the treat-
`ment goals for a specific patient. With DAO, the geometric
`constraints of a MLC associated with either IMRT, IMAT, or
`hybrid IMRT are incorporated during the optimization pro-
`cess, thereby permitting the development ofa treatment plan
`for IMRT, IMAT, and hybrid IMRT in one system. I)/\O is
`an improvement over prior arts optimization methods
`because in the prior art methods each system is dedicated to
`only gantry-fixed IMRT. Inverse planning for IMAT and
`hybrid IMRT was not possible with prior arts. Another
`distinguishing feature of DAO is that all of the geometric
`constraints imposed by the treatment unit are incorporated
`into the optimization. Examples of geometric constraints [or
`the MI..C and linac include, but are not limited to, the dose
`rate, gantry speed, and minimal amount of radiation that can
`be delivered with acceptable accuracy.
`
`[0036] FIG. 1 shows a flow chart of the DA0 procedure.
`In a first step 60, the mode of delivery is selected. The modes
`of delivery include IMRT, IMAT, or hybrid IMRT. If tixed
`field lMR'l‘or hybrid IMR'I‘isselected, in a step 61, the user
`must select the delivery angles and the number of apertures
`assigned to each angle. Then one proceeds to step 62:: ifonc
`selected hybrid IMRT in a step 60. Otherwise, if one selected
`fixed field IMRT in a step 60, then one proceeds immediately
`to a step 63. If the user selects IMAT in a step 60, then the
`user proceeds immediately to step 620.
`
`In a step 62rr, one must select the number of arcs
`[0037]
`and the range for each arc. After the consideration factors
`(the delivery angles and number of apertures assigned to
`each angle for DIRT or the number of arcs and range for
`each are for IMAT} are entered, in a step 626. the treatment
`
`0007
`0007
`
`
`
`US 20l34/0071261 Al
`
`Apr. 15, 2004
`
`planning system automatically calculates evenly spaced
`radiation beams to approximate the range of rotation of the
`gantry. Hybrid lMR'l‘ required both steps 61 and 62 to
`account
`for the combined use of fixed field and arced
`delivery.
`
`In a step 63, each field is divided into a grid of
`[0038]
`discrete pencil beams and the dose distribution for each
`pencil beam is computed. The MLC delivery constraints for
`fixed field delivery are determined in a step 64. For rotation
`delivery in a step 64, the constraints associated with rota-
`tional delivery are also determined to ensure not only
`coordination of MLC movement with radiation delivery, but
`also the synchronization of radiation delivery and gantry
`rotation.
`
`In a step 65, the user defines the clinical objectives
`[0039]
`of the treatment plan. These clinical objectives are used to
`score the quality of the treatment plan throughout
`the
`optimization process. The treatment plan quality can be
`scored by an objective function that reduces the treatment
`plan into a single numerical value. The objective function
`can be, by way of example only, a least-square dose differ-
`ence objective between the desired dose and the achieved
`dose. The objective function can also be based on dose
`volume histograms (DVH) or biological based parameters.
`
`[0040] The optimization process begins in a step 66,
`where the treatment planning system assigns an initial
`aperture shape for each beam angle. in the preferred embodi-
`ment, the radiation beam‘s eye view of the target for each
`beam angle is used for the starting point, but any aperture
`shape for each beam angle can be used. The treatment
`planning system also assigns a relative weight (intensity) to
`each aperture shape.
`In addition,
`the treatment planning
`system calculates the radiation dose,
`the radiation dose
`distribution, and the dose distribution quality (objective
`function}.
`
`[0041] After obtaining an initial score for the dose distri-
`bution quality of the plan, the treatment planning system, in
`a step 67, modifies an optimization variable. The optimiza-
`tion variables that the treatment planning system considers
`include, but are not
`limited to, the positions of the MLC
`leaves used to shape each aperture for each beam angle, and
`the relative weight
`(intensity) of each aperture shape
`assigned to each aperture. A stochastic or deterministic
`approach can be used to determine the variable for modifi-
`cation and the size of the modification.
`
`Prior to calculating the new dose distribution and
`[0042]
`objective function resulting from the modification of the
`optimization variable in a step 67, the treatment planning
`system determines, in El step 68, it‘ one or more geometric
`constraints is violated by the modification. Examples of
`geometric constraints include, but are not
`limited to,
`the
`MLC leaf positions for the particular linear accelerator, the
`linac gantry speed, the dose rate, and MLC leaf travel speed.
`If the proposed modified aperture shape or intensity violates
`any of geometric constraints. the treatment planning system
`rejects the modified aperture shape and returns to a step 67.
`
`If none of the geometric constraints is violated in a
`[0043]
`step 68, then the treatment planning system calculates the
`radiation dose applied to the treatment area as a result of the
`modification. The value of the objective function is calcu-
`lated from the new radiation dose. and the dose distribution
`
`quality is compared to the dose distribution quality prior to
`the modification. If the value of the objective function
`changed in the desired direction,
`the treatment planning
`system accepts the proposed modification of the aperture
`shape. If the radiation dose and dose distribution quality are
`not within acceptable ranges or
`the objective function
`changes in the undesirable direction, the treatment planning
`system either accept or rejects the proposed modification of
`the aperture shapes based on a series of pre-set mles and
`returns to a step 67.
`
`In the preferred embodiment of this invention, a
`[0044]
`simulated annealing algorithm is used in steps 67 through 70
`to determine the optimal aperture shapes and corresponding
`weights. The optimization algorithm randomly selects a
`variable from the set of variables considered in the optimi-
`zation process, i.e., the MLC leaf positions and the weights
`of the aperture shape. For the selected variable, a change of
`random size is sampled from a probability distribution. For
`instance, a Gaussian distribution could be used. In addition,
`the shape of the curve could change with successive iteration
`of the procedure. For instance, the width ofthe Gaussian plot
`could decrease according to some schedule such as in
`Formula (1):
`
`Lt'=l+IA.—l]£'
`
`'?SI!hm_'iJ
`rill“.
`
`I I t
`
`[0045] where A is the initial Gaussian width, nmc is the
`number ofsuccessful iterations, and 'l""°",, quantifies the rate
`of cooling. Although the above schedule is specific, any
`schedule can be used. For instance, the step size could be
`constant throughout the optimization. The goal of this inven-
`tion is to achieve the optimal aperture shape [or each beam
`angle as quickly as possible. Decreasing the amplitude of
`change as the optimization progresses allows coarse samples
`in the beginning and line-tuning at the end of the optimiza-
`tion process.
`
`[0046] Other types ofoptirnization algorithms can be used
`in this invention such as conjugate gradient or genetic
`algorithms.
`
`[0047] Based on pre-defined termination criteria which are
`dictated by the optimization algorithm, the treatment plan-
`ning system will cease the optimization process in step 71.
`The plan with the optimal value of the objective function is
`deemed the optimal plan. This optimum treatment plan is a
`set of deliverable aperture shapes and the intensities asso-
`ciated with each aperture shape. Monitor units are units of
`radiation output from a linac.
`
`In a step 72, the treatment planning system pro-
`[0048]
`vides the optimum treatment plan and linal radiation dose
`distribution to a user for review by displaying the optimum
`treatment plan on a display screen, or printing it out using a
`printer, or placing it on some other user interface which is
`known in the art field.
`
`the final radiation dose
`In an optional step 73,
`[0049]
`distribution resulting from the optimum treatment plan is
`optionally reviewed and approved by a user capable for
`reviewing such information.
`
`In a step 74, after optional review and approval, the
`[0050]
`optimum treatment plan is transferred from the treatment
`
`0008
`0008
`
`
`
`US 2004/0071261 A1
`
`Apr. 15, 2004
`
`planning system performing the direct aperture optimization
`to the I..CS in the form of a Prescription file. The optimal
`treatment plan is loaded onto the LCS via a diskette, a
`computer network link, or any other means known in the art
`field capable of transferring data between two distinct corri-
`puters. This invention allows the direct aperture optimiza-
`tion information to be transmitted from the treatment plan-
`ning system located at one site to the linac control system
`(LCS) located at a diflerent site.
`
`[0051] Because the treatment planning system is distinct
`from the linac control system (LCS), one can optimize
`several ditferent treatment plans for dilferent types of linear
`accelerators in succession or concurrently.
`
`[0052] FIG. 2 illustrates three aperture shapes obtained by
`using the DAO of this invention assigned to a radiation beam
`direction. As compared with the aperture shapes obtained
`from a typical
`leaf sequencing step using the prior art
`treatment planning programs,
`the exposed area of each
`aperture shape is significantly increased, resulting in greater
`eificiency in delivery.
`
`[0053] FIG. 3 illustrates the intensity distribution created
`with the three apertures shown in FIG. 2. Theoretically, the
`number of intensity levels, N, resulting from n apertures can
`be expressed as: N=2"-1. For example, with three aperture
`shapes per beam, seven intensity levels can be created.
`Moreover, because each intensity level
`is a free—floating
`percentage of the maximum intensity as compared to fixed
`percentage ofthe maximum intensity in the previous arts of
`IMRT planning, the seven intensity levels created by over-
`lapping directly optimized apertures give more llexibility to
`the planning system in creating optimal treatment plans. In
`the prior art IMR'l' treatment planning, an intensity pattern
`containing 7' intensity levels would require 15 to 30 aper-
`tures to realize, resulting in very inefficient treatment deliv-
`ery. Moreover, when such large number of apertures is used,
`the aperture shapes are generally small, requiring very high
`accuracy in the positions of the MLC leaves. As the result,
`quality assurance efforts must be intensified to levels much
`beyond conventional
`treatments. With
`the
`invention
`described herein,
`the benefit of IMR'l‘ can be realized
`without such inetficiertcy and labor intcnsiveness associated
`with IMR'l' using prior arts.
`
`5. Yu, C. X.; Syrnons, M. .|.; Du, M. N.; et al.; “A
`