S4

sWHEN DOES lREATMENT PLAN OPTIMIZATION REQUIREINVERSE PLANNING?

George W. Sherouse, Ph.D., Department of Radiation Oncology,Medical University of South Carolina, Charleston, South Carolina,USA

Increasing maturity of image-based computer-aided design of three­dimensional conformal radiotherapy has recently sparked a great dealof work in the areaof trea1rnent plan optimization. Optimization of aconformal photon beamtrea1rnent planis that exercise through whichaset of intensity-modulated static beamsor arcs is specified such that,when the plan is executed, I) a region of homogeneous dose isproduced in the patient with a shape which geometrically conforms(within a specified tolerance) to the three-dimensional shape of adesignated target volume and 2) acceptably low incidental dose isdelivered to non-target tissues. Interest in conformal radiotherapyarises froma fundamental assumption that thereis significant valuetobe gained from aggressive customization of the trea1rnent for eachindividual patient In ourefforts to designoptimal trea1rnents, however,it is important to remember that, given the biological and economicrealities of clinical radiotherapy, mathematical optimization of dosedistribution metrics withrespect to someminimal constraint set is not anecessary or evensufficient condition fordesignof a clinically optimaltrea1rnent. There is wide variation in the complexity of the clinicalsituations encountered in practice and there are a number of non­physical criteria to be considered in planning. There is also acomplementary variety of computational and engineering means forachieving optimization. Todate,thescientific dialogue regarding thesetechniques hasconcentrated on development of solutions to worst-casescenarios, largely in the absence of consideration of appropriatematching of solution complexity to problem complexity. It is the aimof this presentation to propose a provisional stratification of trea1rnentplanning problems, stratified by relative complexity, and to identify acorresponding stratification of necessary trea1rnent planningtechniques. It is asserted that the subsetof clinical radiotherapy casesfor which the most sophisticated techniques of inverse planningrepresent a cost-effective solution is relatively small, and thatsurprisingly simple optimization techniques (and correspondinglysimple trea1rnent delivery techniques) can reliably produce acceptableresults forthemajority of routine conformal radiotherapy practice.

7

STATUSOF THE IAEA CODE OF PRACTICE FORTHE CALIBRATIONANDUSE OF

PLANE·PARALLEL IONIZATIONCHAMBERS

PEDRO ANDREODepartment of Medical Radiation Physics, Karolinska Institutet­University of Stockholm, 17177 Stockholm, Sweden

A new IAEA Code of Practice for parallel-plate ion chamberscoml?lementing and updatingTRS-277 is scheduled for 6-testingwithin 1995. The working group is formed by P. Andreo(chairman), P. Almond,O. Mattsson, A. Nahumand M. Roos. Itsinitial contents was presented during ESTRO-94 (Granada),emphasizing the method of choice for calibration using high­energyelectronbeams. Alternative methods in 6OCO beamsare alsoincludedin the protocol togetherwith an extensive discussion ontheir limitations and increased uncertainty. These are mainlycausedby thelarge variation of the factorsk(ll and pw~ll for plane­parallel chambers of the same design and discrepancies betweenMonteCarlocalculated andexperimental values.

Perturbation effects in electron beamshavereceivedconsiderableattention, yielding changes in the previous outline. The presentknowledge of correction factors for backscatter effectsdue to thedifferentmaterials in COmmercial chambers is inconclusive and arecommendation on numerical valuescannotbe given.

In consistency with the actual trendsfor calibrating ion chambersin terms of absorbed dose to water, both No air and Nowformalisms are included. New symbols are adopted wherethe firstsubscript denotes the calibration quantity and the second is themedium where the quantity is measured; if needed a third indexindicates the quality of the beam used for calibration. AdoptingND w it is no longer possible to sustain the concept of a globalcen'tral electrode correction factor for cylindrical chambers as inTRS-277, as only the contribution ansing during in-phantommeasurements is relevant. The majorconsequence is that the Noail.factorincludeskeel and therefore it is l!:l!!x a factorcharacteristic 'otthe ion chambervolume. The practical (numerical) consequencesof these changes are almost negligible in most cases but it is aconceptual modification. Updated valuesfor peel areincluded. Notethat although the Code of Practice is addressed to plane-parallelionizationchambers. thimblechambersare used as reference forcertaintypesof calibration.

6

A COMPARISON OFFORWARD PLANNING ANDOPTIMISEDINVERSE PLANNING

MarkOldham, Anthony Neal,SteveWebbPhysics Dept.,TheRoyal Marsden NHSTrust:Institute of CancerResearch, London,UK

A radiotherapy treatment plan optimisation algorithm has beenappliedto 48 prostate plansandtheresultscompared withthoseof anexperienced human planner. Twelve patients wereused in the study,and a 3, 4, 6 and 8 field plan(with standard coplanar beamangles foreach plan type) were optimised by both the human planner and theoptimisation algorithm. The human planner "optimised" the planbyconventional forward planning techniques. The optimisationalgorithm was based on fast-simulated-annealing. "Importancefactors" assigned to different regions of the patientprovide a methodfor controlling the algorithm, and it was found that the same valuesgave good results for almost all plans. The plans were compared onthebasisof dosestatistics and normal-tissue-complication-probability(NTCP) and tumour-control-probability (TCP). The results showthat the optimisation algorithm yielded results that were at least asgood as the human planner for all plan types, and on the wholeslightly better. A study of the beam-weights 'chosen by theoptimisation algorithm and the planner will be presented. Theoptimisation algorithm showed greater variation, in response toindividual patient geometry. For simple (e.g. 3 field) plans it wasfound to consistently achieve slightly higherTCP and lower NTCPvalues. For more complicated (e.g. 8 fields) plans the optimisationalso achieved slightly better results with generally less numbers ofbeams. The optimisation timewas always :55 minutes; a factor of upto 20 timesfasterthanthehuman planner.

8

DESIGN AND PERFORMANCE CHARACTERISTICS OF PLANE­PARALLEL CHAMBERS

M. RoosPhysikalisch- TechnischeBundesanstaltD38116 Braunschweig. Germany

Plane-parallel chambers are established as the detectors of choicein clinical electron dosimetry, particularly at low energies. One ofthe mainadvantages is the possibility of minimising the perturbationof the radiation field by the air cavity. Furthermore. these chambersmay be designed so that it may be assumed that they samplealmost exclusively the electron f1uence incident through the frontwindow, justifying the effective pointof measurementto be taken asthe centre of the front surface. of the air cavity. independent ofelectronenergyand depth inthe phantom.The desirable properties of plane-parallel chambers are discussedincluding, forinstance.also the wall materials of the chambers. theirpolarity effect. the stabilization time and the saturation behaviour.Furthermore the results of measurements are reported which wereperformed at the PTBlinearaccelerator in the energy range from 6MeV up to 20 MeV. including variouschambertypes.The range of the perturbation effectby the air cavity at the depth ofmaximum dose extends from negligible values up to more than 2%or even 3% at the lowestenergy. It is shown that in the case of thetypes revealing the largest perturbation effect. the effective pointofmeasurement is essentially displaced from the front surface of theairvolume.A systematic investigation of the polarity effect. its depthdependence. the stabilization time after reversal of the polarity ofthe polarizing voltage etc., also yields strongly chamber typedependent results. which have to be taken into account in practicaldosimetry. With respect to the saturationcorrection it turns out, forinstance. that forsome chambertypes common correction schemeslike the "two-voltage technique" are only applicable withprecautions.