- Email: [email protected]
Physical Models for 3D-Radiotherapy Planning with Photon Beams
TAGUNGSBERICHTE
Physical Models for 3D-Radiotherapy Planning with Photon Beams Anders Ahnesjo Helax AB, Research Department, Uppsala
Abstract Dose calculation methods f or 3D treatment plannin g must be general enough to consider the degrees of fr eedom available for treatment delivery, without loosing accuracy in the final result . This is the motivation f or the development of enery deposition kernel models which will be reviewed here. Zus ammenfassung Dosisberechnungsmethoden fiir 3D-Bestrahlungsplanung miissen allgemein genug sein, urn die Freiheitsgrade der Behandlungsdu rchfiihrung zu beriicksichtigen, ohne jedoch im Enderg ebnis an Genauigkeit zu verlieren. Dies ist die Motivation fii r die Entwicklung von Modellen fiir Energ ieiibertragungskerne, welche hier behandelt werden. Key wor ds : Treatment Planning, Physical Models, Energy Deposition Kernels
The Task of Dose Calculat ion In the present development of 3D radioth erapy there is a will to exploit all degrees of freedom provided by the treatment machines such as dynamic multileaf etc. Hence, the basic design criterion of dose calculation algorithm s is to calculate dose, within clinic ally acceptabl e time limits, for the necessary treatment situations with an accurac y that does not significantly degrade the overall treatment accurac y. The main dosimetric uncertain ties, other than from treatment plann ing, have been estimated [6, 9] to 4.3% with a potential to achieve 2.5% with future development. Hence the ultimate goal for dose calculation algorithms is to achie ve an error of less than I%. At the present state of the art, an error of 2% - 3% does not lower the overall treatment precision significantly, while a 5% error makes dose calcul ations the bottle neck of accuracy.
Energy Deposition Kernels The basic physics for therapeutic photon beams is well known and easy to characterize by use of microscopic
102
cross sections for each type of interaction. The number of interactions required before the energy of a photon entering a patient is completely absorbed is limited, but analytical photon transport methods for dose computation fail due to the complexity of the integrals involved . The only exception is the transport of unscattered, primary photons where simple attenuation and projection laws apply. Hence, Monte Carlo simulation is the only available method that can yield macroscopic dose distributions from the microscopic description using interaction cross sections. To achie ve acceptab le accuracy for dose calculations, however, the use of Monte Carlo require s the simulation of at least 108 photons. Despite the rapid development of computer hardware this is far to time consuming for clinical use. Convolution/superpo sition methods using energy deposition kernels provide a compromise for combining analytical calculations of the primary photon transport with Monte Carlo transport of secondary particles. Calculat ion speed is achieved by doing the Monte Carlo simulation in advance for simplified geometries and reusing the result as convolution kernels. Some accuracy is thereby traded since approximate sea-
ling laws must be applied when using such kernels. The kernel for energy deposition around a primary photon interaction site is commonl y referred to as a photon point spread function (see cover picture of this issuer.Other usable kernels are the pencil beam and planar spread function, cf. fig. I.
Dose Calculation using Point Spread Functions Kernel summa tion techniques It is, of course, impossible to calculate an "exact" kernel for all possible beam configuration s and tissue distributions. The common approach is to scale all dose fractions of a point spread function calculated for a homogeneous medium by the mean density between the point of energy release and deposition. This scaling is consistent with the theorems of O'Connor and Fano. Density scaling of a discrete kernel is equivalent to rebinning a homogeneou s kernel into bins with extensions according to the current inhomogenities. This is trivial for kerne l regions of low gradients but must be done under energy conservation where gradient s are greater to avoid artefact s. The effect
Z. Med, Phys. 4 (1994) 102 - 104
TAGUNGSBERI CHTE
of high kernel gradients is probably the main reason for differences in results from apparently equivalent methods as reported by Wong and Purdy [12]. Several methods can be used for summation of density scaled kernels in dose calculations. These methods, however, are extremely time consuming, so that the search for alternatives has been intensive. In homogeneous media one can use transform based convolution algorithms which yield drastically reduced computation times [7]. In heterogeneous media , however, a full density scaling cannot be combined with transform convo lutions, unless very simp lifying assumptions are made. Boyer and Mok [81 proposed a method that sca led the scatter do. e kernel by the density at the scatter site only. The method has been improved [I] to include the densities at both the scattering and the receiving sites for scaling of the scatter kernel. However, this concept fails to scale the primary dose due to high gradients of the primary dose kernel.
Collapsed cone convolution Polyenergetic point spread functions for treatment machine spectra can be accurately parameterized by a sum of two exponentials over the squa re of the radius. This enable s a semi-analytical approach as used in the colla psed cone method [4]. In this model the point spread function is discretized into a finite number of conical sectors. The energy transport within each cone is approximated to take place on the symmetry axis of the cone, i.e. the cone is "collapsed" onto its axis. A lattice of cone axes is constructed throughout the irradiated volume such that one axis from each direction intersects at each calculation point. The kernel summation is then done by analyt ical convol utions in disc rete steps along each axis using recursive formulas . The recursions reduce the total number of operations needed to a value proportional to N3 for calculation of the dose to 3 point .
Point spread function
Figure I
Pencil beam kernel
Planar spread function
Irradiation geometries for different types of energy deposition kernels.
Dose Calculation using Pencil Beam Kernels The use of pencil beam kernels instead of point spread functions can be viewed as a "preconvolution" over the depth dimension to save computation time. Runtime convolution of the kernel over the beam aperture is then done assuming a flat beam entrance surface. The influence of patient curvature is taken into acco unt by using kernel values for the depth at the calculation point, as beam library models scale measured dose distributions due to patient curvature. Hence, the pencil beam method shares some of the limitations of the older, empirical methods. The main adva ntages of the pencil beam method are that it is generalized for irregular fields, handles latera l fluence varia tions, calculates the dose in energy fluence normalized unit that can be used to derive output factors , and is based on the same set of data as a full 3D convolution model. This type of model com bines naturally with ID heterogeneity corrections that apply the forward part of a planar spread kernel along the beam path to each calculation point. Such simp le methods are surpris ingly accurate provided lateral charge d particle equilibrium is maintained (3).
Beam Data Acquisition for Kernel Based Methods A common approximation is to assume the spectrum to be laterally invariant and handle the lateral variation
of the energy fluence separa tely, i.e. the energy fluence distribution is factorized into a spectrum and a lateral distri bution:
To be clinically practical, the characterization of a therapy machine must be based on easily obtainable data such as depth dose curves and profiles measured in a water phantom. Such a procedure could be ummarized as follows: I. Determination of an effective spectrum WE and charged particle contamination from depth dose data. It is hereby necessary to use constraints while minimizing the difference between measured and reconstructed depth dose curves [2]. Once acqui red, the spectrum is used for calc ulation of kernels and attenuation coefficients from databases for monoenerget ic photons . 2. Characterization of the lateral energy fluence distribution W(x, y) by deconvolution of a measured dose distribution for the largest possib le field [5, II]. The fluence distribu tion for the larges t open field is the base on which beam collimation and modulation (wedges , filter, etc.) is applied. 3. Mode lling dose from contaminating particles [2, 3, 10]. All irradiate d parts of the treatmen t head act as diffuse irradiators of both scattered photons and charged particle s. Separate models are required for calculating the contribution from contaminating particles as they are not mod-
103
TAGUNGSBERICHTE
elled thro ugh co nvo lution of the basic kerne ls for the primary beam. It is espe cially important to consi der the photons scattered in the treatm ent head not only when mode lling output factors, but also for profiles in dyna mically modulated fields.
[2 /
[3)
[4)
Conclusion The main adva ntages of kernel based models. com pared to conventional beam library models. are their greater ge nerality and accuracy for the higher degree of freedom involve d in 3D radiotherapy planning.
References
[5 ]
16)
[71 [8 )
[I ) Ahnesjo A.: lnvariance of convolution kernels applied to dos e calculat ions for pho -
ton beams. Proc 9th Int. Con f. Compo Rad. Ther.• IX ICeR (1987) 99-102 Ahnesjo A.. and And reo P.: Determinatio n of effective bremsstrah lung spec tra and electron co ntami nation for photon dose calc ulation s. Phys. Med. BioI. 34 (1989) 1451- 1464 Ahnesjo A.• M. Saxner, and A. Trepp: A pencil bea m model fo r pho ton dose calculation. Me d. Phys. t 9 (1992) 263-273 Ah nesj o A.: Collapsed co ne convolutio n of radiant energy for photon dose ca lculatio n in heterogeneous media. Med . Phys. 16 ( t 989) 577-592 Ahnesjo A.. and A. Trepp: Acqu isition of the effec tive lateral energ y fluence distributio n for photo n beam dose ca lculations by convo lution mode ls. Phys. Med . BioI. 36 ( 199 1) 973-985 Andreo P.: Uncertainties in dosimetric datu and beam ca libration . tnr, J. Rad . One. BioI. Phys. 19 (1990) 1233 -1247 Boyer A.• and E.C. Mo k: A photon distribution model employing convolution calculat ions. Med. Phys. 12 (198 5) 169-177 Boyer A.L.. and E.C. Mok : Calculation of photon dose distri bu tions in an inhomogeneous medium using co nvolutions . Me d. Phys . 13 ( 1986) 503 -509
Erfahrener
Correspond ing address: Dr. Anders Ahn csjo, Helax AB, Klostergatan 10, S-751 47 Up psala , Sweden
MEVAPLAN, SIDOS, EVADOS, SOMATOM
MEDIZIN·PHYSIKER fur Strah/entherapeutisches Zentrum Hamburg gesucht ! Obertariflich e Bezahlung ! Ggf. Wo hnungsbeschaffung !
[9 ] Brah me A.: Accuracy requi reme nts and q uality ass urance of ex ternal bea m therapy with pho tons and electrons. Acta Oncol., (Ed . 1988) Sup pl. I ( 10) Jaffray D.A.• J. J. Battista. A. Fenster and P. Munro: X-ray so urces of mcdical linear acce lerato rs: Focal an d ex tra-focal radia tion . Me d. Phys. 20 (1993) 14 17- 1427 [I I I Treuer II .. R. Boesecke . W. Schle gel. G.H . Hartm ann. R.P. Miill er and V. Sturm : The source-density fu nctio n: determi natio n from meas ured lateral dose dis tributions and use for co nvo lution dosimetry. Phys. Med . Biol. 37 ( 1992) 1895-1909. (12) Wong. J.W.• and J.A . Purd y: O n method s of inhomogeneity correctio ns fo r photon transport . Med . Phys. 17 (1990) 807-814
in
Bestrahlungsp lanungs·Systeme, CT, DEC-Computer Komp lettsyste me, Ersatzteile , Servicevertraqe , Leasing
SELECTRON, 6 Kanale, 36 Quellen, 40 mCi, neuwertig
Zuschriften unter Chiffre MP 01 an G. Braun Fachverlage, Postf. 17 09, 760J6 Karlsruhe
Ingenieurburo MICHAEL SONNTAG 91054 Erlangen, Spardorfer Str. 76 Tel. 0 91 31·20 54 50, Fax·20 54 51
Medizini sche Universitat zu Lubeck An der Klinik fiir Strahlent herapie und Nuklearmedizin der Medizinisehen Univcrsitat zu Lubeck (D ircktor: Prof. Dr. med. E. Richter) ist ab I. 6. 1994 eine Stelle einer/eines
Medizinphysikerin/Medizinphysikers zu besetzen. Die geratetechn ische Ausstattung der Abteilung Strahle ntherapie umfaBt zwei Linearbeschleuniger mit vem etztem Vertifikations- und Protokolliersystern, Simulator. HDR·Afterloading·Einrichtung. Therapic planungssystcm, Rontgen·Obertlachen.Therapie·Einrichtung und Hypertherrnie, Eine mechanische und eine elektronische Werkstatt steht zur Verfugung, Aufgabengebiet: Bestahlungsplanung, Dosimetric, Betriebssicherheit der Gerate, Qualitatskontrollen, Wahmehm ung der Pfllchten cines Strahlcnschutzbcauftraglen fur den physikalisch-technisc hcn Bereich, Pflegc dcr angcwcndctcn Bestrahlungstcchniken und -methoden und Entwicklung neuer Techniken, Melhod en und Hilfsmittel. Anfordcrungen: Fachkunde im Strahlenschutz gemaB .Ric htlinien Strahlenschutz in der Medizin", Anlage 2. Abschnitt 2.1, und nach der Rontgcnverordnung (Fach kunderichtlinie Technik). Kenntn isse der englischcn Sprachc, Berufscrfahrung in den im Aufgabengebiet genannten Tatigkeiten, EDV-Kenntnisse. Bereitschaft zur Zusammenarbeit. Die Berechtigung zum Filhren der Berufsbezeichnung . Medizinphysikerlin" oder die Fachanerkennung in Medizinischer Physik dUTCh die DGMP sind erwilnscht. jedoch nicht Voraussetzung. Auskiinfte erteilt Herr Dr. E. Ihnen. Tel.: (04 51) 5 00 20 66. Die Medizinische Universitiit zu Liibeck ist bestrebt. den Anteil dcr weiblichcn Beschiiftigten in herausgehobencn Positionen anzuheben und fordert deshalb entsprechend qualifizierte Frauen auf, sich zu bcwcrben. Schwerbchindcrten BewerbcrinncnlBewerbem wird bei entsprechender Eignung der Vorzug gegeben. Bewerbungen mit den iiblichen Unte rlagen (Lebensla uf. Zcugnis, ggf. Schriftenverzeichnis) werden innerhalb von 2 Wochen nach Erscheinen dieser Anzeige erbete n.
An den Verwaltungsdirektor der Med. Universltat Lubeck, Personalwesen 310/9 Ratzeburger Allee 160, 2353 8 Lubeck
Physical Models for 3D-Radiotherapy Planning with Photon Beams Anders Ahnesjo Helax AB, Research Department, Uppsala
Abstract Dose calculation methods f or 3D treatment plannin g must be general enough to consider the degrees of fr eedom available for treatment delivery, without loosing accuracy in the final result . This is the motivation f or the development of enery deposition kernel models which will be reviewed here. Zus ammenfassung Dosisberechnungsmethoden fiir 3D-Bestrahlungsplanung miissen allgemein genug sein, urn die Freiheitsgrade der Behandlungsdu rchfiihrung zu beriicksichtigen, ohne jedoch im Enderg ebnis an Genauigkeit zu verlieren. Dies ist die Motivation fii r die Entwicklung von Modellen fiir Energ ieiibertragungskerne, welche hier behandelt werden. Key wor ds : Treatment Planning, Physical Models, Energy Deposition Kernels
The Task of Dose Calculat ion In the present development of 3D radioth erapy there is a will to exploit all degrees of freedom provided by the treatment machines such as dynamic multileaf etc. Hence, the basic design criterion of dose calculation algorithm s is to calculate dose, within clinic ally acceptabl e time limits, for the necessary treatment situations with an accurac y that does not significantly degrade the overall treatment accurac y. The main dosimetric uncertain ties, other than from treatment plann ing, have been estimated [6, 9] to 4.3% with a potential to achieve 2.5% with future development. Hence the ultimate goal for dose calculation algorithms is to achie ve an error of less than I%. At the present state of the art, an error of 2% - 3% does not lower the overall treatment precision significantly, while a 5% error makes dose calcul ations the bottle neck of accuracy.
Energy Deposition Kernels The basic physics for therapeutic photon beams is well known and easy to characterize by use of microscopic
102
cross sections for each type of interaction. The number of interactions required before the energy of a photon entering a patient is completely absorbed is limited, but analytical photon transport methods for dose computation fail due to the complexity of the integrals involved . The only exception is the transport of unscattered, primary photons where simple attenuation and projection laws apply. Hence, Monte Carlo simulation is the only available method that can yield macroscopic dose distributions from the microscopic description using interaction cross sections. To achie ve acceptab le accuracy for dose calculations, however, the use of Monte Carlo require s the simulation of at least 108 photons. Despite the rapid development of computer hardware this is far to time consuming for clinical use. Convolution/superpo sition methods using energy deposition kernels provide a compromise for combining analytical calculations of the primary photon transport with Monte Carlo transport of secondary particles. Calculat ion speed is achieved by doing the Monte Carlo simulation in advance for simplified geometries and reusing the result as convolution kernels. Some accuracy is thereby traded since approximate sea-
ling laws must be applied when using such kernels. The kernel for energy deposition around a primary photon interaction site is commonl y referred to as a photon point spread function (see cover picture of this issuer.Other usable kernels are the pencil beam and planar spread function, cf. fig. I.
Dose Calculation using Point Spread Functions Kernel summa tion techniques It is, of course, impossible to calculate an "exact" kernel for all possible beam configuration s and tissue distributions. The common approach is to scale all dose fractions of a point spread function calculated for a homogeneous medium by the mean density between the point of energy release and deposition. This scaling is consistent with the theorems of O'Connor and Fano. Density scaling of a discrete kernel is equivalent to rebinning a homogeneou s kernel into bins with extensions according to the current inhomogenities. This is trivial for kerne l regions of low gradients but must be done under energy conservation where gradient s are greater to avoid artefact s. The effect
Z. Med, Phys. 4 (1994) 102 - 104
TAGUNGSBERI CHTE
of high kernel gradients is probably the main reason for differences in results from apparently equivalent methods as reported by Wong and Purdy [12]. Several methods can be used for summation of density scaled kernels in dose calculations. These methods, however, are extremely time consuming, so that the search for alternatives has been intensive. In homogeneous media one can use transform based convolution algorithms which yield drastically reduced computation times [7]. In heterogeneous media , however, a full density scaling cannot be combined with transform convo lutions, unless very simp lifying assumptions are made. Boyer and Mok [81 proposed a method that sca led the scatter do. e kernel by the density at the scatter site only. The method has been improved [I] to include the densities at both the scattering and the receiving sites for scaling of the scatter kernel. However, this concept fails to scale the primary dose due to high gradients of the primary dose kernel.
Collapsed cone convolution Polyenergetic point spread functions for treatment machine spectra can be accurately parameterized by a sum of two exponentials over the squa re of the radius. This enable s a semi-analytical approach as used in the colla psed cone method [4]. In this model the point spread function is discretized into a finite number of conical sectors. The energy transport within each cone is approximated to take place on the symmetry axis of the cone, i.e. the cone is "collapsed" onto its axis. A lattice of cone axes is constructed throughout the irradiated volume such that one axis from each direction intersects at each calculation point. The kernel summation is then done by analyt ical convol utions in disc rete steps along each axis using recursive formulas . The recursions reduce the total number of operations needed to a value proportional to N3 for calculation of the dose to 3 point .
Point spread function
Figure I
Pencil beam kernel
Planar spread function
Irradiation geometries for different types of energy deposition kernels.
Dose Calculation using Pencil Beam Kernels The use of pencil beam kernels instead of point spread functions can be viewed as a "preconvolution" over the depth dimension to save computation time. Runtime convolution of the kernel over the beam aperture is then done assuming a flat beam entrance surface. The influence of patient curvature is taken into acco unt by using kernel values for the depth at the calculation point, as beam library models scale measured dose distributions due to patient curvature. Hence, the pencil beam method shares some of the limitations of the older, empirical methods. The main adva ntages of the pencil beam method are that it is generalized for irregular fields, handles latera l fluence varia tions, calculates the dose in energy fluence normalized unit that can be used to derive output factors , and is based on the same set of data as a full 3D convolution model. This type of model com bines naturally with ID heterogeneity corrections that apply the forward part of a planar spread kernel along the beam path to each calculation point. Such simp le methods are surpris ingly accurate provided lateral charge d particle equilibrium is maintained (3).
Beam Data Acquisition for Kernel Based Methods A common approximation is to assume the spectrum to be laterally invariant and handle the lateral variation
of the energy fluence separa tely, i.e. the energy fluence distribution is factorized into a spectrum and a lateral distri bution:
To be clinically practical, the characterization of a therapy machine must be based on easily obtainable data such as depth dose curves and profiles measured in a water phantom. Such a procedure could be ummarized as follows: I. Determination of an effective spectrum WE and charged particle contamination from depth dose data. It is hereby necessary to use constraints while minimizing the difference between measured and reconstructed depth dose curves [2]. Once acqui red, the spectrum is used for calc ulation of kernels and attenuation coefficients from databases for monoenerget ic photons . 2. Characterization of the lateral energy fluence distribution W(x, y) by deconvolution of a measured dose distribution for the largest possib le field [5, II]. The fluence distribu tion for the larges t open field is the base on which beam collimation and modulation (wedges , filter, etc.) is applied. 3. Mode lling dose from contaminating particles [2, 3, 10]. All irradiate d parts of the treatmen t head act as diffuse irradiators of both scattered photons and charged particle s. Separate models are required for calculating the contribution from contaminating particles as they are not mod-
103
TAGUNGSBERICHTE
elled thro ugh co nvo lution of the basic kerne ls for the primary beam. It is espe cially important to consi der the photons scattered in the treatm ent head not only when mode lling output factors, but also for profiles in dyna mically modulated fields.
[2 /
[3)
[4)
Conclusion The main adva ntages of kernel based models. com pared to conventional beam library models. are their greater ge nerality and accuracy for the higher degree of freedom involve d in 3D radiotherapy planning.
References
[5 ]
16)
[71 [8 )
[I ) Ahnesjo A.: lnvariance of convolution kernels applied to dos e calculat ions for pho -
ton beams. Proc 9th Int. Con f. Compo Rad. Ther.• IX ICeR (1987) 99-102 Ahnesjo A.. and And reo P.: Determinatio n of effective bremsstrah lung spec tra and electron co ntami nation for photon dose calc ulation s. Phys. Med. BioI. 34 (1989) 1451- 1464 Ahnesjo A.• M. Saxner, and A. Trepp: A pencil bea m model fo r pho ton dose calculation. Me d. Phys. t 9 (1992) 263-273 Ah nesj o A.: Collapsed co ne convolutio n of radiant energy for photon dose ca lculatio n in heterogeneous media. Med . Phys. 16 ( t 989) 577-592 Ahnesjo A.. and A. Trepp: Acqu isition of the effec tive lateral energ y fluence distributio n for photo n beam dose ca lculations by convo lution mode ls. Phys. Med . BioI. 36 ( 199 1) 973-985 Andreo P.: Uncertainties in dosimetric datu and beam ca libration . tnr, J. Rad . One. BioI. Phys. 19 (1990) 1233 -1247 Boyer A.• and E.C. Mo k: A photon distribution model employing convolution calculat ions. Med. Phys. 12 (198 5) 169-177 Boyer A.L.. and E.C. Mok : Calculation of photon dose distri bu tions in an inhomogeneous medium using co nvolutions . Me d. Phys . 13 ( 1986) 503 -509
Erfahrener
Correspond ing address: Dr. Anders Ahn csjo, Helax AB, Klostergatan 10, S-751 47 Up psala , Sweden
MEVAPLAN, SIDOS, EVADOS, SOMATOM
MEDIZIN·PHYSIKER fur Strah/entherapeutisches Zentrum Hamburg gesucht ! Obertariflich e Bezahlung ! Ggf. Wo hnungsbeschaffung !
[9 ] Brah me A.: Accuracy requi reme nts and q uality ass urance of ex ternal bea m therapy with pho tons and electrons. Acta Oncol., (Ed . 1988) Sup pl. I ( 10) Jaffray D.A.• J. J. Battista. A. Fenster and P. Munro: X-ray so urces of mcdical linear acce lerato rs: Focal an d ex tra-focal radia tion . Me d. Phys. 20 (1993) 14 17- 1427 [I I I Treuer II .. R. Boesecke . W. Schle gel. G.H . Hartm ann. R.P. Miill er and V. Sturm : The source-density fu nctio n: determi natio n from meas ured lateral dose dis tributions and use for co nvo lution dosimetry. Phys. Med . Biol. 37 ( 1992) 1895-1909. (12) Wong. J.W.• and J.A . Purd y: O n method s of inhomogeneity correctio ns fo r photon transport . Med . Phys. 17 (1990) 807-814
in
Bestrahlungsp lanungs·Systeme, CT, DEC-Computer Komp lettsyste me, Ersatzteile , Servicevertraqe , Leasing
SELECTRON, 6 Kanale, 36 Quellen, 40 mCi, neuwertig
Zuschriften unter Chiffre MP 01 an G. Braun Fachverlage, Postf. 17 09, 760J6 Karlsruhe
Ingenieurburo MICHAEL SONNTAG 91054 Erlangen, Spardorfer Str. 76 Tel. 0 91 31·20 54 50, Fax·20 54 51
Medizini sche Universitat zu Lubeck An der Klinik fiir Strahlent herapie und Nuklearmedizin der Medizinisehen Univcrsitat zu Lubeck (D ircktor: Prof. Dr. med. E. Richter) ist ab I. 6. 1994 eine Stelle einer/eines
Medizinphysikerin/Medizinphysikers zu besetzen. Die geratetechn ische Ausstattung der Abteilung Strahle ntherapie umfaBt zwei Linearbeschleuniger mit vem etztem Vertifikations- und Protokolliersystern, Simulator. HDR·Afterloading·Einrichtung. Therapic planungssystcm, Rontgen·Obertlachen.Therapie·Einrichtung und Hypertherrnie, Eine mechanische und eine elektronische Werkstatt steht zur Verfugung, Aufgabengebiet: Bestahlungsplanung, Dosimetric, Betriebssicherheit der Gerate, Qualitatskontrollen, Wahmehm ung der Pfllchten cines Strahlcnschutzbcauftraglen fur den physikalisch-technisc hcn Bereich, Pflegc dcr angcwcndctcn Bestrahlungstcchniken und -methoden und Entwicklung neuer Techniken, Melhod en und Hilfsmittel. Anfordcrungen: Fachkunde im Strahlenschutz gemaB .Ric htlinien Strahlenschutz in der Medizin", Anlage 2. Abschnitt 2.1, und nach der Rontgcnverordnung (Fach kunderichtlinie Technik). Kenntn isse der englischcn Sprachc, Berufscrfahrung in den im Aufgabengebiet genannten Tatigkeiten, EDV-Kenntnisse. Bereitschaft zur Zusammenarbeit. Die Berechtigung zum Filhren der Berufsbezeichnung . Medizinphysikerlin" oder die Fachanerkennung in Medizinischer Physik dUTCh die DGMP sind erwilnscht. jedoch nicht Voraussetzung. Auskiinfte erteilt Herr Dr. E. Ihnen. Tel.: (04 51) 5 00 20 66. Die Medizinische Universitiit zu Liibeck ist bestrebt. den Anteil dcr weiblichcn Beschiiftigten in herausgehobencn Positionen anzuheben und fordert deshalb entsprechend qualifizierte Frauen auf, sich zu bcwcrben. Schwerbchindcrten BewerbcrinncnlBewerbem wird bei entsprechender Eignung der Vorzug gegeben. Bewerbungen mit den iiblichen Unte rlagen (Lebensla uf. Zcugnis, ggf. Schriftenverzeichnis) werden innerhalb von 2 Wochen nach Erscheinen dieser Anzeige erbete n.
An den Verwaltungsdirektor der Med. Universltat Lubeck, Personalwesen 310/9 Ratzeburger Allee 160, 2353 8 Lubeck