
Medical Physics
Medical Radiation Physics
Current Research Topics
Multidimensional Dose and Activity Patterns in Radiation Medicine
Within recent years radiotherapy has undergone some major and significant changes from static, nearly unmodulated fields towards small and
intensity modulated techniques including the extensive use of dynamic irradiation schemes such as modulated arc techniques allowing the application
of complex dose patterns. Under radiobiological aspects, e.g. the elimination of anoxic conditions in tumour tissues, this technique has sometimes
been named labelled dose painting of a tumor.
Likewise, dosimetry must deal with highly inhomogeneous dose distributions of variable shape, accompanied by strong dose rate differences and
significant spectral changes of the beam, in combination with higher out-of-field doses and increased irradiation times. The research activities
of our group are strongly directed towards the solution of the associated problems with the methods of medical radiation physics. During the last
years, we successfully followed the path of expanding established concepts for the characterization and measurement of dose distributions and introduced
new ones by the development of new detectors and mathematical methods, guided by the concepts of signal theory.
Summarizing the activities that will be described below, we tried to assist in conceiving a new paradigm of dosimetry
which may be called "dose pattern analysis" or shortly "dose imaging".
- Analytical methods in dosimetry
- 1D iterative deconvolution of DAVID signals
- 2D iterative deconvolution of portal images
- Determination of the effective point of measurement
- Correction of volume effect for detectors with finite volume
- Monte Carlo methods
- Entwicklung von Verfahren zur Individualisierung der Dosisbestimmung und Berechnung in der Computertomographie
- Dosimetrie schmaler Photonfelder
Analytical methods in dosimetry
1D iterative deconvolution of DAVID signals
DAVID system, a translucent multiwire ionization chamber placed in the accessory holder of the treatment head below the MLC. Each wire is exactly adjusted along the midline of its associated leaf pair, thereby generating a signal correlated to the leaf pair aperture. Iterative 1D-deconvolution has been implemented to correct for the blurring of the profile of the wire signals across the beam due to the lateral transport of scattered electrons in the air gap of the DAVID chamber. The true photon fluence profile is calculated from the measured signal profile through the deconvolution algorithm, based upon the convolution kernel formed by the lateral wire signal profile when only one leaf pair is opened. Deconvolved lateral fluence profiles are obtained with increased resolution, and errors in MLC positioning are revealed with enhanced sensitivity.2D iterative deconvolution of portal images
Portal imaging has become an integral part of modern radiotherapy techniques such as IMRT and IGRT. It serves to verify the accuracy of day-to-day patient positioning, a prerequisite for treatment success. However, image blurring attributable to different physical and geometrical effects, analysed in this work, impairs the image quality of the portal images, and anatomical structures cannot always be clearly outlined. A 2D iterative deconvolution method was developed to reduce this image blurring. The affiliated data basis was generated by the separate measurement of the components contributing to image blurring. Secondary electron transport and pixel size within the EPID, as well as geometrical penumbra due to the finite photon source size were found to be the major contributors, whereas photon scattering in the patient is less important. The underlying line-spread kernels of these components were shown to be Lorentz functions. This implies that each of these convolution kernels and also their combination can be characterized by a single characteristic, the width parameter λ of the Lorentz function. Portal images were deconvolved using the point-spread function derived from the Lorentz function together with the experimentally determined λ values. The improvement of the portal images was quantified in terms of the modulation transfer function of a bar pattern. The resulting clinical images show a clear enhancement of sharpness and contrast.Determination of the effective point of measurement
The subject of this study is the "shift of the effective point of measurement", Δz, well-known as a method of correction compensating for the "displacement effect" in photon and electron beam dosimetry. Radiochromic EBT 1 films have been used to measure the "true" TPR curves of 6 and 15 MV photons and 6 and 9 MeV electrons in the solid water-equivalent material RW3. The positions of the effective points of measurement have been determined for the Roos and Markus chambers, the cylindrical "PinPoint", "Semiflex" and "Rigid-Stem" chambers, the 2D-Array and the E-type silicon diode (all from PTW-Freiburg) by direct or indirect comparison between their TPR curves and those of the EBT 1 film. According to a theoretical consideration, the shift of the effective point of measurement from the reference point of the detector is caused by a gradient of the fluence of the ionizing particles. Its value depends on the construction of the detector, but remains invariant under changes of radiation quality and depth. Other disturbances, which do not belong to the class of "gradient effects", are not corrected by shifting the effective point of measurement.Correction of volume effect for detectors with finite volume
The measured dose profiles using a detector with a finite sensitive volume is a result of the convolution of the undisturbed true dose profile with the detector's response function. The lateral and longitudinal detector's response function can be approximated by the Gauss function and hence are characterised by the parameters σ_lat und σ_long. The volume effect will cause widening of the measured dose profile and reduction of the dose at the central axis for small fields. These effect can be minimized by choosing a small detector with negligible volume effect. Otherwise, the measured dose profile has to be corrected by deconvolution of the measured signal with the detector's response function. The parameters σ_lat und σ_long has been determined for a number of detectors by comparison of the measured dose profile to the diode's profile, where the later is convolved with a Gauss function. The magnitude of the standard deviation of the Gauss function that produce the best fit between the convolved diode's profile and the measured signal are chosen as the σ_lat und σ_long. The measured signal at the central axis of small fields can be corrected using the correction factor kV = kV,lat . kV,long. kV is derived using the detector's response function weighted mean value of the ideal dose profile around the central axis, which is approximated by the Taylor's series terminated after the quadratic term. back to topMonte Carlo methods
Monte Carlo calculations have found a wide range of applications in the Medical Physics domain and are a major work-tool in our research group. We have implemented this method in numerous situations to assess quantities of interest such as in dose and spectral computations. The EGSnrc Monte Carlo package has been widely utilised and currently, the FLUKA code system is on trial in our group.
Linear Accelerator Modelling
Using BEAMnrc/EGSnrc, models of the 6/15 MV dual-energy SIEMENS Primus linear accelerator (linac) have been constructed using manufacturer data and results from other researchers. The linac models have been bench-marked by comparing in-phantom percent depth dose profiles and transverse dose profiles measured at the reference treatment unit located at the department of Radiotherapy and Oncology of the Pius-Hospital Oldenburg. Dose computations were performed using DOSXYZnrc in a large voxelized water tank in the same geometry as in the experimental measurement setup. The bench-marking has been extended to out-of-field regions, in areas dominated by high Linear Energy Transfer (LET) radiation, keeping in mind extensive reports from other researchers on secondary cancer induction at areas in the vicinity of the primary treatment site.Linac head reconfigurations have also been performed and has led to the design of a novel beam flattening system, the so-called Direction-Selective Flattening filter system, an intermediary between the classical flattening filter (FF) systems and the flattening filter free (FFF) system. Assessment of the neutron source strength has been performed under 15 MV configurations indirectly by assessing the photon energy fluence profiles at all tungsten surfaces hit by photons, assuming their total absorption and using the macroscopic photoneutron cross section in tungsten.
Spectra Computations
Motivated by the need to provide a new descriptive parameter for the beam quality of Co-60 gamma-ray units and megavoltage linear accelerator units, we have performed data compression based on the dose contribution by low-energy scattered photons until a stated cut-off energy, e.g. 200 keV. Using FLURZnrc/EGSnrc, photon fluence spectra have been computed for different field sizes, both for the Theratron 780C Co-60 unit and for the SIEMENS Primus linac operating under 6 MV and 15 MV nominal photon energies using the calculated photon spectra and photon attenuation data libraries of the National Institute of Standards and Technology (NIST).Dose at high-Z interfaces
If a patient, who receives external beam treatment, has an artificial joint, this prosthesis has great influence on the dose distribution of the beam. Artificial joints are usually made of high-Z materials like titanium. To analyse the effects of high-Z inhomogeneities within human tissue during radiation therapy, we constructed a model of titanium slabs in a water phantom and calculated the depth dose distribution for different energy spectra from 4MV to 24MV using EGS-Ray and EGSnrc. back to topEntwicklung von Verfahren zur Individualisierung der Dosisbestimmung und Berechnung in der Computertomographie
Der kontinuierlich steigende Anteil der Computertomographie in der diagnostischen Bildgebung erfordert eine genaue Quantifizierung und Optimierung der Dosis beim CT. Die Berücksichtigung der individuellen Strahlenexposition gewinnt daher auch in der Radiologie immer mehr an Bedeutung. Für die genaue Quantifizierung der Strahlenexposition und speziell für die Bestimmung der effektiven Dosis einer CT-Untersuchung ist eine möglichst exakte Bestimmung der einzelnen Organdosen Voraussetzung. Da eine direkte Messung der Organdosen im Patienten nicht möglich ist, werden sie meist aus messbaren Größen und Konversionsfaktoren berechnet, die z. B. aus Monte Carlo Simulationen an Voxelmodellen des Menschen, die am Helmholtz Zentrum München entwickelt wurden, bestimmt werden.Im Rahmen des Teilprojektes sollen Verfahren zur Berechnung der Dosisverteilung im menschlichen Körper erstellt werden, die zu einer Individualisierung der Strahlenexposition beim CT beitragen. Da bei Patienten individuell durchaus größere Unterschiede zu den verwendeten Voxelmodellen des Menschen auftreten können, soll für spezielle Organe und Untersuchungen geprüft werden, inwieweit sich eine Individualisierung bzw. Parametrisierung verschiedener relevanter Parameter auf die Organdosisbestimmung auswirken würde. Im Wesentlichen werden dabei zwei Ansätze zur Berechnung der Dosisverteilung im Körper und zur Organdosisbestimmung verfolgt, wobei Verfahren aus der Strahlentherapie (Organkonturierung, Organdosisberechnung etc.) adaptiert werden.
Monte Carlo Simulation direkt an den CT-Daten des Patienten sind für die individuelle Dosisbestimmung in der Computertomographie eine geeignete Möglichkeit. Allerdings sind sie nicht nur wegen der erhöhten Rechenzeit, sondern vor allem wegen der durchzuführenden Segmentierung der Organe sehr zeitaufwendig. Zudem ist eine Anpassung exakter MC-Modelle auf die einzelnen Scanner bei der Vielzahl der verfügbaren Gerätetypen zurzeit nicht denkbar. Deshalb soll der Einfluss individueller Patientenparameter auf die Strahlenexposition systematisch untersucht und entsprechende Modelle, die z. B. die Organdosisbestimmung mittels Konversionsfaktoren ergänzen, entwickelt werden. Dazu sollen sowohl Monte Carlo Simulationen als auch Messungen an dosimetrischen Phantomen durchgeführt werden.
Der zweite Ansatz verfolgt die Entwicklung analytischer Modelle zur Berechnung von Dosisverteilungen in der Computertomographie. Dabei sollen analytische Modelle aus der Strahlentherapie übertragen werden auf die Röntgendiagnostik. Ziel ist es Dosisverteilungen einzelner CT-Scans mittels analytischer Funktionen auf Phantomen zu beschreiben. In einem weiteren Schritt wäre dann die Anpassung der Berechnung auf den menschlichen Körper denkbar. back to top