Bernhard Kroupa

Interpretation and evaluation of the γ index and the γ index angle for the verification of IMRT hybrid plans

In IMRT, the method for a quantitative comparison of two-dimensional dose distributions is still under development. The γ evaluation method proposed by Low et al is the most accepted approach and has been adapted by many groups. Based on the concept of Low et al we developed a software tool with an intelligent search algorithm to minimize the calculation time. For the interpretation of deviations a γ angle distribution and other tools (dose difference map, profiles, γ area histograms, etc) are integrated in the software package. Ten hybrid plans are included in the verification study containing 6 IMRT head and neck cases, 2 IMRT prostate cases and one IMRT paravertebral case as well as a standard uniform intensity conformal 4 field box treatment for comparison. IMRT plans are realized with a segmental MLC delivery technique. The fields of a hybrid plan are applied at once and dose distributions are measured with films in three planes of a verification phantom. All γ vector calculations are based on a 3% dose criterion and a 3 mm DTA acceptance criterion. The mean value γmean (mean value in the γ distribution) of the various IMRT plans is 0.45 ± 0.10 (1 SD). On average, the percentage of points exceeding the acceptance criteria of γ ≤ 1 (γ>1) is 5.8 ± 5.4% (1 SD). The mean value of γ1% (1% of points have an equal or higher γ value) is 1.47 ± 0.59 (1 SD) for IMRT plans. In 5 out of 27 planes, γ>1 is substantially larger than the average. This is also indicated in γ area histograms. Planes with large areas outside the tolerance criteria were further evaluated using γ angle distributions. This additional information indicates that the large areas with high γ values are dominated by the dose difference. It is shown that the deviations are influenced by tongue and groove effects. From the statistical evaluation of γ values (e.g. γ area histogram), acceptance criteria for IMRT hybrid plans can be defined. For the interpretation of the γ maps, distributions of the γ angle and traditional evaluation methods, such as dose profiles, are still very useful.

Flattening filter free beams in SBRT and IMRT: Dosimetric assessment of peripheral doses

PURPOSE Recently, there has been a growing interest in operating medical linear accelerators without a flattening filter. Due to reduced scatter, leaf transmission and radiation head leakage a reduction of out-of-field dose is expected for flattening filter free beams. The aim of the present study was to determine the impact of unflattened beams on peripheral dose for advanced treatment techniques with a large number of MUs. MATERIAL AND METHODS An Elekta Precise linac was modified to provide 6 and 10 MV photon beams without a flattening filter. Basic beam data were collected and implemented into the TPS Oncentra Masterplan (Nucletron). Leakage radiation, which predominantly contributes to peripheral dose at larger distances from the field edge, was measured using a Farmer type ionisation chamber. SBRT (lung) and IMRT (prostate, head&neck) treatment plans were generated for 6 and 10 MV for both flattened and unflattened beams. All treatment plans were delivered to the relevant anatomic region of an anthropomorphic phantom which was extended by a solid water slab phantom. Dosimetric measurements were performed with TLD-700 rods, radiochromic films and a Farmer type ionisation chamber. The detectors were placed within the slab phantom and positioned along the isocentric longitudinal axis. RESULTS Using unflattened beams results in a reduction of treatment head leakage by 52% for 6 and 65% for 10 MV. Thus, peripheral doses were in general smaller for treatment plans calculated with unflattened beams. At about 20 cm distance from the field edge the dose was on average reduced by 23 and 31% for the 6 and 10 MV SBRT plans. For the IMRT plans (10 MV) the average reduction was 16% for the prostate and 18% for the head&neck case, respectively. For all examined cases, the relative deviation between peripheral doses of flattened and unflattened beams was found to increase with increasing distance from the field. CONCLUSIONS Removing the flattening filter lead to reduced peripheral doses for advanced treatment techniques. The relative difference between peripheral doses of flattened and unflattened beams was more pronounced when the nominal beam energy was increased. Patients may benefit by decreased exposure of normal tissue to scattered dose outside the field.

Inverse Planning – a Comparative Intersystem and Interpatient Constraint Study

Purpose:To compare commercial treatment-planning systems (TPS) for inverse planning (IP) and to assess constraint variations for specific IMRT indications.Material and Methods:For IP, OTP, XiO and BrainSCAN were used and step-and-shoot intensity-modulated radiotherapy (IMRT) delivery was assumed. Based on identical constraints, IP was performed for a prostate, head and neck, brain, and gynecologic case. IMRT plans were compared in terms of conformity/homogeneity, dose-volume histograms (DVHs), and delivery efficiency. For ten patients each of a class of indications, constraint variations were evaluated.Results:IMRT plans were comparable concerning minimum target dose, homogeneity, conformity, and maximum doses to organs at risk. Larger differences were seen in dose gradients outside the target, monitor units, and segment number. Using help structures proved efficient to shape isodoses and to reduce segmentation workload. For IMRT class solutions, IP constraint variations depended on anatomic site.Conclusion:IP systems requiring doses as input and having objective functions based on physical parameters had a very similar performance. Constraint templates can be established for a class of IMRT indications.Ziel:Die Funktionalität der inversen Planung (IP) von kommerziellen Bestrahlungsplanungssystemen (TPS) sowie die Unterschiede von Dosiszielgrößen bei typischen IMRT-Indikationen (intensitätsmodulierte Strahlentherapie) wurden untersucht.Material und Methodik:Für die IP fanden OTP, XiO und BrainSCAN Verwendung, und eine „Step-and-shoot“-IMRT wurde angenommen. Basierend auf gleichen Zielgrößenvorgaben wurde für je einen Fall mit Prostatakarzinom, Hirntumor, HNO-Tumor sowie gynäkologischem Tumor eine IP durchgeführt. Die IMRT-Pläne wurden anhand von Dosis-Volumen-Histogrammen (DVHs), Konformität, Homogenität und Bestrahlungseffizienz bewertet. Für je zehn Patienten mit bestimmten Indikationen wurde die Schwankung der IP-Zielgrößen untersucht.Ergebnisse:Mit allen drei TPS konnten ähnliche IMRT-Pläne mit vergleichbarer Zielgebietsauslastung, Dosishomogenität, Konformität und maximaler Dosisbelastung der Risikoorgane erstellt werden. Größere Unterschiede wurden hinsichtlich des Dosisgradienten außerhalb des Zielgebiets, der Monitoreinheiten sowie der Segmentzahlen beobachtet. Die Verwendung von Hilfsstrukturen erwies sich als zeitsparend. Für eine IMRT-Indikation schwanken die Zielgrößenvorgaben in Abhängigkeit von der Patientenanatomie.Schlussfolgerung:Mit TPS, deren Zielgrößen und Zielfunktionen auf physikalischen Dosen beruhen, konnten ähnliche IMRT-Pläne erzielt werden. Für IMRT-Konzepte lassen sich Standardzielvorgaben festlegen.

Normalized sensitometric curves for the verification of hybrid IMRT treatment plans with multiple energies

With the clinical implementation of time-variable dose patterns and intensity modulated radiotherapy (IMRT) film dosimetry has regained popularity. Films are currently the most frequently used dosimetric means for patient specific quality assurance in IMRT. A common method is to verify a so-called hybrid IMRT plan, which is the patient specific treatment plan with unmodified fluence patterns recalculated in a dedicated phantom. For such applications the sensitometric curve, i.e., the relation between optical density (OD) and absorbed dose, should not depend critically on beam energy, field size and depth, or film orientation. In order to minimize the influence of all these variables a normalization of sensitometric curves is performed at various photon beam energies (6 MV, 10 MV, 25 MV). By doing so one unique sensitometric curve can be used for these three beam qualities. This holds for both film types investigated: Kodak X-Omat V films and EDR-2 films. Additionally, the influence of field size, depth and film orientation on a normalized sensitometric curve is determined for both film types. For doses smaller than 0.8 Gy for X-Omat V and doses smaller than 3 Gy for EDR-2 films the field size variation of normalized sensitometric curves is much smaller than 3% for fields up to 20 x 20 cm2. For X-Omat V films all differences between sensitometric curves determined at depths of 5, 10, and 15 cm are smaller than 3%. For EDR-2 films deviations larger than 3% are only observed at low net OD smaller than 0.25. The dependence of film orientation (parallel versus perpendicular) on a normalized sensitometric curve is found to be not critical. However, processing conditions have the largest influence and can result in differences up to 20% for sensitometric curves derived from films of the same batch but using different film processors. When normalizing sensitometric curves to the dose value necessary to obtain a net OD=1 for that respective geometry and energy the large energy dependence of sensitometric curves can be almost eliminated. This becomes especially important for the verification of hybrid IMRT plans with multiple energies. Additionally, such a normalization minimizes other influences such as field size, depth, and film orientation. This method is generally applicable to both Kodak X-Omat V and EDR-2 films. In order to achieve the highest accuracy level an upper dose limit of 0.8 Gy for X-Omat V films and 3 Gy for EDR-2 films should be taken into account. However, these dose limits may vary with film reading instrument and film processor.

Clinical Comparison of Dose Calculation Using the Enhanced Collapsed Cone Algorithm vs. a New Monte Carlo Algorithm

Purpose:Comparison of the dosimetric accuracy of the enhanced collapsed cone (eCC) algorithm with the commercially available Monte Carlo (MC) dose calculation for complex treatment techniques.Material and Methods:A total of 8 intensity-modulated radiotherapy (IMRT) and 2 stereotactic body radiotherapy (SBRT) lung cases were calculated with eCC and MC algorithms with the treatment planning systems (TPS) Oncentra MasterPlan 3.2 (Nucletron) and Monaco 2.01 (Elekta/CMS). Fluence optimization as well as sequencing of IMRT plans was primarily performed using Monaco. Dose prediction errors were calculated using MC as reference. The dose–volume histrogram (DVH) analysis was complemented with 2D and 3D gamma evaluation. Both algorithms were compared to measurements using the Delta4 system (Scandidos).Results:Recalculated with eCC IMRT plans resulted in lower planned target volume (PTV) coverage, as well as in lower organs-at-risk (OAR) doses up to 8%. Small deviations between MC and eCC in PTV dose (1–2%) were detected for IMRT cases, while larger deviations were observed for SBRT (up to 5%). Conformity indices of both calculations were similar; however, the homogeneity of the eCC calculated plans was slightly better. Delta4 measurements confirmed high dosimetric accuracy of both TPS.Conclusion:Mean dose prediction errors < 3% for PTV suggest that both algorithms enable highly accurate dose calculations under clinical conditions. However, users should be aware of slightly underestimated OAR doses using the eCC algorithm.Ziel:Vergleich der Dosisberechnungsgenauigkeit eines erweiterten Collapsed-Cone-(eCC-)Algorithmus mit einem kommerziell verfügbaren Monte-Carlo-(MC-)Algorithmus anhand von komplexen Bestrahlungstechniken.Material und Methodik: 8 IMRT- und 2 SBRT-Pläne wurden jeweils mit dem eCC-Algorithmus des Bestrahlungsplanungssystems Oncentra 3.2 (Nucletron) und mit dem Monte-Carlo-Algorithmus von Monaco 2.01 (Elekta/CMS) berechnet. Die Fluenz-Optimierung und Sequenzierung der IMRT-Pläne wurden mittels Monaco durchgeführt. Zusätzlich zur DVH-Analyse wurden eine 2D und 3D Gamma-Index-Evaluierung durchgeführt. Darüber hinaus erfolgte eine dosimetrische Überprüfung der Bestrahlungspläne mit dem Delta4-System (Scandidos).Ergebnisse:Im Vergleich zum MC-Algorithmus resultierte die Nachberechnung der IMRT-Pläne mit dem eCC in einer geringeren Abdeckung der Zielvolumina sowie in einer um bis zu 8% niedrigeren Dosis für Risikoorgane (OAR). Bei den IMRT-Fällen waren die Abweichungen zwischen den PTV-Dosen der beiden Algorithmen gering (1–2%), wohingegen die größten Abweichungen (bis zu 5%) für SBRT festgestellt wurden. Die ermittelten Konformitätsindizes waren für beide Algorithmen sehr ähnlich; die Homogenität war für eCC-Pläne geringfügig besser. Die Delta4-Messungen bestätigten die hohe Dosisberechnungsgenauigkeit beider Algorithmen.Schlussfolgerung:Die mittlere Unsicherheit der berechneten PTV-Dosen betrug weniger als 3%, d.h. mit beiden Algorithmen kann eine zuverlässige Dosisvorhersage unter klinischen Bedingungen erzielt werden. Bei der Beurteilung von mit eCC berechneten Plänen sollte die geringfügige Unterschätzung der Dosis in OAR berücksichtigt werden.

Pre-clinical evaluation of an inverse planning module for segmental MLC based IMRT delivery

Phantom tests are performed for pre-clinical evaluation of a commercial inverse planning system (HELAX TMS, V 6.0) for segmented multileaf collimator (MLC) intensity modulated radiotherapy (IMRT) delivery. The optimization module has available two optimization algorithms: the target primary feasibility and the weighted feasibility algorithm, only the latter allows the user to specify weights for structures. In the first series, single beam tests are performed to evaluate the outcome of inverse planning in terms of plausibility for the following situations: oblique incidence, presence of inhomogeneities, multiple targets at different depths and multiple targets with different desired doses. Additionally, for these tests a manual plan is made for comparison. In the absence of organs at risk, both the optimization algorithms are found to assign the highest priority to low dose constraints for targets. In the second series, tests resembling clinical relevant configurations (simultaneous boost and concave target with critical organ) are performed with multiple beam arrangements in order to determine the impact of the systems configuration on inverse planning. It is found that the definition of certain segment number and segment size limitations does not largely compromise treatment plans when using multiple beams. On the other hand, these limitations are important for delivery efficiency and dosimetry. For the number of iterations and voxels per volume of interest, standard values in the systems configuration are considered to be sufficient. Additionally, it is demonstrated that precautions must be taken to precisely define treatment goals when using computerized treatment optimization. Similar phantom tests could be used for a direct dosimetric verification of all steps from inverse treatment planning to IMRT delivery.

Analysis of the dose calculation accuracy for IMRT in lung: A 2D approach

The purpose of this study was to compare the dosimetric accuracy of IMRT plans for targets in lung with the accuracy of standard uniform-intensity conformal radiotherapy for different dose calculation algorithms. Tests were performed utilizing a special phantom manufactured from cork and polystyrene in order to quantify the uncertainty of two commercial TPS for IMRT in the lung. Ionization and film measurements were performed at various measuring points/planes. Additionally, single-beam and uniform-intensity multiple-beam tests were performed, in order to investigate deviations due to other characteristics of IMRT. Helax-TMS V6.1(A) was tested for 6, 10 and 25 MV and BrainSCAN 5.2 for 6 MV photon beams, respectively. Pencil beam (PB) with simple inhomogeneity correction and ‘collapsed cone’ (CC) algorithms were applied for dose calculations. However, the latter was not incorporated during optimization hence only post-optimization recalculation was tested. Two-dimensional dose distributions were evaluated applying the γ index concept. Conformal plans showed the same accuracy as IMRT plans. Ionization chamber measurements detected deviations of up to 5% when a PB algorithm was used for IMRT dose calculations. Significant improvement (deviations ∼2%) was observed when IMRT plans were recalculated with the CC algorithm, especially for the highest nominal energy. All γ evaluations confirmed substantial improvement with the CC algorithm in 2D. While PB dose distributions showed most discrepancies in lower (<50%) and high (>90%) dose regions, the CC dose distributions deviated mainly in the high dose gradient (20–80%) region. The advantages of IMRT (conformity, intra-target dose control) should be counterbalanced with possible calculation inaccuracies for targets in the lung. Until no superior dose calculation algorithms are involved in the iterative optimization process it should be used with great care. When only PB algorithm with simple inhomogeneity correction is used, lower energy photon beams should be utilized.

Comparison of Monte Carlo Dose Calculation with Advanced Kernel Methods in External Photon Beam Treatment Planning

After successful implementation in commercial treatment planning systems (TPS) for high energy electron beams, Monte Carlo dose calculation algorithms are becoming also commercially available for high energy photon beams. On the other hand, advanced kernel based methods are in clinical use for many years. The aim of this study was to compare the accuracy of both types of dose calculation in phantom studies and clinical situations. The treatment planning systems investigated were Oncentra Masterplan (Nucletron, V3.0) with a collapsed cone (CC) convolution algorithm, Eclipse (Varian, V8.1), iPlan (BrainLAB, V4.0.0) and Monaco (CMS, V1.0.2). The latter two TPS offer XVMC based Monte Carlo (MC) dose calculation. Commissioning of the first three systems was performed for an ELEKTA Synergy platform providing 6, 10 and 18MV beams. The Eclipse system was commissioned for a Varian Clinac with 6 and 15 MV. Measurements were performed with a calibrated ionization chamber and radiochromic EBT type films. Single field test cases and IMRT treatments were delivered to an inhomogeneous phantom where cork was used to simulate lung. In addition, the calculation was compared in patient geometry using 7 different IMRT cases. More specifically, MC generated plans were recalculated with the CC algorithm. Dosimetric comparisons included single points, and 1D as well as 2D gamma evaluations together with the use of independent fluence recalculation software. Absolute dosimetry revealed no large differences between MC and advance kernel dose calculation. 1D γ -evaluation showed that CC and the AAA algorithms were not able to model the penumbra broadening, build up and re-build up region behind the polystyrene plate and target as correctly as the MC algorithms. The dose calculation accuracy for the IMRT cases in patient geometry was found to be similar for CC and MC. Dose calculation speed was also assessed in the light of future demands for on-line treatment planning.