Asa K Carlsson Tedgren | PubMed Health

PubMed2005

Brachytherapy source characterization for improved dose calculations using primary and scatter dose separation.

In brachytherapy, tissue heterogeneities, source shielding, and finite patient/phantom extensions affect both the primary and scatter dose distributions. The primary dose is, due to the short range of secondary electrons, dependent only on the distribution of material located on the ray line between the source and dose deposition site. The scatter dose depends on both the direct irradiation pattern and the distribution of material in a large volume surrounding the point of interest, i.e., a much larger volume must be included in calculations to integrate many small dose contributions. It is therefore of interest to consider different methods for the primary and the scatter dose calculation to improve calculation accuracy with limited computer resources. The algorithms in present clinical use ignore these effects causing systematic dose errors in brachytherapy treatment planning. In this work we review a primary and scatter dose separation formalism (PSS) for brachytherapy source characterization to support separate calculation of the primary and scatter dose contributions. We show how the resulting source characterization data can be used to drive more accurate dose calculations using collapsed cone superposition for scatter dose calculations. Two types of source characterization data paths are used: a direct Monte Carlo simulation in water phantoms with subsequent parameterization of the results, and an alternative data path built on processing of AAPM TG43 formatted data to provide similar parameter sets. The latter path is motivated of the large amounts of data already existing in the TG43 format. We demonstrate the PSS methods using both data paths for a clinical 192Ir source. Results are shown for two geometries: a finite but homogeneous water phantom, and a half-slab consisting of water and air. The dose distributions are compared to results from full Monte Carlo simulations and we show significant improvement in scatter dose calculations when the collapsed-cone kernel-superposition algorithm is used compared to traditional table based calculations. The PSS source characterization method uses exponential fit functions derived from one-dimensional transport theory to describe both the primary and scatter dose contributions. We present data for the PSS characterization method to different 192Ir, 137Cs, and 60Cs brachytherapy sources. We also show how TG43 formatted data can be derived from our data to serve traditional treatment planning systems, as to enable for a gradual transfer to algorithms that provides improved modeling of heterogeneities in brachytherapy treatment planning.

Algorithms↗

PubMed2003

Accounting for high Z shields in brachytherapy using collapsed cone superposition for scatter dose calculation.

Common clinical brachytherapy treatment planning algorithms perform at best one-dimensional corrections for high Z heterogeneities that will be inaccurate for intermediate energies (60-100 keV). The development of fast methods for a three-dimensional dose calculation to account for high Z materials in this energy range is important, e.g., to fully utilize the potential of patient individualized shields using isotopes such as 241Am and 169Yb. In this work we use the collapsed cone superposition algorithm to calculate the scatter dose contribution around partly lead-shielded point sources at 60, 100, and 350 keV. Methods to scale point kernels for water into kernels for high Z materials are derived. The scaling accounts for scattered photon spectral differences between materials and thus goes beyond the common density scaling approach. Compared to Monte Carlo simulations, the results of our algorithm yield agreements on the unshielded side to within 3% at 350 and 60 keV and to within 7% at 100 keV out to distances of 8 cm from the source. The effect of the shield in the center of the unshielded region is small at 350 keV but significant and occurs at short distances at 100 and 60 keV. At 60 keV, the shield causes a dose reduction of around 10%, 1 cm from the source on the unshielded side. At 100 keV, the reverse effect is seen, the insertion of shields leading to the total dose being increased by about 10% at 1 cm. That one-dimensional algorithms are incapable of predicting these changes shows the importance of accounting for the full three-dimensional geometry in correctly determining the scatter dose contribution.

Algorithms↗