Why the scatter-to-primary ratio of photon beams in water follows $\mathrm{SPR}=aZ+bZ^{2}$ with $Z=rd/(r+d)$: deriving the Bjärngard-Petti (1988) empirical law

In external-beam radiotherapy dosimetry, the dose at a point in a water phantom is conventionally split into a primary component (photons arriving without interacting in the phantom) and a scatter component. Bjärngard and Petti (Phys. Med. Biol. 33, 21-32, 1988) computed the kerma scatter-to-primary ratio (SPR) by Monte Carlo for water irradiated by 200 keV monoenergetic photons, 60Co gamma rays, and 4 and 15 MV x-rays, and found that on the central axis of a circular field of radius r at depth d it obeys the compact empirical law $\mathrm{SPR}=aZ+bZ^{2}$, where $Z=rd/(r+d)$ is a combined geometric variable; the coefficient $a$ is determined by the spectrum-averaged linear attenuation coefficient $\bar{\mu}$ and the ratio $r/d$, and the coefficient $b$ depends on $r/d$ and on energy, the energy dependence being approximately proportional to $\bar{\mu}^{3}$. The paper itself states the relation is "not fully explained": only a qualitative first-order-scatter argument for $\mathrm{SPR}\simeq aZ$ was given, its Monte Carlo order separation shows both $a$ and $b$ contain first-order and multiply scattered contributions, and its extraction of scatter from measured 60Co tissue-air-ratio (TAR) tables was limited by the normalization convention of the zero-field quantity $\mathrm{TAR}(0,d)$. Later work adopted and re-fit the law (Nizin; Iwasaki; McDonough-Bjärngard; Wang-Zhu) and substantially reworked the TAR normalization question (McKenzie; Nizin; Rice-Chin), but no first-principles derivation of the law, its coefficient scalings, or its scattering-order decomposition has been located. The open problem is to give this empirical law a physical foundation.

Background

In external-beam radiation therapy, tumors are treated with high-energy photon beams: gamma rays from cobalt-60 radioactive sources or bremsstrahlung x-rays of a few megavolts (MV) from linear accelerators. When such a beam enters a water phantom (the standard stand-in for soft tissue), the absorbed dose at any point splits into a primary component, deposited by photons that reach the neighbourhood of that point without interacting in the phantom, and a scatter component, deposited by photons deflected (mainly by Compton scattering off electrons) one or more times within the phantom. Separating these components is a recurring task of clinical dosimetry, because beam data such as depth-dose curves and output factors must be interpolated, extrapolated and quality-checked, and the scatter part is what makes them deviate from simple exponential attenuation.

A convenient measure of the scatter burden at a point is the scatter-to-primary ratio (SPR), the ratio of the scattered to the primary contribution there; since both components scale with the photon fluence, SPR is insensitive to absolute calibration. Bjärngard and Petti (1988) studied the SPR of kerma (the kinetic energy released to charged particles per unit mass, which equals absorbed dose where charged particles are in equilibrium) on the central axis of circular fields. Using a Monte Carlo code that scored photons scattered exactly once and more than once separately, for 200 keV monoenergetic photons, 60Co gamma rays, and 4 and 15 MV x-rays in water, they found that the SPR at depth $d$ in a field of radius $r$ is accurately described by

$\mathrm{SPR}=aZ+bZ^{2}, \qquad Z=\frac{rd}{r+d},$

where the combined geometric variable $Z$ — half the harmonic sum of field radius and depth — collapses data taken at different ratios $r/d$ onto nearly common curves, far better than conventional plots against $r$ alone. The fitted coefficients showed a striking regularity: $a$ is determined by the effective (spectrum-averaged) linear attenuation coefficient $\bar{\mu}$ of the beam in water and by the ratio $r/d$, and $b$ depends on $r/d$ and on energy, the energy dependence being approximately proportional to $\bar{\mu}^{3}$. A qualitative argument suggests why the leading term is linear: to first order in the scattering probability, scattered kerma is produced in proportion to the mass of irradiated material effectively contributing to the point, and that effective amount grows like $Z$. The same argument cannot explain the accuracy of the quadratic form over the studied ranges, nor the $\bar{\mu}^{3}$ scaling of $b$, and the authors' own order-separated Monte Carlo showed that identifying the term $aZ$ with first-order scatter fails away from the limit $Z\to 0$, because attenuation of single-scattered photons and accumulation of multiple scatter pull in opposite directions as the geometry grows.

The relation is practically useful: measured beam data such as tissue-air ratios — TAR, the ratio of the dose at a point at depth $d$ in the phantom to the dose at the same point in free air, tabulated per field size — converge when plotted against $Z$, which aids extrapolation to the small fields and shallow depths where direct measurement is hardest. That same use exposed a second issue: extracting SPR from published 60Co TAR tables via $\mathrm{SPR}=\mathrm{TAR}(s,d)/\mathrm{TAR}(0,d)-1$ (square fields of side $s$ converted to an equivalent circular radius through the Bjärngard-Siddon equivalent-circle rule) and extrapolating to zero field produces a nonzero intercept, suggesting that the zero-field quantity $\mathrm{TAR}(0,d)$, conventionally taken to be pure primary dose, itself contains a few percent of scatter. The paper left the physical origin of the law, the physical interpretation of its coefficients, and a consistent definition of the zero-field primary reference unresolved, while later beam-data quality-assurance tools and zero-field extrapolation methods continued to build on the law.

Problem Statement

Derive, from photon-transport first principles — energy-dependent Compton and photoelectric cross sections, attenuation of primaries and of scattered photons, and the geometry of a circular field of radius $r$ at depth $d$ in a water phantom — why the central-axis kerma scatter-to-primary ratio obeys

$\mathrm{SPR}(r,d)=aZ+bZ^{2}, \qquad Z=\frac{rd}{r+d},$

over the broad energy range studied (from about 200 keV monoenergetic photons through 60Co gamma rays to 4 and 15 MV x-ray spectra), and account for the observed coefficient structure: (i) $a$ determined by the spectrum-averaged linear attenuation coefficient $\bar{\mu}$ of the beam in water and by the ratio $r/d$; (ii) $b$ depending on the ratio $r/d$ and on energy, the energy dependence being approximately proportional to $\bar{\mu}^{3}$; and (iii) an explanation of what contributions beyond first-order scatter are required to account for the quadratic term and the accuracy of the two-parameter form, given that the source demonstrated only qualitatively that $\mathrm{SPR}\simeq aZ$ if first-order scatter alone is accounted for, and stated that the relation between SPR and $Z$ is not fully explained. The derivation should delimit the domain in $Z$, $r/d$ and energy within which the quadratic truncation is accurate and predict where it fails (for example, regimes where higher-order terms or backscatter become significant).

References

1. B. E. Bjärngard and P. L. Petti, 'Description of the scatter component in photon-beam data', Physics in Medicine and Biology 33 (1988) 21-32. DOI: 10.1088/0031-9155/33/1/002. https://doi.org/10.1088/0031-9155/33/1/002
2. B. E. Bjärngard and R. L. Siddon, 'A note on equivalent circles, squares, and rectangles', Medical Physics 9 (1982) 258-260. DOI: 10.1118/1.595161. https://doi.org/10.1118/1.595161
3. P. S. Nizin, 'Geometrical aspects of scatter-to-primary ratio and primary dose', Medical Physics 18 (1991) 153-160. DOI: 10.1118/1.596701. https://doi.org/10.1118/1.596701
4. A. Iwasaki, 'Characteristics of 60Co gamma-ray SPR (scatter-primary ratio), SF (scatter factor), beta (dose-kerma ratio), and dmax (depth of maximum dose)', Physics in Medicine and Biology 39 (1994) 1081-1088. DOI: 10.1088/0031-9155/39/7/002. https://doi.org/10.1088/0031-9155/39/7/002
5. R. K. Rice and L. M. Chin, 'Monte Carlo calculations of scatter to primary ratios for normalisation of primary and scatter dose', Physics in Medicine and Biology 35 (1990) 333-338. DOI: 10.1088/0031-9155/35/3/003. https://doi.org/10.1088/0031-9155/35/3/003
6. A. L. McKenzie, 'Should 60Co tissue-air ratios be re-evaluated?', Physics in Medicine and Biology 37 (1992) 1601-1610. DOI: 10.1088/0031-9155/37/7/011. https://doi.org/10.1088/0031-9155/37/7/011
7. A. L. McKenzie, 'What is the explanation for the changes to cobalt-60 tissue-air ratios in BJR Supplement 25?', Physics in Medicine and Biology 42 (1997) 1055-1064. DOI: 10.1088/0031-9155/42/6/005. https://doi.org/10.1088/0031-9155/42/6/005
8. A. Iwasaki, '10 MV x-ray SMRs obtained using zero-area S(p) correction factors derived by means of the Bjärngard-Petti method', Physics in Medicine and Biology 41 (1996) 625-636. DOI: 10.1088/0031-9155/41/4/004. https://doi.org/10.1088/0031-9155/41/4/004
9. A. Iwasaki, '10 MV x-ray zero-area phantom scatter correction factors (Sp) obtained using three extrapolation methods', Physics in Medicine and Biology 41 (1996) 2627-2634. DOI: 10.1088/0031-9155/41/12/003. https://doi.org/10.1088/0031-9155/41/12/003
10. P. S. Nizin, D. M. Bellezza and R. B. Mooij, 'The elements of tissue-air ratio and systematic error', Medical Physics 23 (1996) 1191-1197. DOI: 10.1118/1.597864. https://doi.org/10.1118/1.597864
11. Y. Xiao, M. D. Altschuler and B. E. Bjärngard, 'Quality assurance of central axis dose data for photon beams by means of a functional representation of the tissue phantom ratio', Physics in Medicine and Biology 43 (1998) 2195-2206. DOI: 10.1088/0031-9155/43/8/014. https://doi.org/10.1088/0031-9155/43/8/014
12. P. S. Nizin and R. B. Mooij, 'Tissue-air ratios for narrow 60Co gamma-ray beams', Medical Physics 25 (1998) 1458-1463. DOI: 10.1118/1.598319. https://doi.org/10.1118/1.598319
13. J. McDonough, P. Bloch and B. E. Bjärngard, 'Determination of scatter factor parameters and electron disequilibrium for monoenergetic photon beams', Medical Physics 26 (1999) 200-205. DOI: 10.1118/1.598504. https://doi.org/10.1118/1.598504
14. X. A. Li, 'Peak scatter factors for high energy photon beams', Medical Physics 26 (1999) 962-966. DOI: 10.1118/1.598489. https://doi.org/10.1118/1.598489
15. K. K. Wang and T. C. Zhu, 'Modeling scatter-to-primary dose ratio for megavoltage photon beams', Medical Physics 37 (2010) 5270-5278. DOI: 10.1118/1.3480480. https://doi.org/10.1118/1.3480480
16. A. Iwasaki, 'A method for 10MV X-ray scatter dose-spread kernel construction based on the Bjärngard scatter factor formulation', Radiation Physics and Chemistry 64 (2002) 89-98. DOI: 10.1016/S0969-806X(01)00449-2. https://doi.org/10.1016/S0969-806X(01)00449-2