PHE comments on ICRU/ICRP draft report on the operational quantities: combined comments of several PHE representatives
The aspiration to make the operational quantities better reflect the risk/detriment associated with an exposure is laudable. This will, if it is implementable, improve radiation protection. It could also make the epidemiological interpretation of occupational doses more accurate, but occupational doses nowadays are so low that they will still carry very little weight in epidemiological studies. The changes will, however, add a complication to epidemiological studies because they will create an additional disconnect between doses monitored versus risk when implemented, with care needing to be taken about which data refer to the new and old quantities. This will be a particular problem if there is not a common implementation date in different countries, but these problems already exist in historic dose data.
The change to quantities in gray for limits to control tissue reactions will add clarity. The current dose limit for the skin in terms of Sv can confuse, since a 500 mSv dose limit is not appropriate for stochastic effects. For the eye lens the change is probably a good thing, but ICRP Publication 118 leaves open the possibility that cataract induction is stochastic or partially stochastic, so ICRP need to make a decision on whether this change is appropriate for the eye lens.
It is understood that one of the main drivers for this change is the need for a common set of quantities that can be used for operational purposes for all radiation fields. It does not make sense for personnel at accelerators, air crew and astronauts to have different quantities defined for their exposures, and the current quantities as defined in ICRU57/ICRP74 do not meet the needs of high energy fields. But astronauts already have different quantities, defined in ICRU84. It is also important that proposals for the replacement of the current quantities do not create more problems than they solve.
Some of the problems stem from the application of kerma factors in the calculations used to produce the conversion coefficients in ICRU57/ICRP74. It should be recognized that for neutrons at that time no other option was possible, and for photons, the development of electron transport and implementation of radiative processes in the cross sections was in its infancy. So whilst this was not in accordance with the definitions, the definitions were in advance of computational capability. In practice, for the vast majority of workplaces, this does not cause any problem. For higher energies, it does, but full transport of the charged particles generated within the phantom/body, and simulation of the radiative processes, causes perhaps as many problems as it solves. By not using a discrete depth for H*, the kerma approximation version of the new quantities does not change very significantly compared to the reference values of H* with full charged particle transport, for the energy range used. It could then be argued that it would be accurate enough to use the kerma approximation in the anthropomorphic phantom for H* as part of the definition. Computation times would be much reduced, and the new quantity would be more easily and accurately applied in the radiation protection community.
In terms of high energy fields there are distinct areas of significance:
The proposals may cause too much change in routine radiation protection, which has benefitted from the stability afforded by almost no change to the quantities used since the 1980s. The new quantities will require almost all photon instruments to be redesigned, if they aspire to measure accurately in low energy fields. Further, neutron instruments may need redesign, and all types of personal dosemeters will not respond adequately in terms of these quantities. Cost benefits are lacking from the report. The costs may be significant.
In addition to the issues of redesign, there are real concerns that for anyone wearing personal dosemeters in highly scattered fields, the difference between ROT and AP may need them to have more than one dosemeter to estimate this quantity. Analysis of the PHE neutron dosimetry systems implies that this is the case for the method currently used for calibration. Personal dosimetry services will need to determine the response of their dosemeters for reverse angles of incidence, but those conversion coefficients are not given. Conversion coefficients for 135° would help a lot, but it may still be very difficult to design dosemeters worn on the front of the body to respond correctly.
With all the changes proposed, there will need to be changes to many international standards. All of the reference field standards produced by ISO will need major revision. There need to be reference field standards and reference fields for protons, helium ions, muons and pions. Currently we cannot calibrate such dosemeters. IEC will need to amend their large number of standards that deal with instrument and/or dosemeter characteristics. All of this work will need to be synchronized with the changes in legislation that are required to implement these recommendations.
Further, ICRP are currently reviewing the range of reference phantoms available, with different sizes/ages being considered, and moves away from the reference voxel structure. In this context, the operational quantities are not just used for occupational protection: environmental exposures to the public will include age dependent phantoms, so logically, for such exposures the envelope used for H* should include other phantoms? This is also true of exposures on aircraft, where a range of ages are exposed, though that is not an issue for occupational purposes. If nurbs/mesh phantoms become the ICRP reference, then presumably those should be used for the calculation of effective dose and H* and Hp will change? If ICRP review the tissue and/or radiation weighting factors as better data become available for biological effects, then presumably H* and Hp will again change? This period of decades of stability could be replaced by an era in which the operational quantities have to change to keep consistency with the protection quantities of the ICRP, which would repeat the impact on ISO and IEC standards. Again, evolving H* and Hp as new scientific evidence becomes available is a good idea in principle, but in practice the financial and logistical implications might be immense. The only alternative would be to ‘fix’ H* and Hp at the 2017 values of E, which would solve the problem, but also defeat their purpose.
The kerma approximation is used in the report for the conversion coefficients for practical calibrations, which is necessary. It would help if there were a different symbol used for the quantity with the kerma approximation applied. This report needs to recommend the appropriate build-up for the highest energies: a 50 MeV electron has a range of 160 m in air, and the current build up used in calibration laboratories would be completely inadequate. In practice, for these high energies a set of values with different amounts of secondary charged particle build up would be preferable.
What is the purpose of H* data for photons from 50 MeV to 1 TeV when calculated without the kerma approximation. The same applies everywhere that the quantities are calculated above the energy range for calibration purposes given in this report.
Much thought needs to be given for the lowest energy for which conversion coefficients are tabulated. Low energies do not make any sense in the reference phantom because the skin is not modelled accurately. There should not be any H* data published for energies lower than are used for the proposed skin model. For photons this would limit the lower energy to 10 keV, which would be consistent with ICRP 116. The problems are seen in a comparison between the ICRP 116 value for dose to the skin of 1.74 pGy cm2 for 10 keV photons, compared with the value of 7.2 pGy cm2 in this report (Table A.4.1.2a). The difference is caused by the skin in the reference phantom having a very high mass compared to the 50 micron layer used in this report: the skin is one voxel thick with no sub-division. This problem is complicated by the different voxel sizes of the male and female phantoms leading to artificial differences in the skin dose results. So, the skin contribution to H* in this report will be much too low. For energies below 10 keV this will become more severe, but those energies are really not of concern for external dosimetry.
The reference (Otto, 2017) needs to be published in a peer reviewed journal before this report is published. A private communication for these important data is not good enough. The same is true of (Endo, 2016b). ICRU have had problems with this in the past.
The neutron tables give a value for 25 meV. Conventionally, this should be 25.3 meV to represent thermal neutrons at STP. However, it needs to be stated whether this is a monoenergetic field or a Maxwell-Boltzmann distribution. The latter would be more useful since it would give a “thermal” point.
There is a problem with the data for Hp used for positive angles of incidence (30° to 75°) because of the “staircase effect” – the voxelization does not permit proper coverings of breast and brain tissue, in particular, when not irradiated from the front, back or side. For charged particles, this can enable them to reach tissues that should be beyond their range. So, where the Hp data increase away from 0° it is likely to be caused by artefacts of the phantom rather than the real risk to an individual.
Overall, it is clear that the new quantities will provide more realistic estimates of risk than the current ones. But, that advantage will be countered by a potential need to redesign and recalibrate every instrument and dosemeter currently in use. There is thus a significant cost-benefit issue associated with the recommendations, the resolution of which is not clear. However, given the low doses typically relevant in operational protection and the uncertainties associated with their routine measurement, one wonders: how many lives will actually be saved as a result of changing the definitions of these quantities at the base level? What is needed, therefore, is a follow up report from ICRU, with strong input from operational health physicists, ISO and IEC, to assess the costs and practicality of implementing these proposals, and to give recommendations on precisely how they should be implemented.
Line 218: Change “… the study the values of …” in “... the study of the values ...”.
Figure 1: there needs to be an arrow from Measured Quantities to Protection Quantities, because the measurements are entered into dose records as estimates of the protection quantities.
Lines 303-306: the reference (Harrison et al 2016) does not support this statement and it also does not seem to be in any ICRP Publication. Epidemiology indicates that there is a linear relationship between effective dose and stochastic effects ABOVE 100 mSv. It is used to extrapolate using the linear no-threshold model below 100 mSv. Stochastic effects are even more important above 100 mSv, which may not be received in a very short time interval and hence may carry no risk of tissue reactions.
Lines 306-307: “For evaluating tissue reactions, mean absorbed dose in the organ or tissue is assessed, ...” has exceptions, such as for instance local skin, for skin redness or necrosis, and spinal cord for paralysis effects, where only a local part of the organ or tissue needs to be assessed.
Lines 312-313: the reference is not found by a simple internet search. This is probably not a reference that will be accessible to the reader – a better reference should be found.
Lines 326-329: (Endo, 2017) compares data for common energies and geometries with the data in ICRP 116 as part of the validation process. However, the data presented for validation within ICRP 116 are scant, and the comparison in (Endo, 2017) is difficult to interpret for the same reason. The reliance in this work on a single code is unfortunate and not good practice. It is odd that the fundamental data use just one code, but there are three codes used to calculate doses to the eye lens and skin.
Lines 326-329. Focussing just on the 1% statistical error gives the impression that the results, and hence the risk estimates, are much more accurate than they actually are. It is worth pointing out that whilst the statistical uncertainties are less than 1% for the calculations, the uncertainties in the cross-sections are larger, and the uncertainties produced at higher energies by the nuclear models are significantly greater. Some discussion should be included on the likely magnitudes of the non-statistical uncertainties on the data. Parameter variation of, say, these cross section or interaction model choices may be a first step in analysing this, the latter of which may be critical at high energies.
Lines 326-329. There are some issues with the validation:
Lines 333-335: it is understood that the operational quantities are defined at a point, but the choice of phantoms will lead to considerable confusion when these quantities are implemented because it will not be obvious that this is a point quantity.
2. CONCEPTS AND TERMS
Line 339: this is not how Ep is used. Ep is used for a variety of particles not just photons. The index p is used for personal in this report and should not also be used for particle. The particle type is i so why not Ei?
Lines 341-345: the kerma approximation is exactly that, an approximation that is valid, within experimental uncertainties, across a volume. Charged particle equilibrium at a point is a difficult concept, but in no practical situation is this definition of charged particle equilibrium ever true. The kerma approximation is being abandoned to cope with high energy fields, where the charged particles generated have long ranges, and this sort of equilibrium is never obtained because of the different materials present. The definition as given should say constant with respect to particle energy AND direction. [Note: later on we find that all the data that will ever be used are calculated with the kerma approximation]
Line 386 (First occurrence): Here and elsewhere, describing dX/dY as dX divided by dY etc., or some such variation, is mathematically wrong, and weakens the perceived rigour of the document. It is also unnecessary in many cases, as in practice concepts such as doses are only ever of interest when averaged over a finite volume. Moreover, extending them to infinitesimals implies that they meaningful at all scales, which disregards the impact of e.g. structure or nanodosimetric complications. Why not simply define dose as delta(E) / delta(m), for example, or even just as “the dose to a given volume is the energy deposited in that volume divided by its mass, D=E/m”, and likewise for the other differential definitions? This would also make more sense in the subsequent Monte Carlo calculation of these quantities.
3. DEFINITONS OF QUANTITIES
Line 400, Equation 3.4: where a parameter or quantity has a functional dependence on another, that should be expressed as X(Y) not XY. Subscripts should be reserved for things like essential labels, e.g. a p to indicate ‘personal’ as in Hp, or else used as discrete indices to delineate members, e.g. where summation is required. Either way, whichever notation is adopted it should apply consistently throughout the entire document.
Line 401: Similarly, use of the subscript ‘p’ to denote a generic particle (c.f. ‘i’ used later for species) is not defined specifically, and is not always applied consistently in the notation. The clash with Hp later is also unfortunate.
Line 403: it is even worse to use the parameters as a subscript and in a functional relationship.
Lines 401-403: a reference direction needs to be defined for the sphere (line 387) for this definition to be used.
Lines 409-422: the fluence has just been defined as having a functional dependence on the direction, now the radiance is being defined as the functional dependence of the fluence on the direction. Are both definitions needed?
Lines 425-428: there needs to be a comma between p and tr or this is very confusing.
Line 433 (First occurrence): Here and elsewhere, referring to the kinetic energy of a photon (or any other massless particle) is physically dubious.
Line 435: The definition of collision kerma still holds when radiative losses are negligible: it’s just that its value approaches that of kerma in such cases.
Line 467: more care needs to be taken throughout with the equations. In this instance, it is not obvious whether the integral is part of the numerator or denominator, from the way in which the equation is written.
Line 474-479: The reference or definition of the physical description of the eye lens is missing as is given for the local skin above.
Lines 490-505: it should be explicitly stated that H* is a scalar quantity, with no directional dependence.
Line 492: this would be much better written as h* without the subscript.
Line 494: this is one of many instances of confusing subscripts. Ep was earlier only for photons, and now there is no room for an index i for particle type? Ei could be used, or better still, using epsilon or something instead for energy (which is what ICRP use) to avoid the clash with effective dose. Again, whichever notation is chosen, it must be made consistent throughout the entire document.
Line 504: the summation needs to make it clear this is summation over i (also applies elsewhere in the document)
Lines 517-518: this equation should get its own space and number – it looks a mess in the line.
Line 518: “exposure of the whole body of the stylized eye model” is not clear. It is presumed that the eye model is inserted into the reference phantom and the whole body of the reference male and female phantoms exposed? Since so few particles would be scored, this must require some clever variance reduction to get 1% uncertainty: the report should explain how this was achieved.
Lines 529-534: this definition of the angles is not clear. A diagram is needed.
Lines 547-554: it is good that there is now a defined depth range and area in skin rather than a specific depth/point as before. This makes it “obvious” how the quantity should be scored. It is, however, odd that the slab phantom should be used in the definition rather than one of the extremity phantoms, which are much more important for skin dosimetry. The move to Gy rather than Sv, so that the quantity is more focussed on tissue reactions, makes this choice less appropriate. Rather than saying inside the phantom is a layer of skin, it should be made clear that the front 2 mm is replaced by ICRP skin. How does this definition work with the rod and pillar phantom? They have curved faces. It becomes evident in section 3.4.9 that the intention is that area monitoring is not to be used for estimation of doses to the extremities, which may not always be the case. It is not clear that “local” is needed in this definition.
Line 580: is this footnote actually used? Nothing has these angle dependences. It is in any case very confusingly written. Later it appears in the appendices so reference should be given or it should be moved. The definition is very unclear.
Line 577: the definition needs discussion of the reference direction.
Lines 577-585: how the male and female phantoms are dealt with should be explained.
Lines 580-588: An explanation needs to be provided for why the maximum of left-right angles is taken, but not up-down angles, since the asymmetry of the body is much greater up-down than the left-right. Why not define Hp(omega) asymmetrically as Hp(theta,phi) rather than Hp(theta,Max(phi))… conservatism perhaps, but then if so, why not Hp(Max(theta),Max(phi))? Presumably dosemeters will need to be redesigned anyhow to measure the new dose quantity, so including asymmetries in them might not be any greater issue. But then the location where they are worn on the body may impact upon their readings. But then, perhaps dosemeter location should be specified if ICRU consider that Omega is important? This would confuse calibration, however, as Hp(omega) is a point quantity, so ICRU would need also to recommend what calibration phantom should be used in practice. Thus, ICRU need to explain why their proposal for Hp reflects the optimum compromise: this is a complicated issue that has not been dealt with adequately in the report, with the proposed choice appearing a little arbitrary.
Line 592: it is understandable, that these quantities are based simply on the field at a point, but to define eye lens dose at a point on the “head or body” is very confusing. Do we really want an eye lens quantity defined on (but not in) some arbitrary point on the body? The reference direction then needs to be added to this definition. Should the position and orientation of the phantom, for the calculations, not be dictated by the reference point and the direction the head is facing? Perhaps the most significant eye lens exposures are for an interventional radiologist looking down, which fits very uncomfortably with this definition. In that case the obvious reference direction for Dlens would differ from the natural reference direction for Hp.
Line 595: this footnote is also not used?
Line 598: it is hard to see why there would be different maximum doses for positive and negative angles for the chosen geometries, other than those caused by statistics, because the model has left-right symmetry. The average would be better for the data in this report. Up and down would impact more on these data but those angles have not been considered.
Line 611: no equation has a functional dependence on theta. Is this footnote used?
Lines 614-625: Given the overall aims of this work, perhaps it would be useful to include a discussion examining why these specific phantom dimensions and shapes are considered appropriate and optimal, reminding readers that these are not simply arbitrary decisions
Lines 618-625: why are the densities of the rod and pillar higher? Is this from a previous ICRU recommendation? The reference should be given. The definitions seem to define two volumes, “front” and “back”? This needs clarification.
Line 626: the layer is not inside.
4. CONVERSION COEFFICIENTS
The method recommended for bone dosimetry in ICRP 116 to account for the breakdown of kerma conditions was not then applied therein to generate the conversion coefficients it subsequently reported in its appendices. For Hp and H*, which rely on effective dose data, ICRU should clarify whether this method was adopted in their calculations, and explain that choice.
Line 659: “caudal” is being used for “from below”, but derives from tail? It is an ambiguous word in this instance because it could be synonymous with PA. Top and bottom would be better.
The purpose and added value of this section are not really clear, especially 5.2. Perhaps it should be shortened, and/or merged into the Introduction in Section 1.
A number of very general statements are made regarding the impact of the proposed changes on detector responses, but are not substantiated or properly explained. These assertions should be fully justified, with proper references provided if appropriate. In fact, this section needs to be expanded significantly (or a new section added), with a proper cost-benefit analysis added for each type of measurement, a full discussion of the relative pros and cons anticipated from making the proposed change, and a thorough defence of its conclusions justifying why ICRU recommend that it should be carried forward.
Line 1269: Probably “If this were to take place, …” is meant instead of “In this were to take place, …”.
The data presented in this appendix reference just a handful of papers written by a small group of authors, most of which are also on the report committee. Although the abilities of these individuals are not in dispute, given the importance of the results and their very wide impact within the field, considerably more effort could have been made to have them produced and cross-checked by a much larger pool of contributors using different codes, different physical data, different interaction models, and different approaches to the underlying assumptions. This would be relatively easy given the global ubiquity of Monte Carlo codes and their users, and would lend significant support to the quality assurance of the results, as well as making the ICRU process much more transparent.
Table A.2.3. The data in this table at acute angles, especially at low energies, are likely to be affected by the artificial ‘stair casing’ effect due to impinging on the corners/edges between cubic voxels. This will be true of all other charged particle, i.e. short-range, data. That these data can become limiting might therefore be nothing more than an artefact of the calculation method. This should be discussed.
A. 4 Directional and Personal Absorbed Dose to Local Skin. Why is the local skin dose expressed in ICRU 4 element tissue instead of skin (ICRP, 2009), especially because the phantoms contains an outer layer of 2 mm skin and the dose is determined at 50 to 100 micrometre depth, it is inside the skin layer. It needs to be explained why ICRU 4-element tissue (density of 1 g/cm3) has been used with a density of 1.11 g/cm3. ICRP skin (ICRP 110) has a density of 1.09 g/cm3.
This appendix needs to be expanded significantly: as it stands, it’s not really clear what it contributes. The descriptions of the MC codes are just generic summaries of them...all very interesting, but readily available elsewhere. Conversely, missing from this section is explicit documentation of the precise methods, models, and assumptions that were made in each case to generate all of the data given in Appendix A: for good scientific practice and openness, it should be possible for any independent modeller to be able to use the descriptions provided to recreate all of the published results. These choices will be dependent on factors such as particle species and energy, especially at high energies where they may be critical, and should be recorded in full within the document.
Line 1897-1905: Please consider writing Dual Parton Model (DPM) instead of Dual Parton Model, so the reader is familiar with the abbreviation DPM used in DPMJET code below.
Line 1930: It is noted that the thermal neutron treatment for water is applied for the skin and lens of the eye in the calculations that used MCNP6. However, no mention is made of it being applied for the calculations made using PHITS or FLUKA: the method used should be stated for those codes as well. Justification should be given for why water was chosen.
Line 1935 and 1944: According to Table B.1 the used cross-section data for H-1 are in 1001.62c but the latest cross section data for H-1 are in 1001.80c according to the Conlin et al reference cited. This commented sentence is not in accordance with the Table B.1. Please specify why 1001.62c is preferred over 1001.80c.
Perhaps some discussion could be made as to the recommended choice of critical volume for the eye lens, rather than just providing competing datasets. Without this recommendation, the practical benefit of this appendix is less apparent.
Lines 2095-2099: The sentence "The conversion coefficients are to the value of the absorbed dose to the lens of the eye, the maximum value of that to the radiation sensitive cells or the complete lens, ..." seems to be ambiguous. The table gives one value for 3 different things:
1) The absorbed dose to the lens of the eye integrated over the whole lens of the eye.
2) The (locally) maximum absorbed dose to the radiation sensitive cells,
3) The (locally) maximum dose to the complete lens.
It remains unclear how for one eye these three different quantities have to be combined to one quantity. It is clear that the right eye lens conversion coefficient and the left eye lens conversion coefficient are combined by taking the maximum conversion coefficient.