Page 395 - 2014 Printable Abstract Book
P. 395
exposure, the possibility of other cancers at short lags cannot be arbitrarily excluded. Furthermore, as
cancer follow-up of atomic bomb survivors did not commence until five years after exposure, the Life Span
Study (3) can provide no data relevant to short lags. Using individual estimates of radiation dose, we are
now testing the possibility that the ERR/dose coefficient for solid cancers is greater at low-doses (<100
mGy from medical X-rays), than at the higher average doses following atomic bomb radiation. We are also
working to assess the possible magnitude of any residual bias from reverse causation. References: 1.
Mathews, J.D., et al., British Medical Journal, 2013. 346: p. f2360. 2. Pearce, M.S., et al. Lancet, 2012.
380(9840): p. 499-505. 3. Preston, D.L., et al. Radiation research, 2007. 168(1): p. 1-64.
(SND02) Reconstructing organ doses from paediatric CT scans within the EPI-CT study. Isabelle Thierry-
3
1
4
2
5
6
7
Chef ; Jeremie Dabin ; Tore S. Istad ; Andreas Jahnen ; Neige Journy ; Lucian Krille ; Choonsik Lee ; Carlo
3
8
7
9
Maccia ; Arvid Nordenskjöld ; Hilde M. Olerud ; and Steven L. Simon, International Agency for Research
1
2
on Cancer, Lyon, France ; Belgian Nuclear Research Centre, SCK•CEN, Mol, Belgium ; Norwegian Radiation
4
3
Protection Authority, Østerås, Norway ; Public Research Center Henri Tudor, Luxembourg, Luxembourg ;
5
Radiation Protection and Nuclear Safety Institute, IRSN, Fontenay-aux-Roses, France ; University Medical
Center of the Johannes Gutenberg, University Mainz, Mainz, Germany ; National Cancer Institute,
6
8
7
9
Bethesda, MD ; CAATS, Bourg-la-Reine, France ; and Karolinska University Hospital, Stockholm, Sweden
The EPI-CT study, funded by the European Union, was designed to quantify risks from a
multinational cohort study of around 1,000,000 children who have been exposed to ionizing radiation
from computed tomographic (CT) imaging. For assessment of risks, accurate estimate of organ-specific
doses is essential. A flexible approach for dose reconstruction was developed that can accommodate
patient and imaging data available from historical sources as well as data automatically extracted from
images. For the distant past (prior to 2000), only sparse information about scanner settings can be
obtained. An approach for those years was used to combine data from a questionnaire, surveys, scientific
publications, and expert interviews in order to assess typical radiological protocols. For recent years,
scanner settings can be directly extracted from the Digital Imaging and Communications in Medicine
(DICOM) headers of recorded images available in the Picture Archiving Communication System (PACS).
Radiation fields and the x-ray interactions within the body are simulated using hybrid phantoms of
different sex and ages and Monte-Carlo-based radiation transport calculations. Individual organ doses are
estimated for each child using the above calculation strategy with the available input data for the time of
exposure. To account for uncertainties in our dose estimates due to missing data a simulation method is
proposed. This approach maintains correlations of doses for persons within subgroups with similar
exposure attributes and simulates uncertain dose-model parameters values that could otherwise lead to
biases if a single imputation strategy was used. We anticipate that the main parameters that are likely to
be missing are: manufacturer and model of the CT machine, scan parameter settings for each examination
type/body region and age group (mAs, kVp, pitch, FOV/type of bow tie filter, number of series/use of
intravenous contrast), and scanned area and body size of the patient. Each missing parameter will be
represented by a probability density function (PDF) representative of the state of knowledge for the
appropriate time period. For each calculation of the cohort dose distribution, values of parameters will be
selected from the appropriate PDFs while maintaining proper correlations between parameters.
cancer follow-up of atomic bomb survivors did not commence until five years after exposure, the Life Span
Study (3) can provide no data relevant to short lags. Using individual estimates of radiation dose, we are
now testing the possibility that the ERR/dose coefficient for solid cancers is greater at low-doses (<100
mGy from medical X-rays), than at the higher average doses following atomic bomb radiation. We are also
working to assess the possible magnitude of any residual bias from reverse causation. References: 1.
Mathews, J.D., et al., British Medical Journal, 2013. 346: p. f2360. 2. Pearce, M.S., et al. Lancet, 2012.
380(9840): p. 499-505. 3. Preston, D.L., et al. Radiation research, 2007. 168(1): p. 1-64.
(SND02) Reconstructing organ doses from paediatric CT scans within the EPI-CT study. Isabelle Thierry-
3
1
4
2
5
6
7
Chef ; Jeremie Dabin ; Tore S. Istad ; Andreas Jahnen ; Neige Journy ; Lucian Krille ; Choonsik Lee ; Carlo
3
8
7
9
Maccia ; Arvid Nordenskjöld ; Hilde M. Olerud ; and Steven L. Simon, International Agency for Research
1
2
on Cancer, Lyon, France ; Belgian Nuclear Research Centre, SCK•CEN, Mol, Belgium ; Norwegian Radiation
4
3
Protection Authority, Østerås, Norway ; Public Research Center Henri Tudor, Luxembourg, Luxembourg ;
5
Radiation Protection and Nuclear Safety Institute, IRSN, Fontenay-aux-Roses, France ; University Medical
Center of the Johannes Gutenberg, University Mainz, Mainz, Germany ; National Cancer Institute,
6
8
7
9
Bethesda, MD ; CAATS, Bourg-la-Reine, France ; and Karolinska University Hospital, Stockholm, Sweden
The EPI-CT study, funded by the European Union, was designed to quantify risks from a
multinational cohort study of around 1,000,000 children who have been exposed to ionizing radiation
from computed tomographic (CT) imaging. For assessment of risks, accurate estimate of organ-specific
doses is essential. A flexible approach for dose reconstruction was developed that can accommodate
patient and imaging data available from historical sources as well as data automatically extracted from
images. For the distant past (prior to 2000), only sparse information about scanner settings can be
obtained. An approach for those years was used to combine data from a questionnaire, surveys, scientific
publications, and expert interviews in order to assess typical radiological protocols. For recent years,
scanner settings can be directly extracted from the Digital Imaging and Communications in Medicine
(DICOM) headers of recorded images available in the Picture Archiving Communication System (PACS).
Radiation fields and the x-ray interactions within the body are simulated using hybrid phantoms of
different sex and ages and Monte-Carlo-based radiation transport calculations. Individual organ doses are
estimated for each child using the above calculation strategy with the available input data for the time of
exposure. To account for uncertainties in our dose estimates due to missing data a simulation method is
proposed. This approach maintains correlations of doses for persons within subgroups with similar
exposure attributes and simulates uncertain dose-model parameters values that could otherwise lead to
biases if a single imputation strategy was used. We anticipate that the main parameters that are likely to
be missing are: manufacturer and model of the CT machine, scan parameter settings for each examination
type/body region and age group (mAs, kVp, pitch, FOV/type of bow tie filter, number of series/use of
intravenous contrast), and scanned area and body size of the patient. Each missing parameter will be
represented by a probability density function (PDF) representative of the state of knowledge for the
appropriate time period. For each calculation of the cohort dose distribution, values of parameters will be
selected from the appropriate PDFs while maintaining proper correlations between parameters.
