POTENTIAL OF HYBRID COMPUTATIONAL PHANTOMS IN
RADIOTHERAPY FOR HEART AND CORONARIES DOSIMETRY
Alexandra MOIGNIER1, David BROGGIO1, Sylvie DERREUMAUX2, Michel CHEA3,
Gilbert BOISSERIE3, Bernard AUBERT2, Didier FRANCK1 and Jean-Jacques MAZERON3
1
Institut de Radioprotection et de Sûreté Nucléaire, IRSN/PRP-HOM/SDI/LEDI, BP-17, 92262
Fontenay-aux-Roses, France
2
Institut de Radioprotection et de Sûreté Nucléaire, IRSN/PRP-HOM/SER/UEM, BP-17, 92262
Fontenay-aux-Roses, France
3
Groupe Hospitalier Pitié Salpêtrière (GHPS), AP-HP, Service de Radiothérapie, Paris, France
Purpose. Many studies have highlighted that radiotherapy induces an increased risk of
heart diseases. However, current retrospective cardiovascular dosimetry studies are based on a
representative patient [1] or simple mathematical phantoms [2]. In this study, a process of
patient modeling is developed to personalize the thorax anatomy and to add a heart model
with coronary arteries [3].
Methods and Materials. The patient models are hybrid computational phantoms (HCPs)
on which a detailed heart model is inserted using computer-aided design software. A CT
acquisition (pseudo-CT) was derived from HCP and imported in DICOM format into a
Treatment Planning System where hospital treatment conditions were reproduced. From a
database of patients treated for left-side breast cancer, six patients were selected: three were
modeled from their CT images (“A” patients) and the others from two orthogonal radiographs
(“B” patients). The method performance and limitation were investigated through quantitative
comparison between the initial CT and the pseudo-CT. Namely, the morphology and the dose
calculation were compared. For the “B” patients, a comparison with two kinds of
representative patients was also conducted. Finally, dose assessment was focused on the
whole coronary artery tree and the left anterior descending coronary.
Main results. When 3D anatomical information was available, the dose calculations
performed on the initial CT and the pseudo-CT were in good agreement. For the models built
from radiographs, comparison of doses derived from HCP and representative patient showed
that the HCP doses were either better or equivalent. In the left-breast radiotherapy context and
for the studied cases, coronary mean doses were at least 5-fold higher than heart mean doses.
Conclusion. For retrospective dose studies, it is suggested that a library of HCPs offers a
better surrogate, in terms of dose accuracy, than representative patients. The use of a detailed
heart model partially eliminates the problem of coronaries identification on the patient CT.
[1] Taylor CW, Nisbet A, McGale P, et al. Cardiac exposures in breast cancer radiotherapy:
1950s-1990s. Int. J. Radiat. Oncol. Biol. Phys. 2007;69(5):1484-1495.
[2] Shamsaldin A, Grimaud E, Hardiman C, et al. Dose distribution throughout the body from
radiotherapy for hodgkin's disease in childhood. Radiother. Oncol. 1998;49(1):85-90.
[3] Moignier A, Derreumaux S, Broggio D, et al. Potential of hybrid computational phantoms
for retrospective heart dosimetry following breast radiotherapy. Accepted in Int. J. Radiat. Oncol.
Biol. Phys. March 2012.
KEYWORDS. Hybrid computational phantom, Breast radiotherapy, Cardiovascular
dosimetry, Heart model