Premium
SU‐FF‐J‐36: Efficient 4D Treatment Planning Using CT Datasets Synthesized Via Deformable Image Registration
Author(s) -
Zhong H,
Fragoso M,
Chetty I
Publication year - 2009
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.3181328
Subject(s) - image registration , radiation treatment planning , computed tomography , benchmark (surveying) , nuclear medicine , computer science , interpolation (computer graphics) , displacement (psychology) , hounsfield scale , medicine , artificial intelligence , image (mathematics) , radiation therapy , radiology , geology , psychology , psychotherapist , geodesy
Purpose: 4D treatment planning is often a time‐intensive process, requiring a series of procedures on multiple (up to 10) CT‐datasets. We hypothesize that more efficient strategies exist for lung cancer, 4D planning, and that dosimetrically, these methods are equivalent to plans based on the typical 4D process. Method and Materials: Three strategies were tested: (a) Exhale and Inhale CT‐datasets were registered using deformable image registration (ITK “demons”) and the resulting displacement vector field (DVF) was used to “synthesize” 8 additional CT datasets, at equally spaced phases between inhale and exhale. A single dataset, based on average CT‐number of the 10 datasets was then generated for planning purposes, CT AVE_SYN . (b) Ten CT‐datasets were retrospectively reconstructed during a patient 4D scan, and an average dataset was generated for planning purposes, CT AVE_4D . (c) Dose is computed on each of the ten datasets independently, dose accumulation is performed (trilinear interpolation), to yield a “warped” 4D, benchmark dataset, CT WARPED_4D . A step‐and‐shoot IMRT plan was developed and applied to the CT AVE_SYN , CT AVE_4D , and CT WARPED_4D datasets. Dose calculations were performed using Monte Carlo (BEAMnrc/DOSXYZnrc) integrated with Pinnacle. Doses were mapped to the reference image dataset (end‐exhale) for dosimetric comparison. Results: PTV and ITV mean doses generated using the CT AVE_SYN and CT AVE_4D datasets agreed within 0.5% of each other. Mean doses generated using CT WARPED_4D agreed with the CT AVE_SYN and CT AVE_4D datasets within 1% and 1.5% for the ITV and PTV, respectively. Biological dose indices, including EUD (ITV and PTV), and NTCP and mean lung dose (normal lung tissue) were within 1% agreement for treatment plans using all 3 datasets. Conclusion: Results are suggestive that 2 CT‐datasets used to synthesize an average 4D dataset (via a DVF and CT‐number averaging) may be a feasible approach, offering the potential for improved efficiency in the lung cancer 4D planning process. Acknowledgement_NIH‐R01CA106770