Program Information
Deformable Dose Mapping Accuracy Using a Novel Framework for User-Intervened Correction of Deformable Registration
AS Nelson1 , M Duchateau2*, JW Piper1 , D Verellen2 , M De Ridder2 , (1) MIM Software, Inc, Beachwood, OH, (2) Universitair Ziekenhuis Brussel, Brussels
Presentations
SU-E-J-216 Sunday 3:00PM - 6:00PM Room: Exhibit HallPurpose:
To evaluate the accuracy of deformable dose mapping using a method for user-intervened correction of deformable registration.
Methods:
Two treatment plans were calculated on a 4DCT dataset from the POPI model (Vandemeulebrouke 2011). POPI2 was chosen due to significant respiratory motion between end-inspiration to end-expiration (14.0mm +/- 7.2mm). Plans were created on the 0% phase for a mediastinal primary (MP) and a lung primary (LP). The 0% phase CT was deformed to the 50% phase CT and the dose was deformed using the same transformation. Doses were recorded at 100 corresponding points between the phases. Differences were calculated between the expected dose on the 50% phase and the deformed dose for all corresponding points. This process was repeated using an interactive tool to influence the deformation through local alignments. Seven local alignments were used to guide the deformation and dose differences were calculated.
Results:
The absolute % difference in dose (Gy) for MP before and after locally guided deformation was 5.6 +/- 10.6% and 1.7 +/- 2.5% respectively (p=0.0002). The percentage of points with < 2% dose difference were 59% and 74% respectively. The percentage of points with > 5% dose difference were 25% and 7% respectively. The absolute % difference in dose for LP before and after locally guided deformation was 10.3 +/- 22.4% and 1.6 +/- 2.5% respectively (p=0.0001). The percentage of points with < 2% dose difference were 47% and 65% respectively. The percentage of points with > 5% dose difference were 27% and 5% respectively.
Conclusion:
A novel method for user-intervened correction of deformable registration provided accurate results for dose mapping with average dose deviations < 1.7% even for a challenging subject with significant respiratory motion.
References:
1. Vandemeulebrouke et al. Med Phys 2011.
2. http://www.creatis.insa-lyon.fr/rio/popi-model/
Funding Support, Disclosures, and Conflict of Interest: AS Nelson is an employee with ownership at MIM Software, Inc. JW Piper is an employee with ownership at MIM Software, Inc.
Contact Email: