Program Information
Development of a Plug-in Based Image Analysis Tool for Integration Into Treatment Planning
D Owen*, C Anderson , C Mayo , I El Naqa , R Ten Haken , Y Cao , J Balter , M Matuszak , University of Michigan, Ann Arbor, MI
Presentations
SU-F-J-94 (Sunday, July 31, 2016) 3:00 PM - 6:00 PM Room: Exhibit Hall
Purpose:
To extend the functionality of a commercial treatment planning system (TPS) to support (i) direct use of quantitative image-based metrics within treatment plan optimization and (ii) evaluation of dose-functional volume relationships to assist in functional image adaptive radiotherapy.
Methods:
A script was written that interfaces with a commercial TPS via an Application Programming Interface (API). The script executes a program that performs dose-functional volume analyses. Written in C#, the script reads the dose grid and correlates it with image data on a voxel-by-voxel basis through API extensions that can access registration transforms. A user interface was designed through WinForms to input parameters and display results. To test the performance of this program, image- and dose-based metrics computed from perfusion SPECT images aligned to the treatment planning CT were generated, validated, and compared.
Results:
The integration of image analysis information was successfully implemented as a plug-in to a commercial TPS. Perfusion SPECT images were used to validate the calculation and display of image-based metrics as well as dose-intensity metrics and histograms for defined structures on the treatment planning CT. Various biological dose correction models, custom image-based metrics, dose-intensity computations, and dose-intensity histograms were applied to analyze the image-dose profile.
Conclusion:
It is possible to add image analysis features to commercial TPSs through custom scripting applications. A tool was developed to enable the evaluation of image-intensity-based metrics in the context of functional targeting and avoidance. In addition to providing dose-intensity metrics and histograms that can be easily extracted from a plan database and correlated with outcomes, the system can also be extended to a plug-in optimization system, which can directly use the computed metrics for optimization of post-treatment tumor or normal tissue response models.
Funding Support, Disclosures, and Conflict of Interest: Supported by NIH - P01 - CA059827
Contact Email: