Encrypted login | home

Program Information

Evaluation of Tumor Hypoxic Fraction Using Serial Volumetric Imaging During Radiation Therapy


A Chvetsov

A Chvetsov, University of Washington, Seattle, Washington

Presentations

SU-F-T-685 (Sunday, July 31, 2016) 3:00 PM - 6:00 PM Room: Exhibit Hall


Purpose: To develop a tumor response model which could be uses to compute tumor hypoxic fraction using serial volumetric tumor imaging. This algorithm may be used for treatment response assessment and also for guidance of more expensive PET imaging of hypoxia.

Methods: Previously developed two-level cell population tumor response model was modified to include a third cell level describing hypoxic and necrotic cells. This third level was considered constant value during radiotherapy treatment; therefore, inclusion additional parameter did not compromise stability of model fitting to imaging data. Fitting the model to serial volumetric imaging data was performed using a least squares objective function and simulated annealing algorithm. The problem of reconstruction of radiobiological parameters from serial imaging data was considered as inverse ill-posed problem described by the Fredholm integral equation of the first kind. Variational regularization was used to stabilize solutions.

Results: To evaluate performance of the algorithm, we used a set of serial CT imaging data on tumor-volume for 14 head and neck cancer patients. The hypoxic fractions were reconstructed for each patient and the distribution of hypoxic fractions was compared to the distribution of initial hypoxic fractions previously measured using histograph. The measured and reconstructed from imaging data distributions of hypoxic fractions are in good agreement. The reconstructed distribution of cell surviving fraction was also in better agreement with in vitro data than previously obtained using the two-level cell population model.

Conclusion:Our results indicate that it is possible to evaluate the initial hypoxic tumor fraction using serial volumetric imaging and a tumor response model. This algorithm can be used for treatment response assessment and guidance of more expensive PET imaging.


Contact Email: