Author: Dante PI Capaldi, Lu Jiang, Qihui Lyu, Ke Sheng 👨🔬
Affiliation: Department of Radiation Oncology, University of California at San Francisco, Department of Radiation Oncology, University of California, San Francisco 🌍
Purpose:
Preclinical small animal studies help understand radiation-induced biological responses, toxicities, and mechanisms, facilitating the translation of new therapies to patient treatment. Intensity-modulated radiotherapy(IMRT) has been developed for preclinical systems to more closely mimic human radiotherapy. However, current approaches rely on fixed beam angles not optimized for dosimetry. This study introduces a novel 4π non-coplanar beam orientation optimization framework with rectangular apertures for the sparse orthogonal collimator(SOC) to achieve superior preclinical IMRT.
Methods:
We developed an integrated 4π rectangular aperture-based beam orientation optimization(4π-RA-BOO) framework for small animal IMRT, which includes an L2-norm term for dose fidelity, an L1-norm term for aperture sparsity, and an L2,1/2 group sparsity term for beam selection. Computational efficiency was improved by replacing the Kronecker product of dose-loading matrices with smaller matrix multiplications. Solving the optimization problem was further accelerated using the Fast Iterative Shrinkage-Thresholding Algorithm(FISTA). The optimal number of beams was determined by the tradeoff between converged dose fidelity values and the number of beams. Three preclinical cases(brain, liver tumor, and spine) involving five mice were analyzed and compared with static 5-beam Rectangular Aperture Optimization(5F-RAO) and Fixed-Size collimator plans(5F-FS). Plan evaluation metrics included D2%, D98%, homogeneity index for planning target volumes(PTVs), and mean and maximum doses for organs at risk(OARs).
Results:
For all treatment sites, 4π-BOO delivered superior plan quality by reducing hot spots and maintaining target coverage. In spine cases, hot spots were reduced up to 13% compared to 5F-FS. Mean doses to the lungs, kidneys, and liver were reduced by 7%, 3%, and 40% compared to 5F-RAO; and by 52%, 60% and 46% compared to 5F-FS, respectively. 4π-BOO plans require 2-3X more apertures compared to 5F-RAO.
Conclusion:
Incorporating beam orientation optimization in small animal IMRT allows dosimetrically superior plans that more closely mimic human IMRT quality in target dose homogeneity and OAR sparing.