A Pilot Study for Piecewise Uniform Dose Prescription and Optimization Method Based on PET/CT Images

Purpose: During intensity modulated radiation therapy (IMRT) planning, the tumor target volume is commonly given a uniform dose prescription (UDP) clinically, which does not consider the heterogeneous characteristics of tumor such as hypoxia, clonogenic cell density, tumor proliferation rate and radio-sensitivity. To better spare organs at risk (OARs) while not decrease the tumor control probability (TCP), we propose to use a piecewise uniform dose prescription (PUDP) method, which is based on the heterogeneous characteristics of tumor extracted from its PET/CT medical images. Methods: First, we automatically delineate several sub-gross target volumes (sub-GTVs) inside the gross tumor volume (GTV) by a hierarchical Mumford-Shah vector model, and make each sub-GTV is of almost homogeneous [18F] fluorodeoxyglucose positron emission tomography (FDG-PET) standard uptake value (PETSUV), PET-contrast, and PET-busyness of tumors. Then, we make the GTV mean dose minimized while restricting the TCP to a constant and derive an optimal redistribution of dose prescription based on a linear-quadratic (LQ) model. Finally, we use the equivalent uniform dose (EUD) in each sub-GTV as its final dose prescription, which composes a PUDP for the GTV. Results: We have evaluated the IMRT planning for 9 patients with nasopharyngeal carcinoma respectively using PUDP and clinical UDP. The results show that almost all OARs are better spared by PUDP than by UDP. Although PUDP and UDP have the similar TCPs for GTV, PUDP can reduce the max dose to brain stem by $8.53\pm 7.71$Gy and the mean dose by $2.86\pm 1.30\mathrm{G}\mathrm{y}$. The mean normal tissue complication probability (NTCP) of the brain stems for 9 patients is reduced from $3.55\times 10^{-6}$ (UDP) to $1.17\times 10^{-6}$ (PUDP). Conclusion: Comparing with the UDP, the PUDP can better spare almost all OARs while keeping the similar TCP of GTV.