Using Overlap Volume Histogram Analysis of a Prior Plan Dataset to Generate Clinically Acceptable Plans for CyberKnife Robotic Radiosurgery Treatment of Localized Prostate Cancer
B Wu1*, D Pang1, J Gatti1, S Lei1, S Colin2, T McNutt3, T Kole1, S.P Collins1, A Dritschilo1, (1) Georgetown University Hospital (2) Accuray Inc (3) Johns Hopkins UniversitySU-E-T-670 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: CyberKnife offers the potential benefits of non-isocentric, non-coplanar treatment delivery. However, its planning is a laborious manual, trial-and-error process. This study is to investigate whether an overlap volume histogram (OVH)-driven planning approach can produce clinically acceptable plans for treatment of localized prostate cancer.
Methods: It is assumed that given consistent target coverage, patients with a relevant target-organ spatial relationship should have similar organ sparing. The OVH is used to characterize the 3-D spatial relationship between an organ and a target. A database containing the OVH and DVH of prior plans is built to serve as an external reference. During the initial planning, the OVH is used to search through the database to find a prior patient group whose target-organ relationship is related to that of a new patient. The planning objectives for the new patient are then estimated from the group and input into the CyberKnife TPS for optimization. To demonstrate the effectiveness of the method, the plans of 12 prostate patients (prescription: 36.25Gy in 5 fractions) are generated by the OVH approach and compared to the corresponding clinical plans using the in-house dosimetric guidelines.
Results: Physicians confirm that OVH plans are clinically acceptable. OVH plans: on average, Vcc(37Gy) and V(18.12Gy) to the bladder decrease 1cc and 6% (p<0.05); Vcc(36Gy) and V(18.12Gy) to the rectum decrease 0.02cc and 4.3% (p=0.3); Vcc(40Gy) to the prostatic urethra decreases 0.08cc (p=0.01); V(14.5Gy) to the femur heads decreases 0.58% (p=0.01). V(37Gy) to the membranous urethra increases from 23.1% to 24.5% (p=0.75); V(29.5Gy) to the penile bulb increases from 5.5% to 10.7% (p=0.35); V(36.25Gy) to PTV increases 0.2% (p=0.4); the estimated delivery time deceases 3.5 minutes (p=0.01).
Conclusion: With respect to planner-created clinical plans, this approach offers an alternative way to generate clinically acceptable plans. It advances the possibility of automated CyberKnife planning.
Funding Support, Disclosures, and Conflict of Interest: C. Sims, employed by Accuray, Inc;S.P Collins, Accuray clinical consultant