CyberKnife MLC-based treatment planning for abdominal and pelvic SBRT: Analysis of multiple dosimetric parameters, overall scoring index and clinical scoring

Phys Med. 2018 Dec:56:25-33. doi: 10.1016/j.ejmp.2018.11.012. Epub 2018 Nov 21.

Abstract

Purpose: This study evaluated the plan quality of CyberKnife MLC-based treatment planning in comparison to the Iris collimator for abdominal and pelvic SBRT. Multiple dosimetric parameters were considered together with a global scoring index validated by clinical scoring.

Methods and materials: Iris and MLC plans were created for 28 liver, 15 pancreas and 13 prostate cases including a wide range of PTV sizes (24-643 cm3). Plans were compared in terms of coverage, conformity (nCI), dose gradient (R50%), homogeneity (HI), OAR doses, PTV gEUD, MU, treatment time both estimated by TPS (tTPS) and measured. A global plan quality score index was calculated for IRIS and MLC solutions and validated by a clinical score given independently by two observers.

Results: Compared to Iris, MLC achieved equivalent coverage and conformity without compromising OAR sparing and improving R50% (p < 0.001). MLC gEUD was slightly lower than Iris (p < 0.05) for abdominal cases. MLC reduced significantly MU (-15%) and tTPS (-22%). Time reduction was partially lost when measured. The global score index was significantly higher for MLC solutions which were selected in 73% and 64% of cases respectively by the first and second observer.

Conclusion: Iris and MLC comparison was not straightforward when based on multiple dosimetric parameters. The use of a mathematical overall score index integrated with a clinical scoring was essential to confirm MLC plans advantages over Iris solutions.

Keywords: CyberKnife; MLC; Plan quality scoring; Robotic SBRT.

Publication types

  • Comparative Study

MeSH terms

  • Abdomen
  • Humans
  • Liver
  • Male
  • Pancreas
  • Pelvis
  • Prostate
  • Quality of Health Care
  • Radiometry
  • Radiosurgery* / methods
  • Radiotherapy Dosage
  • Radiotherapy Planning, Computer-Assisted* / methods