11. Sensitivity of patient specific quality assurance to simulated delivery errors for CyberKnife MLC treatments and effects on DVH. (December 2018)

• Record Type: Journal Article
• Authors: Zani, M.; Calusi, S.; Doro, R.; Bellosi, N.; Cassinelli, M.; Pazzaglini, S.; Tempobono, M.; Masi, L.
• Is Part Of: Physica medica. Volume 56(2018)Supplement 2
• Page Start: 67
• Page End: 68
• Publication Date: 2018-12
• Subjects: Medical physics -- Periodicals; Biophysics -- Periodicals; Biophysics -- Periodicals; Imagerie médicale -- Périodiques; Radiothérapie -- Périodiques; Rayons X -- Sécurité -- Mesures -- Périodiques; Physique -- Périodiques; Médecine -- Périodiques; 610.153
• DOI: 10.1016/j.ejmp.2018.04.021
• Languages: English
• ISSNs: 1120-1797

Abstract

Abstract : Purpose: Accurate and reliable patient-specific QA is required for CyberKnife (CK) MLC treatment plans. Films are the current method of choice, but filmless solutions are starting to be introduced in the routine[1] . Aim of this study is to test the sensitivity of a filmless QA technique to simulated delivery errors for CK MLC plans, and to discuss which γ -index criteria are adequate; DVH implications were also analyzed. Methods: Preliminarily, routine garden fence tests were employed to assess MLC leaf positioning accuracy. Test plans were then delivered to Octavius 1000SRS array (PTW) and EBT3 (Ashland), editing leaf positions of five clinical plans (prostate/pancreas) with three error types: MLC banks opened, both of 0.3 mm (03 mm OUT)/0.5 mm (05 mm OUT), one of 1 mm (1 mm X2)), closed (both of 0.5 mm (05 mm IN)) and shifted (both of 1 mm (1 mm X1X2)). Γ-index analysis was performed comparing the measured plans to the unedited dose distributions using three criteria: 2%2 mm, 3%1 mm, 2%1 mm. For patient plans, DVH implications were also considered. Results: The garden fence leaf positioning distribution mean over 11578 values is 0.07 ± 0.16 mm. The 95% percentile is 0.34 mm, with 141 values (1.2%) outside the 0.5 mm tolerance. In Fig. 1 1000SRS passing rates for the edited and the unedited plans are reported for two cases. For all plans, the highest influence on pass-rate is found for increased segments size. With both banks opened of 0.3 mm a mean passing rate decrease from 94.0% to 80.2% was observed (3%1 mm). EBT3 show analogous results. InTable 1 the simulated errors effect on DVH parameters is reported for Plan2. A 0.3 mm opening produced a total of 7 OAR constraint violations over all plans. Conclusions: High sensitivity in the detection of small delivery errors, slightly affecting DVHs, is achievable for CK MLC plans delivery QA even employing filmless solutions. A 2%2 mm criteria is not always tight enough.