Xing Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Xing Li has a total of 11 co-authors. Here are the top co-authors by number of shared publications:
Ann Miner
Department of Advanced Radiation and Proton Therapy INOVA Schar Cancer Institute Fairfax Virginia USA
1 shared publicationAshish Chawla
Department of Advanced Radiation and Proton Therapy INOVA Schar Cancer Institute Fairfax Virginia USA
1 shared publicationFan-Chi Su
1 shared publicationFan‐Chi Su
Department of Radiation Oncology University of Texas Southwestern Medical Center Dallas Texas USA
1 shared publicationJiajin Fan
Department of Advanced Radiation and Proton Therapy INOVA Schar Cancer Institute Fairfax Virginia USA
1 shared publicationJingzhu Xu
1 shared publication
Latest Publications
- Open Access
- 1 March 2026
- Published by AME Publishing Company in Therapeutic Radiology and Oncology
- Vol. 10, 6
Evaluating the impact of volume averaging in pencil beam scanning proton therapy
Background: Pencil beam scanning (PBS) proton therapy delivers therapeutic doses with exceptional precision through its unique spot-by-spot delivery technique. However, the resulting composite dose profiles—dependent on spot size and spacing—introduce volume averaging that complicates accurate dose determination in treatment planning system (TPS), a critical step in beam commissioning. In this study, we aim to quantify the clinical impact of volume averaging in TPS for PBS proton therapy and establish an optimal dose acquisition strategy in TPS to minimize volume averaging effect that facilitates accurate and safe patient treatment. Methods: Treatment plans were designed within RayStation TPS using PBS proton beams of 70, 150 and 225 MeV with spot spacings from 0.18 to 0.6 cm. The plans incorporated a PPC05 parallel-plate (PP) chamber in a virtual water phantom. Dose density, defined as dose per unit area, was introduced to quantify the impact of spot size and spacing on volume averaging. Following beam configuration guidelines in the TPS, the chamber was placed at various depths. Volume doses () were obtained from a contoured volume of the chamber’s active measurement element, and point doses () were obtained from dose specification point (DSP) at the inner surface of the chamber’s entrance window along the central axis. Subsequently, the plans were also delivered in a water phantom on an IBA ProteusPlus PBS machine, and the measured doses in the phantom, , were compared with the TPS values of and to assess volume averaging. Results: Significant discrepancies were observed between and at lower dose densities. The maximum deviations occurred at 0.1 cGy·cm−2, where was higher than by +4.5% for 70 MeV, +2.2% for 150 MeV, and +4.3% for 225 MeV. The in-phantom verifications showed similar results. Overall, point doses in TPS showed better agreement with phantom verification than volumetric doses. At vendor-recommended commissioning condition (0.4 cGy·cm−2), all discrepancies fell below ±1%. Conclusions: We recommend using point-based dose acquisitions in TPS to reduce volume averaging in PBS proton therapy. Furthermore, adherence to vendor-recommended commissioning settings in beam configuration minimized the deviations between volumetric and point doses observed in this study.- 1 March 2026
- Published by AME Publishing Company in Therapeutic Radiology and Oncology
- Vol. 10, 6
Open AccessEvaluating the impact of volume averaging in pencil beam scanning proton therapy
Background: Pencil beam scanning (PBS) proton therapy delivers therapeutic doses with exceptional precision through its unique spot-by-spot delivery technique. However, the resulting composite dose profiles—dependent on spot size and spacing—introduce volume averaging that complicates accurate dose determination in treatment planning system (TPS), a critical step in beam commissioning. In this study, we aim to quantify the clinical impact of volume averaging in TPS for PBS proton therapy and establish an optimal dose acquisition strategy in TPS to minimize volume averaging effect that facilitates accurate and safe patient treatment. Methods: Treatment plans were designed within RayStation TPS using PBS proton beams of 70, 150 and 225 MeV with spot spacings from 0.18 to 0.6 cm. The plans incorporated a PPC05 parallel-plate (PP) chamber in a virtual water phantom. Dose density, defined as dose per unit area, was introduced to quantify the impact of spot size and spacing on volume averaging. Following beam configuration guidelines in the TPS, the chamber was placed at various depths. Volume doses () were obtained from a contoured volume of the chamber’s active measurement element, and point doses () were obtained from dose specification point (DSP) at the inner surface of the chamber’s entrance window along the central axis. Subsequently, the plans were also delivered in a water phantom on an IBA ProteusPlus PBS machine, and the measured doses in the phantom, , were compared with the TPS values of and to assess volume averaging. Results: Significant discrepancies were observed between and at lower dose densities. The maximum deviations occurred at 0.1 cGy·cm−2, where was higher than by +4.5% for 70 MeV, +2.2% for 150 MeV, and +4.3% for 225 MeV. The in-phantom verifications showed similar results. Overall, point doses in TPS showed better agreement with phantom verification than volumetric doses. At vendor-recommended commissioning condition (0.4 cGy·cm−2), all discrepancies fell below ±1%. Conclusions: We recommend using point-based dose acquisitions in TPS to reduce volume averaging in PBS proton therapy. Furthermore, adherence to vendor-recommended commissioning settings in beam configuration minimized the deviations between volumetric and point doses observed in this study. - Open Access
- 18 November 2025
- Published by Wiley in Journal of Applied Clinical Medical Physics
- Vol. 26 (12) , e70369
Radiation incident learning in multi‐site centers
Purpose: Incident learning in large, multi‐site hospital systems faces challenges from resource constraints, diverse site cultures, and deviations from system‐wide policies. This study introduces a novel, resource‐efficient hybrid approach to facilitate cross‐site learning and proactive risk management, ultimately enhancing patient safety. Methods: A hybrid approach, integrating statistical, and root cause analyses was used to analyze safety events recorded from 2023 to 2024 at five sites. Events were categorized by time, site, type, and harm score with statistics tracking the event frequency per category. Failure modes (FMs) were identified with summative keywords. The frequency and cross‐site correlation of keywords were innovatively displayed by word clouds. The hybrid approach was exemplified at a high‐throughput satellite clinic, where treatments and incidents were tracked quarterly. A proactive model was also employed to predict the number of potentially affected patients at this site. Results: An analysis of 228 reported events revealed that most incidents occurred within the treatment planning category, where the keywords “carepath” and “contour” were identified as the top two frequently reported FMs, repeated 13 and 10 times respectively. The keywords “contour” and “documentation” appeared across four sites, followed by “carepath” and “couch” which appeared across three sites. At the featured satellite clinic, a total of 18 FMs were identified from yearly 8,714 treatments. The proactive model showed an average risk priority number decreased from 106 to 77, and the estimated number of affected patients decreased from 27 to 20 at this site. Conclusions: We developed a comprehensive, system‐wide approach that enhanced efficiency in incident learning across multi‐site clinics. Site‐specific shifts in safety culture were effectively benchmarked and cross‐site learning was facilitated through measurable endpoints. A novel word cloud analysis was developed, enhancing visualization of cross‐site root causes. Additionally, a proactive model was adopted to increase system‐wide preparedness and improve patient safety outcomes.
- 18 November 2025
- Published by Wiley in Journal of Applied Clinical Medical Physics
- Vol. 26 (12) , e70369
Open AccessRadiation incident learning in multi‐site centers
Purpose: Incident learning in large, multi‐site hospital systems faces challenges from resource constraints, diverse site cultures, and deviations from system‐wide policies. This study introduces a novel, resource‐efficient hybrid approach to facilitate cross‐site learning and proactive risk management, ultimately enhancing patient safety. Methods: A hybrid approach, integrating statistical, and root cause analyses was used to analyze safety events recorded from 2023 to 2024 at five sites. Events were categorized by time, site, type, and harm score with statistics tracking the event frequency per category. Failure modes (FMs) were identified with summative keywords. The frequency and cross‐site correlation of keywords were innovatively displayed by word clouds. The hybrid approach was exemplified at a high‐throughput satellite clinic, where treatments and incidents were tracked quarterly. A proactive model was also employed to predict the number of potentially affected patients at this site. Results: An analysis of 228 reported events revealed that most incidents occurred within the treatment planning category, where the keywords “carepath” and “contour” were identified as the top two frequently reported FMs, repeated 13 and 10 times respectively. The keywords “contour” and “documentation” appeared across four sites, followed by “carepath” and “couch” which appeared across three sites. At the featured satellite clinic, a total of 18 FMs were identified from yearly 8,714 treatments. The proactive model showed an average risk priority number decreased from 106 to 77, and the estimated number of affected patients decreased from 27 to 20 at this site. Conclusions: We developed a comprehensive, system‐wide approach that enhanced efficiency in incident learning across multi‐site clinics. Site‐specific shifts in safety culture were effectively benchmarked and cross‐site learning was facilitated through measurable endpoints. A novel word cloud analysis was developed, enhancing visualization of cross‐site root causes. Additionally, a proactive model was adopted to increase system‐wide preparedness and improve patient safety outcomes.