Cross-institutional knowledge-based planning (KBP) implementation and its performance comparison to Auto-Planning Engine (APE)

B. Wu, M. Kusters, M. Kunze-Busch, T. Dijkema, T. McNutt, G. Sanguineti, K. Bzdusek, A. Dritschilo, D. Pang

Research output: Contribution to journalArticlepeer-review


BACKGROUND AND PURPOSE: To investigate (1) whether a plan library established at one institution can be applied for another institution's knowledge-based planning (KBP); (2) the performance of cross-institutional KBP compared to Auto-Planning Engine (APE). MATERIAL AND METHODS: Radboud University Medical Center (RUMC) provided 35 oropharyngeal cancer patients (68Gy to PTV68 and 50.3Gy to PTV50.3) with clinically-delivered and comparative APE plans. The Johns Hopkins University (JHU) contributed a three-dose-level plan library consisting of 179 clinically-delivered plans. MedStar Georgetown University Hospital (MGUH) contributed a KBP approach employing overlap-volume histogram (OVH-KBP), where the JHU library was used for guiding RUMC patients' KBP. Since clinical protocols adopted at RUMC and JHU are different and both approaches require protocol-specific planning parameters as initial input, 10 randomly selected patients from RUMC were set aside for deriving them. The finalized parameters were applied to the remaining 25 patients for OVH-KBP and APE plan generation. A Wilcoxon rank-sum test was used for statistical comparison. RESULTS: PTV68 and PTV50.3's V95 in OVH-KBP and APE were similar (p>0.36). Cord's D0.1 cc in OVH-KBP was reduced by 5.1Gy (p=0.0001); doses to other organs were similar (p>0.2). CONCLUSION: APE and OVH-KBP's plan quality is comparable. Institutional-protocol differences can be addressed to allow cross-institutional library sharing.
Original languageEnglish
Pages (from-to)57-62
Number of pages6
JournalRadiotherapy and Oncology
Issue number1
Publication statusPublished - Apr 1 2017


  • APE
  • IMRT
  • KBP
  • OVH


Dive into the research topics of 'Cross-institutional knowledge-based planning (KBP) implementation and its performance comparison to Auto-Planning Engine (APE)'. Together they form a unique fingerprint.

Cite this