Knowledge-based IMRT treatment planning for prostate cancer

Authors

  • Chanyavanich Vorakarn,

    1. Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University Medical Center, Durham, North Carolina 27705 and Medical Physics Graduate Program, Duke University Medical Center, Durham, North Carolina 27705
    Search for more papers by this author
    • a)

      Author to whom correspondence should be addressed. Electronic mail: vc17@duke.edu; Telephone: (919) 684-1440; Fax: (919) 684-1491.

  • Das Shiva K.,

    1. Medical Physics Graduate Program, Duke University Medical Center, Durham, North Carolina 27705 and Department of Radiation Oncology, Duke University Medical Center, Durham, North Carolina 27710
    Search for more papers by this author
  • Lee William R.,

    1. Department of Radiation Oncology, Duke University Medical Center, Durham, North Carolina 27710
    Search for more papers by this author
  • Lo Joseph Y.

    1. Carl E. Ravin Advanced Imaging Laboratories, Department of Radiology, Duke University Medical Center, Durham, North Carolina 27705; Medical Physics Graduate Program, Duke University Medical Center, Durham, North Carolina 27705; and Department of Biomedical Engineering, Duke University, Durham, North Carolina 27708
    Search for more papers by this author

Abstract

Purpose:

To demonstrate the feasibility of using a knowledge base of prior treatment plans to generate new prostate intensity modulated radiation therapy (IMRT) plans. Each new case would be matched against others in the knowledge base. Once the best match is identified, that clinically approved plan is used to generate the new plan.

Methods:

A database of 100 prostate IMRT treatment plans was assembled into an information-theoretic system. An algorithm based on mutual information was implemented to identify similar patient cases by matching 2D beam's eye view projections of contours. Ten randomly selected query cases were each matched with the most similar case from the database of prior clinically approved plans. Treatment parameters from the matched case were used to develop new treatment plans. A comparison of the differences in the dose–volume histograms between the new and the original treatment plans were analyzed.

Results:

On average, the new knowledge-based plan is capable of achieving very comparable planning target volume coverage as the original plan, to within 2% as evaluated for D98, D95, and D1. Similarly, the dose to the rectum and dose to the bladder are also comparable to the original plan. For the rectum, the mean and standard deviation of the dose percentage differences for D20, D30, and D50 are 1.8% ± 8.5%, −2.5% ± 13.9%, and −13.9% ± 23.6%, respectively. For the bladder, the mean and standard deviation of the dose percentage differences for D20, D30, and D50 are −5.9% ± 10.8%, −12.2% ± 14.6%, and −24.9% ± 21.2%, respectively. A negative percentage difference indicates that the new plan has greater dose sparing as compared to the original plan.

Conclusions:

The authors demonstrate a knowledge-based approach of using prior clinically approved treatment plans to generate clinically acceptable treatment plans of high quality. This semiautomated approach has the potential to improve the efficiency of the treatment planning process while ensuring that high quality plans are developed.

Ancillary