Cite this article as: Song T, Tian Z, Jia X, Zhou L, Jiang SB, Gu X. A plan quality classifier derived with overlap-wall-histogram of hollow organs for automatic IMRT plan quality control of prostate cancer cases. Int J Cancer Ther Oncol 2014; 2(2):020241.
DOI: http://dx.doi.org/10.14319/ijcto.0202.41

Innovation/Impact: We regard this work as innovative because it proposed a new descriptor called overlap-wall- histogram (OWH) (Figure 1) to describe the spatial relationship between PTV and a hollow organ, and two metrics to qualitatively evaluate patient geometry difficulty for planning and plan quality in terms of OAR sparing. Besides, a model was built to establish the boundary of poor quality and high quality plans (e.g. the blue lines in Figure 2) after studying from a training database of prostate cancer IMRT cases with acceptable plan quality. Thus contributes a plan quality classifier to assess IMRT plan quality of prostate cancer patients for automatic plan quality control. A query plan falling above the boundary is assessed as high quality, vice versa poor quality. Furthermore, for a query plan with poor quality, potential improvement of plan quality can be estimated and a good reference plan with similar or harder geometry can be automatically chosen from our database to help guide the re-planning if necessary.

Key Results: 15 prostate IMRT plans were used to test our model. Their distribution in our IMRT plan quality map is shown in Figure2. One was identified as poor quality and the others were common-level. After re-planning all plans, the dose constraints for bladder wall W75 (the percentage of wall receiving more than 75Gy),W₇₀,W₆₅ and W₆₀ can be reduced by 3.34%, 3%, 6.99%, 6.54% for that poor quality plan and 1.11%, 0.95%, 1.45% and 1.81% averagely for the  common-level quality group, without sacrificing PTV coverage and rectum dose sparing(Figure 3).

FIG.1: An OWH example. Shape of a target and OAR (above). OWH (below)

FIG. 2: Established boundary of poor quality and high quality plans studied from a training database

FIG. 3: An example of re-planning improvement for a poor quality plan (original plan- solid line, after re-planning- dashed line)