Encrypted login | home

Program Information

An Overlap-Wall-Histogram Plan Quality Classifier for Automatic Prostate Cancer IMRT Treatment Plan Quality Control

no image available
T Song

T Song1,2*, Z Tian1 , X Jia1 , L Zhou2 , S Jiang1 , X Gu1 , (1) UT Southwestern Medical Center, Dallas, TX, (2) Southern Medical University, Guangzhou, China


SU-E-T-38 Sunday 3:00PM - 6:00PM Room: Exhibit Hall

Purpose:A new patient geometry descriptor, PTV and a hollow organ overlapping wall histogram (OWH), is proposed to combine with dose to wall histogram (DWH) to develop a plan quality classification model for automatic prostate IMRT plan quality assessment and improvement.

Methods:In this study, we define a geometric metric named COP (closeness of an OAR to PTV) to quantify the difficulty of patient geometry for planning and a dosimetric metric named DSO (dose spillage over an OAR) to evaluate treatment plan quality in terms of OAR dose sparing. COP and DSO are defined as the weighted sum of OWH and DWH respectively. Fifteen clinically acceptable prostate treatment plans were selected to study the correlation between COP and DSO. After plotting COP as X axis and DSO as Y axis, a linear correlation was established and the entire space is classified into high and poor plan quality regions by a linear fitting with these COP-DSO pairs. This proposed classifier is evaluated with other 25 prostate plans. Any evaluation plan falling into the poor plan quality region will be re-planned, and the re-planning can be guided by a selected reference plan having similar COP but lower DSO values. Re-planned high-quality plans will be added to database, affording more options during reference plan selection.

Results:Using our classification model, 8 out of 25 plans are identified as poor quality plans. After re-planning, the dose volume for bladder wall W₇₅(wall receiving more than 75Gy),W₇₀,W₆₅ and W₆₀ can be reduced by 3.34%, 3%, 6.99%, 6.54% for the worst quality plan and 1.11%, 0.95%, 1.45% and 1.81% averaged over 8 plans, without sacrificing PTV coverage and rectum dose sparing.

Conclusion:Our proposed model is capable to identify poor quality plans and find a specific reference plan to guide re-planning, thus improve plan quality.

Contact Email: