Clinical validation of a graphical method for radiation therapy plan quality assessment

Area of Science:

• Radiation Oncology
• Medical Physics
• Clinical Informatics

Background:

• Developing tools for objective radiotherapy treatment plan assessment is crucial.
• Radiation oncologists' (ROs) clinical preferences are essential for treatment plan selection.
• SPIDERplan is a graphical tool designed for treatment plan assessment.

Purpose of the Study:

• To clinically validate the SPIDERplan tool by comparing its scoring-based plan choices with ROs' clinical preferences.
• To assess SPIDERplan's performance in nasopharynx cancer treatment planning.

Main Methods:

• Validation involved three ROs from three departments evaluating 20 pairs of nasopharynx treatment plans.
• ROs blindly ranked plans and assigned qualitative evaluations ('Good', 'Admissible', 'Not Admissible').
• SPIDERplan was applied using local and international structure groups, with weights derived from RO preferences via mixed linear programming.

Main Results:

• Two-thirds of plans were rated 'Good' by ROs.
• Fair to substantial agreement was observed in ROs' plan selection and evaluation.
• SPIDERplan's plan choices largely agreed with ROs, especially when score differences were significant (threshold 0.005).

Conclusions:

• SPIDERplan successfully reproduced ROs' treatment plan selections for nasopharynx cases.
• The tool's performance is robust across different configurations of structure groups and weight assignments.
• SPIDERplan demonstrates clinical validity and can support ROs in treatment planning decisions.