Treatment plan quality control using multivariate control charts | JoVE Visualize

Area of Science:

Radiotherapy Quality Assurance
Medical Physics
Statistical Process Control

Background:

The American Association of Physicists in Medicine (AAPM) Task Group 218 recommends statistical process control (SPC) tools like control charts for radiotherapy quality assurance.
Existing SPC tools lack the ability to analyze multivariate, correlated data common in radiotherapy treatment plan quality measures.
There is a need for advanced quality control tools that can model complex plan data and account for patient-specific factors without disrupting clinical workflows.

Purpose of the Study:

To develop novel quality control tools for radiotherapy that can model multivariate plan quality measures with correlations.
To incorporate patient-specific risk factors into the quality control process.
To create a system that enhances radiotherapy quality assurance without adding significant burden to clinical workflow.

Main Methods:

Developed a multivariate quality control chart incorporating risk-adjustment, Hotelling's T2 statistic, and Principal Component Analysis (PCA).
PCA was used to manage correlations among organ-at-risk (OAR) dose-volume histogram (DVH) points.
Risk-adjustment models estimated principal components using patient- and treatment-specific factors; residuals informed the Hotelling's T2 statistic and control chart plotting.
Box-Cox transformation addressed non-normality in DVH points; three charts were evaluated: conventional, risk-adjusted, and PCA-based risk-adjusted.

Main Results:

Conventional charts generated numerous false alarms due to ignoring patient-specific risk factors and correlations.
Risk-adjusted charts reduced false alarms but did not account for DVH point correlations.
The PCA-based, risk-adjusted control chart effectively identified unusual plans by accounting for correlations.
Re-planning based on identified unusual plans demonstrated measurable improvements.

Conclusions:

The developed multivariate risk-adjusted control chart facilitates pre-delivery quality control of radiotherapy plans.
This methodology is adaptable and can be applied to other radiotherapy quality assurance protocols, such as gamma analysis pass rates.
The approach enhances the accuracy and efficiency of radiotherapy quality assurance by addressing complex data characteristics.