# Independent verification system for intracavitary brachytherapy based on a reference plan and statistical model

## TL;DR

This paper introduces a new verification system for brachytherapy treatments in cervical cancer, ensuring safety and accuracy across different equipment and techniques.

## Contribution

A novel verification system using reference plans and statistical models to detect errors in brachytherapy treatment planning.

## Key findings

- The system accurately estimated total dwell time with 0.26 ± 0.49% accuracy for Flexitron afterloader.
- The system successfully detected intentional human errors in treatment plans.
- The approach is feasible for new afterloaders, applicators, and hybrid brachytherapy techniques.

## Abstract

High dose rate (HDR) intracavitary brachytherapy (ICBT) with a remote afterloading system plays a vital role in the treatment of cervical cancer. We aimed to develop a new verification system for ICBT for cervical cancer and evaluate the feasibility for clinical plans (PlanClin) generated for different remote afterloaders, applicators and treatment techniques. In total, 517 PlansClin of patients were treated with Elekta 192Ir microSelectron HDR v2r. Reference plans (PlanRef) were generated for the ICBT applicators. An equation to predict total dwell time (Tdwell) of PlanClin was generated by evaluating the relationship between the volume receiving 100% of the prescribed dose (V100%) and the Tdwell. We also developed software to detect human errors in PlanClin by comparing parameters, including applicator and reference point geometries, dwell position and weight patterns and reference point dose, with those of PlanRef. Feasibility was evaluated for 83 PlanClin cases treated with the Elekta Flexitron remote afterloader and six ICBT plans with extra catheters (hybrid BT). The linear fitting function showed good agreement with the correlation between V100% and Tdwell. The developed equation accurately estimated the Tdwell of the PlanClin treated with the Flexitron with an accuracy of 0.26 ± 0.49%. Our system successfully detected intentional human errors including incorrect channel mapping, applicator tip offset, incorrect plan templates, an applicator digitization model and incorrect reference points. A verification system based on PlanRef and a statistical approach is feasible for the new remote afterloaders, applicators and hybrid BT techniques. This system contributes to the implementation of safe treatments.