Workflow Optimization In Cervical Cancer Radiotherapy Planning Using AI/Scripts In Middle Income Countries
Abstract
Purpose
The purpose of this work was to optimize the radiotherapy treatment planning workflow for cervical cancer using artificial intelligence-based tools and optimization trough scripts, with the goal of reducing planning time and improving the implementation of clinical protocols in a middle-income country.
Methods
CT images acquired with intracavitary brachytherapy applicators in place, including fletcher, ring, cylinder or central tandems, were imported into Eclipse (v18) for treatment planning of cervical cancer patients. An AI-assisted and script-based workflow was implemented for cervical cancer external beam radiotherapy, focusing on automated structure generation and standardized plan initialization and dose calculation using the RapidPlan model, where is was also adapted for SIB and non-SIB treatment techniques. Brachytherapy treatment plans were subsequently generated and evaluated. The combined external beam and brachytherapy dose distributions were assessed using Radformation tools by evaluating the biologically effective dose (BED) to the target volumes and compliance with dose limits for organs at risk. Planning efficiency and workflow performance were evaluated by comparing planning time and manual interactions with the conventional manual planning process.
Results
The AI-assisted workflow along with the scripts reduced median treatment planning time compared to manual planning. Plan to plan variability was reduced, while target coverage and OAR dose constraints remained clinically acceptable and comparable to the conventional workflow.
Conclusion
The use of scripting and artificial intelligence resulted in improved efficiency in structure generation, treatment planning, and plan approval for cervical cancer brachytherapy. When integrated with prior external beam radiotherapy planning, the proposed workflow demonstrated reduced planning time while maintaining clinically acceptable plan quality, supporting its applicability in resource-limited clinical settings.