A carregar...
Publicação
Two automated VMAT segmentation approaches : iCycleEclipse integration and a native SCP-based algorithm
| Nome: | Descrição: | Tamanho: | Formato: | |
|---|---|---|---|---|
| 14.31 MB | Adobe PDF |
Orientador(es)
Resumo(s)
Volumetric Modulated Arc Therapy (VMAT) is a modern technique in external radiotherapy that allows efficient and highly conformal dose delivery through continuous gantry rotation and dynamic modulation of beam intensity, dose rate, and multileaf collimator (MLC) positions. Despite its advantages, translating idealized fluence maps into clinically deliverable VMAT plans remains a challenging task. This thesis explores two complementary approaches to optimize VMAT segmentation from a Beam’s Eye View (BEV) perspective, aiming to preserve the dosimetric quality of Erasmus-iCycle plans while ensuring compatibility with clinical treatment planning systems (TPS) and linear accelerators (LINACs). The first approach involves integrating Erasmus-iCycle with the Eclipse TPS. Ten prostate cancer cases were analyzed by converting iCycle-optimized fluence plans into Eclipse-based VMAT formats. Dose-volume histogram (DVH) analysis was used to compare target coverage (V95%, D98%), dose homogeneity (Dmean, D2%), and organ-at-risk (OAR) sparing. While Eclipse preserved the general dosimetric quality, minor trade-offs in heterogeneity and rectal dose were observed, suggesting the need for refined optimization settings. The second approach introduces a native segmentation algorithm based on Sequential Convex Programming (SCP), designed to emulate the Erasmus-iCycle dose distribution using direct aperture optimization. Implemented in MATLAB, the SCP method was first validated on a test case and then applied across the patient cohort. It produced plans with strong agreement to reference distributions, offering greater flexibility and allowing the inclusion of constraints to enhance deliverability and mechanical stability. Overall, this work demonstrates the clinical feasibility and value of BEV-driven segmentation strategies, providing a foundation for automated, customizable VMAT planning that can be extended to other treatment sites and advanced radiotherapy techniques.
Descrição
Tese de Mestrado, Engenharia Biomédica e Biofísica, 2025, Universidade de Lisboa, Faculdade de Ciências
Palavras-chave
VMAT Erasmus-iCycle Treatment Planning Optimization BEV Segmentation SCP