Beam geometry generation: Reverse beam with fused opacity map

Loading...
Thumbnail Image

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Master's thesis

Department

Mcode

SCI3044

Language

en

Pages

74+12

Series

Abstract

Cancer is hard to cure and radiation therapy is one of the most popular treatment modalities. Even though the benefits of radiation therapy are undeniable, it still has possible side effects. To avoid severe side effects, with clinical evidence, delivering optimal radiation doses to patients is crucial. Intensity-modulated radiation therapy (IMRT) is an advanced radiation therapy technique and will be discussed in this thesis. One important step when creating an IMRT treatment plan is radiation beam geometry generation, which means choosing the number of radiation beams and their directions. The primary goal of this thesis was to find good gantry angles for IMRT plans by combing computer graphics and machine learning. To aid the plan generation process, a new method called reverse beam was introduced in this work. The new solution consists of two stages: angle discovery and angle selection. In the first stage, an algorithm based on the ray casting technique will be used to find all potential angles of the beams. For the second stage, with a predefined beam number, K-means clustering algorithm will be employed to select the gantry angles based on the clusters. The proposed method was tested against non-small cell lung cancer dataset from The Cancer Imaging Archive. By using IMRT plans with seven equidistant fields with 45-degree collimator rotations generated by the Ethos therapy system from Varian Medical Systems as a baseline for comparison, the plans generated by the reverse beam method illustrated good performance with the capability of avoiding organs while targeting tumors.

Description

Supervisor

Laaksonen, Jorma

Thesis advisor

Victor, Lopez

Other note

Citation