Segmentation of target organs and organs at risk is a fundamental task in radiotherapy treatment planning. Since its completion carried out by a radiation oncologist is really time-consuming, there is the need to perform it automatically. Unfortunately there is not a universal method capable to segment accurately every anatomical structure in every medical image, so each problem requires a study and an own solution. In this paper we analyze the problem of segmentation of bladder, prostate and rectum in lower abdomen CT images and propose a novel algorithm to solve it. It builds a statistical model of the organs analyzing a training set, generates potential solutions and chooses the segmentation result evaluating them on the basis of an aprioristic knowledge and the characteristics of patient image, using Genetic Algorithms. Out method has been tested qualitatively and quantitatively and offered good performance.
|Autori interni:||MASTRONARDI, Giuseppe|
|Titolo:||An Evolutionary Method for Model-Based Automatic Segmentation of Lower Abdomen CT Images for Radiotherapy Planning|
|Titolo del libro:||Applications of evolutionary computation : EvoApplications 2010, Istanbul, Turkey, April 7-9, 2010 : proceedings. Part I, EvoCOMPLEX, EvoGAMES, EvoIASP, EvoINTELLIGENCE, EvoNUM, and EvoSTOC|
|Data di pubblicazione:||2010|
|Digital Object Identifier (DOI):||10.1007/978-3-642-12239-2_33|
|Appare nelle tipologie:||2.1 Contributo in volume (Capitolo o Saggio)|