Algoritmo de controle do dano ao tecido saudável no tratamento de câncer por hipertermia
Loading...
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Universidade Federal de Juiz de Fora (UFJF)
DOI
Abstract
According to the World Health Organization, cancer is a global health concern. The high
mortality rate drives the scientific community to study new treatments, among which we
can mention hyperthermia by magnetic nanoparticles. This treatment involves subjecting
the target region to a low-frequency magnetic field, causing the nanoparticles to increase
the temperature above 43◦C, considered the target temperature to damage the tissue
and induce cell necrosis. This work uses the three-dimensional in silico model of Pennes,
described by a partial differential equation (PDE), to estimate the percentage of tissue
damage due to hyperthermia treatment. Differential evolution is used to optimize the
treatment, suggesting the best locations for injecting magnetic nanoparticles to maximize
the damage to tumor cells and minimize harm to healthy tissue. Tests were performed
considering two- and three-dimensional domains, and for each one, three scenarios were
considered to evaluate the suggestions obtained by the optimization method. The results
indicate that the proposed technique is promising: a reduction in the percentage of damage
to healthy tissue while keeping the complete damage to the tumor tissue were observed.
Considering the three-dimensional domain in its most complex scenario, the optimization
process was responsible for reducing the damage to healthy tissue by 59% when the
nanoparticle injections are positioned in a non-intuitive location, i.e., different from the
center of the tumors. The numerical solution of the PDE combined with differential
evolution increases the computational effort required for algorithm execution. Due to this
factor, a parallel strategy using CUDA architecture was implemented to solve the PDE
at a lower cost using NVIDIA GPUs. Comparing the parallel version of the code with its
sequential version executed in a single CPU, an 84.4-fold performance gain was observed.
Description
Keywords
Citation
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwise noted, this item's license is described as Acesso Aberto

