An important challenge in cancer radiotherapy treatment, is to administer the exact dosage for the tumor, avoiding damage to the surrounding healthy tissue, since there is always the possibility that the patient moves, and this may cause organs to slightly shift their position.
Within this framework, a Fondecyt project from the School of Computer Engineering at the PUCV – led by the scholar and faculty director, Ignacio Araya – proposes the development of a computer method capable of calculating safer ranges of radiation intensity, that respond robustly to those subtle changes. According to the researchers, in this way “we can contribute to the design of the medical treatment, so it is more precise and reliable for the patient”.
The project, named "Hybrid Filtering for Global Optimization: Combining Affine Relaxations, Constraint Programming and Incremental Bounding” – is carried out by a team comprised by undergraduate students, PUCV professors and European experts.
In practical terms, Araya explained that, although the core of the study is mathematical and computational, its impact is concrete and its more direct application field is in healthcare, particularly in radiotherapy planning (IMRT).
“With this project we hope to decrease computing time and increase the robustness of the algorithms, which could be applied in radiotherapy planning, where precision is critical for patient safety. The idea is that computers find guaranteed solutions using less resources, rapidly eliminating the areas where the solution is not optimal”, the scholar commented.
Likewise, he indicated that the initiative addresses the fundamental challenge of global optimization and seeks to “improve the Branch & Bound methods, that explore the solution space in systematic fashion, combining rapid domain filtering techniques with convex relaxation methods, to reduce the search space in a more effective way and accelerate convergence”.
Finally, Araya indicated that, from a mathematical standpoint, the great expectation of the project is to develop a more robust and efficient algorithmic framework to solve complex optimization problems. “The main contribution will be solving problems that currently have many variables and nonlinear restrictions, which are very costly in terms of time and computer resources”, he stated, adding that the project “seeks to create an hybrid method that combines the best of two worlds: the speed of programming by restrictions and the global vision of convex relaxations”.
By Jenny Díaz
Strategic Communication Department
