Publication Details
Accelerated Design of HIFU Treatment Plans Using Island-Based Evolutionary Strategy
Evolutionary strategy, Island model, HIFU, Treatment planning, k-Wave toolbox
High Intensity Focused Ultrasound (HIFU) is an emerging technique for non-invasive cancer treatment where malignant tissue is destroyed by thermal ablation. Such a treatment consists of a series of short sonications destroying small volumes of tissue. High-quality treatment plans allow to precisely target malignant tissue and protect surrounding healthy tissue. Recently, we developed an evolutionary strategy to design such HIFU treatment plans using patient-specific material properties and a realistic thermal model. Unfortunately, the execution time was prohibitive for routine use. Here, we present an optimized evolutionary strategy based on island model parallelization and a revised fitness function implementation. The proposed improvements allow to develop a good treatment plan 4 times faster and with 5% higher success rate.
@inproceedings{BUT162282,
author="Filip {Kukliš} and Marta {Jaroš} and Jiří {Jaroš}",
title="Accelerated Design of HIFU Treatment Plans Using Island-Based Evolutionary Strategy",
booktitle="Applications of Evolutionary Computation",
year="2020",
series="Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
volume="12104",
pages="463--478",
publisher="Springer International Publishing",
address="Cham",
doi="10.1007/978-3-030-43722-0\{_}30",
isbn="978-3-030-43721-3",
url="https://link.springer.com/chapter/10.1007/978-3-030-43722-0_30"
}