Data di Pubblicazione:
2019
Abstract:
VoxLogicA [1] is a free and open source, multi-platform tool, catering for a novel approach to image segmentation, bringing in ideas from formal methods in software engineering, rapid-development, and declarative programming languages, that have been successful in other domains. In a few lines of code, complex analyses can be specified, translating domain knowledge into logical properties. For instance, in brain tumor segmentation, logical properties encode facts such as ''the oedema touches the tumor'', or ''the tumor contains hyperintense areas; furthermore, very intense areas that are very close to hyperintense ones, are part of the tumor''. Such constraints are extremely effective at filtering noise in automated analysis. The logical core is extended by including imaging primitives, e.g., texture similarity or image normalisation. The language is ''a query language for image analysis''. Its innovation potential can be compared to that of the ''Structured Query Language'' SQL, that revolutionised automated data analysis, by permitting queries on large datasets to be designed by experts of the domain to which the data belongs, instead of computer programmers.
Tipologia CRIS:
04.02 Abstract in Atti di convegno
Keywords:
Model Checking; Spatial Logics; Medical Imaging; segmentation; radiotherapy; glioblastoma
Elenco autori:
Massink, Mieke; Latella, Diego; Ciancia, Vincenzo
Link alla scheda completa:
Link al Full Text:
Titolo del libro:
ESMRMB 2019 Congress October 3-5, Rotterdam/NL Book of Abstracts Electronic Posters / Paper Posters / Clinical Review Posters / Software Exhibits
Pubblicato in: