Learning label diffusion maps for semi-automatic segmentation of lung CT images with COVID-19

Página do item simplificado
dc.contributor.authorBruzadin, Aldimir [UNESP]
dc.contributor.authorBoaventura, Maurílio [UNESP]
dc.contributor.authorColnago, Marilaine
dc.contributor.authorNegri, Rogério Galante [UNESP]
dc.contributor.authorCasaca, Wallace [UNESP]
dc.contributor.institutionUniversidade Estadual Paulista (UNESP)
dc.contributor.institutionUniversidade de São Paulo (USP)
dc.date.accessioned2023-07-29T15:42:23Z
dc.date.available2023-07-29T15:42:23Z
dc.date.issued2023-02-14
dc.description.abstractDeep Learning (DL) has become one of the key approaches for dealing with many challenges in medical imaging, which includes lung segmentation in Computed Tomography (CT). The use of seeded segmentation methods is another effective approach to get accurate partitions from complex CT images, as they give users autonomy, flexibility and easy usability when selecting specific targets for measurement purposes or pharmaceutical interventions. In this paper, we combine the accuracy of deep contour leaning with the versatility of seeded segmentation to yield a semi-automatic framework for segmenting lung CT images from patients affected by COVID-19. More specifically, we design a DL-driven approach that learns label diffusion maps from a contour detection network integrated with a label propagation model, used to diffuse the seeds over the CT images. Moreover, the trained model induces the diffusion of the seeds by only taking as input a marked CT-scan, segmenting hundreds of CT slices in an unsupervised and recursive way. Another important trait of our framework is that it is capable of segmenting lung structures even in the lack of well-defined boundaries and regardless of the level of COVID-19 infection. The accuracy and effectiveness of our learned diffusion model are attested to by both qualitative as well as quantitative comparisons involving several user-steered segmentations methods and eight CT data sets containing different types of lesions caused by COVID-19.en
dc.description.affiliationSão Paulo State University (UNESP) IBILCE, SP
dc.description.affiliationUniversity of São Paulo (USP) ICMC, SP
dc.description.affiliationSão Paulo State University (UNESP) ICT, SP
dc.description.affiliationUnespSão Paulo State University (UNESP) IBILCE, SP
dc.description.affiliationUnespSão Paulo State University (UNESP) ICT, SP
dc.format.extent24-38
dc.identifierhttp://dx.doi.org/10.1016/j.neucom.2022.12.003
dc.identifier.citationNeurocomputing, v. 522, p. 24-38.
dc.identifier.doi10.1016/j.neucom.2022.12.003
dc.identifier.issn1872-8286
dc.identifier.issn0925-2312
dc.identifier.scopus2-s2.0-85144008582
dc.identifier.urihttp://hdl.handle.net/11449/249474
dc.language.isoeng
dc.relation.ispartofNeurocomputing
dc.sourceScopus
dc.subjectCOVID-19
dc.subjectDeep contour learning
dc.subjectLung CT
dc.subjectSeeded segmentation
dc.titleLearning label diffusion maps for semi-automatic segmentation of lung CT images with COVID-19en
dc.typeArtigo

Arquivos

Coleções

Artigos - Matemática Aplicada - IBILCE