EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Adapted generative latent diffusion models for accurate pathological analysis in chest X-ray images

Daniel I Morís, Joaquim de Moura, Jorge Novo, Marcos Ortega

Medical & Biological Engineering & Computing · 2024-07 · 62 (7) · 2189-2212

DOI: 10.1007/s11517-024-03056-5 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Medical & Biological Engineering & Computing, online erschienen am 19.03.2024. Verfasst von Daniel I Morís und 3 weiteren Autor:innen. Beteiligte Einrichtung: Centro de Investigación CITIC, Universidade da Coruña, A Coruña, Spain. Hauptthemen laut MEDLINE: Algorithms, Radiography, Thoracic. Die Arbeit wurde bisher 20-mal zitiert (Crossref) und nennt 68 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Ministerio de Ciencia e Innovación, Xunta de Galicia.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Respiratory diseases have a significant global impact, and assessing these conditions is crucial for improving patient outcomes. Chest X-ray is widely used for diagnosis, but expert evaluation can be challenging. Automatic computer-aided diagnosis methods can provide support for clinicians in these tasks. Deep learning has emerged as a set of algorithms with exceptional potential in such tasks. However, these algorithms require a vast amount of data, often scarce in medical imaging domains. In this work, a new data augmentation methodology based on adapted generative latent diffusion models is proposed to improve the performance of an automatic pathological screening in two high-impact scenarios: tuberculosis and lung nodules. The methodology is evaluated using three publicly available datasets, representative of real-world settings. An ablation study obtained the highest-performing image generation model configuration regarding the number of training steps. The results demonstrate that the novel set of generated images can improve the performance of the screening of these two highly relevant pathologies, obtaining an accuracy of 97.09%, 92.14% in each dataset of tuberculosis screening, respectively, and 82.19% in lung nodules. The proposal notably improves on previous image generation methods for data augmentation, highlighting the importance of the contribution in these critical public health challenges. Graphical abstract

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

AlgorithmsRadiography, ThoracicHumansDeep LearningLungTuberculosisTuberculosis, PulmonaryLung NeoplasmsImage Processing, Computer-AssistedDiagnosis, Computer-Assisted

Stichwörter der Autor:innen

Chest X-rayDeep learningLung nodulesStable diffusionTuberculosis

Autor:innen und Institutionen

  • Centro de Investigación CITIC, Universidade da Coruña, A Coruña, Spain.
  • Grupo VARPA, Instituto de Investigación Biomédica de A Coruña (INIBIC), Universidade da Coruña, A Coruña, Spain.
  • Centro de Investigación CITIC, Universidade da Coruña, A Coruña, Spain. joaquim.demoura@udc.es.
  • Grupo VARPA, Instituto de Investigación Biomédica de A Coruña (INIBIC), Universidade da Coruña, A Coruña, Spain. joaquim.demoura@udc.es.

Förderung

  • Ministerio de Ciencia e Innovación
  • Xunta de Galicia

Zitieren

APA

Morís, D. I., Moura, J., Novo, J., & Ortega, M. (2024). Adapted generative latent diffusion models for accurate pathological analysis in chest X-ray images. Medical & Biological Engineering & Computing, 62(7), 2189-2212. https://doi.org/10.1007/s11517-024-03056-5

Vancouver

Morís DI, Moura J, Novo J, Ortega M. Adapted generative latent diffusion models for accurate pathological analysis in chest X-ray images. Medical & Biological Engineering & Computing. 2024;62(7):2189-2212. doi:10.1007/s11517-024-03056-5

BibTeX
@article{Mors2024,
  author = {Morís, Daniel I and Moura, Joaquim de and Novo, Jorge and Ortega, Marcos},
  title = {Adapted generative latent diffusion models for accurate pathological analysis in chest X-ray images},
  journal = {Medical & Biological Engineering & Computing},
  year = {2024},
  volume = {62},
  number = {7},
  pages = {2189--2212},
  doi = {10.1007/s11517-024-03056-5}
}
Kostenlos starten →