Publication detail

Chest X-ray Image Analysis using Convolutional Vision Transformer

MEZINA, A. BURGET, R.

Original Title

Chest X-ray Image Analysis using Convolutional Vision Transformer

Type

conference paper

Language

English

Original Abstract

In recent years, computer techniques for clinical image analysis have been improved significantly, especially because of the pandemic situation. Most recent approaches are focused on the detection of viral pneumonia or COVID-19 diseases. However, there is less attention to common pulmonary diseases, such as fibrosis, infiltration and others. This paper introduces the neural network, which is aimed to detect 14 pulmonary diseases. This model is composed of two branches: global, which is the InceptionNetV3, and local, which consists of Inception modules and a modified Vision Transformer. Additionally, the Asymmetric Loss function was utilized to deal with the problem of multilabel classification. The proposed model has achieved an AUC of 0.8012 and an accuracy of 0.7429, which outperforms the well-known classification models.

Keywords

deep learning, multilabel classification, chest Xray images, Vision transformer, InceptionNetV3

Authors

MEZINA, A.; BURGET, R.

Released

25. 4. 2023

Publisher

Brno University of Technology, Faculty of Electrical Engineering and Communication

Location

Brno

ISBN

978-80-214-6154-3

Book

Proceedings II of the 29th Conference STUDENT EEICT 2023 Selected papers

Edition

1

ISBN

2788-1334

Periodical

Proceedings II of the Conference STUDENT EEICT

State

Czech Republic

Pages from

161

Pages to

165

Pages count

5

URL

BibTex

@inproceedings{BUT183898,
  author="Anzhelika {Mezina} and Radim {Burget}",
  title="Chest X-ray Image Analysis using Convolutional Vision Transformer",
  booktitle="Proceedings II of the 29th Conference STUDENT EEICT 2023 Selected papers",
  year="2023",
  series="1",
  journal="Proceedings II of the Conference STUDENT EEICT",
  pages="161--165",
  publisher="Brno University of Technology, Faculty of Electrical Engineering and Communication",
  address="Brno",
  doi="10.13164/eeict.2023.161",
  isbn="978-80-214-6154-3",
  issn="2788-1334",
  url="https://www.eeict.cz/eeict_download/archiv/sborniky/EEICT_2023_sbornik_2_v2.pdf"
}