Cascade of convolutional neural networks for lung texture classification: Overcoming ontological overlapping

Sebastian Roberto Tarando, Catalin Fetita, Pierre Yves Brillet

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

The infiltrative lung diseases are a class of irreversible, non-neoplastic lung pathologies requiring regular follow-up with CT imaging. Quantifying the evolution of the patient status imposes the development of automated classification tools for lung texture. Traditionally, such classification relies on a two-dimensional analysis of axial CT images. This paper proposes a cascade of the existing CNN based CAD system, specifically tuned-up. The advantage of using a deep learning approach is a better regularization of the classification output. In a preliminary evaluation, the combined approach was tested on a 13 patient database of various lung pathologies, showing an increase of 10% in True Positive Rate (TPR) with respect to the best suited state of the art CNN for this task.

Original languageEnglish
Title of host publicationMedical Imaging 2017
Subtitle of host publicationComputer-Aided Diagnosis
EditorsNicholas A. Petrick, Samuel G. Armato
PublisherSPIE
ISBN (Electronic)9781510607132
DOIs
Publication statusPublished - 1 Jan 2017
Externally publishedYes
EventMedical Imaging 2017: Computer-Aided Diagnosis - Orlando, United States
Duration: 13 Feb 201716 Feb 2017

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume10134
ISSN (Print)1605-7422

Conference

ConferenceMedical Imaging 2017: Computer-Aided Diagnosis
Country/TerritoryUnited States
CityOrlando
Period13/02/1716/02/17

Keywords

  • Convolutional networks
  • Deep learning
  • Emphysema
  • Fibrosis
  • Ground glass
  • Infiltrative lung diseases
  • Lung texture classification

Fingerprint

Dive into the research topics of 'Cascade of convolutional neural networks for lung texture classification: Overcoming ontological overlapping'. Together they form a unique fingerprint.

Cite this