Classification of In-vivo Endomicroscopic Images of the Alveolar Respiratory System

Caroline Petitjean (1) , Jonathan Benoist (1) , Luc Thiberville (1) , Mathieu Salaün (1) , Laurent Heutte (1)

1 LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes

This paper presents an original system for the automatic classification of normal versus abnormal endomicroscopic images of the respiratory alveolar system. Images from the alveoli were acquired in-vivo using a newly developed technique, based on confocal microscopy. The classification process includes a 120-feature extraction step followed by an SVM based classification. Results are excellent for non-smoker images and demonstrate the great help of this system for computer-aided diagnosis, using confocal microscopy.

Domaines: Imagerie médicale

Caroline Petitjean, Jonathan Benoist, Luc Thiberville, Mathieu Salaün, Laurent Heutte. Classification of In-vivo Endomicroscopic Images of the Alveolar Respiratory System. IAPR Conference on Machine Vision Applications (MVA), May 2009, Yokohama, Japan. pp.471-474. ⟨hal-00439139⟩