Non-rigid Target Tracking in 2D Ultrasound Images Using Hierarchical Grid Interpolation

Lucas Royer 1 Marie Babel 1 Alexandre Krupa 1
1 Lagadic - Visual servoing in robotics, computer vision, and augmented reality
CRISAM - Inria Sophia Antipolis - Méditerranée , Inria Rennes – Bretagne Atlantique , IRISA-D5 - SIGNAUX ET IMAGES NUMÉRIQUES, ROBOTIQUE
Abstract : n this paper, we present a new non-rigid target tracking method within 2D ultrasound (US) image sequence. Due to the poor quality of US images, the motion tracking of a tumor or cyst during needle insertion is considered as an open research issue. Our approach is based on well-known compression algorithm in order to make our method work in real-time which is a necessary condition for many clinical applications. Toward that end, we employed a dedicated hierarchical grid interpolation algorithm (HGI) which can represent a large variety of deformations compared to other motion estimation algorithms such as Overlapped Block Motion Compensation (OBMC), or Block Motion Algorithm (BMA). The sum of squared difference of image intensity is selected as similarity criterion because it provides a good trade-off between computation time and motion estimation quality. Contrary to the others methods proposed in the literature, our approach has the ability to distinguish both rigid and non-rigid motions which are observed in ultrasound image modality. Furthermore, this technique does not take into account any prior knowledge about the target, and limits the user interaction which usually complicates the medical validation process. Finally, a technique aiming at identifying the main phases of a periodic motion (e.g. breathing motion) is introduced. The new approach has been validated from 2D ultrasound images of real human tissues which undergo rigid and non-rigid deformations.
Type de document :
Communication dans un congrès
SPIE Medical Imaging 2014 conference,, Feb 2014, San Diego, United States. 2014
Liste complète des métadonnées

Littérature citée [7 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-00933688
Contributeur : Eric Marchand <>
Soumis le : lundi 20 janvier 2014 - 21:37:38
Dernière modification le : jeudi 15 novembre 2018 - 11:57:06
Document(s) archivé(s) le : mardi 22 avril 2014 - 12:51:25

Fichier

royer_spie_2014.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-00933688, version 1

Citation

Lucas Royer, Marie Babel, Alexandre Krupa. Non-rigid Target Tracking in 2D Ultrasound Images Using Hierarchical Grid Interpolation. SPIE Medical Imaging 2014 conference,, Feb 2014, San Diego, United States. 2014. 〈hal-00933688〉

Partager

Métriques

Consultations de la notice

522

Téléchargements de fichiers

269