Multi-Atlas-Based Segmentation of the Left Atrium and Pulmonary Veins

Zulma Sandoval (1) , Julian Betancur (1) , Jean-Louis Dillenseger (1)

1 LTSI - Laboratoire Traitement du Signal et de l'Image

This paper presents a multi-atlas segmentation approach concerning the left atrium and pulmonary veins in pre-operative CT images in order to plan ablation therapy in patients with atrial fibrillation. The segmentation procedure is composed of an atlas-based segmentation followed by a region-growing method. The atlas-based segmentation step exploits the a priori knowledge of existing structures to extract the inner region of the left atrium. The output of the atlas-based segmentation is then eroded to be used as seed volume in the region-growing procedure. This step adds new voxels according to a criterion which uses the intensity information of the input image.

Mots clés: Multi-atlas segmentation; affine registration; elastic registration; region-growing; left atrium; pulmonary veins; Atrial fibrillation therapy

Domaines: Ingénierie biomédicale

Zulma Sandoval, Julian Betancur, Jean-Louis Dillenseger. Multi-Atlas-Based Segmentation of the Left Atrium and Pulmonary Veins. Statistical Atlases and Computational Models of the Heart - MICCAI 2013, Sep 2013, Nagoya, Japan. pp.24-30, ⟨10.1007/978-3-642-54268-8_3⟩. ⟨inserm-00950446⟩