Papillary muscle segmentation from a multi-atlas database: A feasibility study

Abstract

Planning of mitral valve replacement would benefit from preprocedural 3D models that could allow the clinician to fully understand the patient anatomical and functional condition. However, no single image modality can provide the complete picture alone but 3D echocardiography and magnetic resonance imaging (MRI) could be combined to leverage the advantages of each modality. The fusion of cardiac echo and MR images is a challenging task that currently requires the use of anatomical landmarks to drive the registration. In mitral valve treatment planning, the papillary muscles represent an ideal landmark set as they can be clearly identified in both image modalities. In this paper, we address the problem of papillary muscles automatic segmentation from MRI by proposing an atlas-based segmentation method. Results show that a good quality segmentation (Dice score 0.60±0.14 and 0.73±0.06 for anterior and posterior papillary muscle, respectively) can be achieved within the straightforward pipeline provided by this approach, also on images acquired with different scanners. Hence, our atlas-based segmentation method could represent the first key step towards a novel, automated echo and MRI fusion algorithm.