Abstract

In many segmentation scenarios, labeled images contain rich structural information about spatial arrangement and shapes of the ob- jects. Integrating this rich information into supervised learning tech- niques is promising as it generates models which go beyond learning class association, only. This paper proposes a new supervised forest model for joint classification-regression which exploits both class and structural information. Training our model is achieved by optimizing a joint ob- jective function of pixel classification and shape regression. Shapes are represented implicitly via signed distance maps obtained directly from ground truth label maps. Thus, we can associate each image point not only with its class label, but also with its distances to ob ject boundaries, and this at no additional cost regarding annotations. The regression com- ponent acts as spatial regularization learned from data and yields a pre- dictor with both class and spatial consistency. In the challenging context of simultaneous multi-organ segmentation, we demonstrate the potential of our approach through experimental validation on a large dataset of 80 three-dimensional CT scans.