Vertebra partitioning with thin-plate spline surfaces steered by a convolutional neural network

TL;DR

This paper introduces a neural network-based method for partitioning vertebra segmentation masks into substructures by predicting boundary surfaces modeled as thin-plate splines, trained with unpaired data.

Contribution

It presents a novel approach combining CNNs and thin-plate spline surfaces for vertebra segmentation boundary prediction, enabling training with unpaired data.

Findings

Accurate boundary prediction between vertebral substructures.

Effective segmentation with unpaired training data.

Improved vertebra partitioning performance.

Abstract

Thin-plate splines can be used for interpolation of image values, but can also be used to represent a smooth surface, such as the boundary between two structures. We present a method for partitioning vertebra segmentation masks into two substructures, the vertebral body and the posterior elements, using a convolutional neural network that predicts the boundary between the two structures. This boundary is modeled as a thin-plate spline surface defined by a set of control points predicted by the network. The neural network is trained using the reconstruction error of a convolutional autoencoder to enable the use of unpaired data.