Reformulating Level Sets as Deep Recurrent Neural Network Approach to Semantic Segmentation

TL;DR

This paper introduces a novel deep recurrent neural network approach called Recurrent Level Set (RLS) for semantic segmentation, improving upon traditional variational level set methods by integrating learnable deep features and addressing multi-instance and initialization sensitivity issues.

Contribution

The paper proposes RLS, a deep recurrent neural network that reformulates level set evolution, and extends it to CRLS for semantic segmentation, combining variational principles with end-to-end learning.

Findings

RLS improves segmentation accuracy over classic level set methods.

CRLS achieves competitive performance with state-of-the-art segmentation models.

The approach reduces computational time compared to traditional variational level set techniques.

Abstract

Variational Level Set (LS) has been a widely used method in medical segmentation. However, it is limited when dealing with multi-instance objects in the real world. In addition, its segmentation results are quite sensitive to initial settings and highly depend on the number of iterations. To address these issues and boost the classic variational LS methods to a new level of the learnable deep learning approaches, we propose a novel definition of contour evolution named Recurrent Level Set (RLS)} to employ Gated Recurrent Unit under the energy minimization of a variational LS functional. The curve deformation process in RLS is formed as a hidden state evolution procedure and updated by minimizing an energy functional composed of fitting forces and contour length. By sharing the convolutional features in a fully end-to-end trainable framework, we extend RLS to Contextual RLS (CRLS) to address semantic segmentation in the wild. The experimental results have shown that our proposed RLS improves both computational time and segmentation accuracy against the classic variations LS-based method, whereas the fully end-to-end system CRLS achieves competitive performance compared to the state-of-the-art semantic segmentation approaches.