Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LIBR+: Improving Intraoperative Liver Registration by Learning the Residual of Biomechanics-Based Deformable Registration

Dingrong Wang, Soheil Azadvar, Jon Heiselman, Xiajun Jiang, Michael Miga, Linwei Wang

TL;DR

A novel \textit{hybrid} registration approach that leverage a linearized iterative boundary reconstruction (LIBR) method based on linear elastic biomechanics, and use deep neural networks to learn its residual to the ground-truth deformation (LIBR+).

Abstract

The surgical environment imposes unique challenges to the intraoperative registration of organ shapes to their preoperatively-imaged geometry. Biomechanical model-based registration remains popular, while deep learning solutions remain limited due to the sparsity and variability of intraoperative measurements and the limited ground-truth deformation of an organ that can be obtained during the surgery. In this paper, we propose a novel \textit{hybrid} registration approach that leverage a linearized iterative boundary reconstruction (LIBR) method based on linear elastic biomechanics, and use deep neural networks to learn its residual to the ground-truth deformation (LIBR+). We further formulate a dual-branch spline-residual graph convolutional neural network (SR-GCN) to assimilate information from sparse and variable intraoperative measurements and effectively propagate it through the geometry of the 3D organ. Experiments on a large intraoperative liver registration dataset demonstrated the consistent improvements achieved by LIBR+ in comparison to existing rigid, biomechnical model-based non-rigid, and deep-learning based non-rigid approaches to intraoperative liver registration.

LIBR+: Improving Intraoperative Liver Registration by Learning the Residual of Biomechanics-Based Deformable Registration

TL;DR

A novel \textit{hybrid} registration approach that leverage a linearized iterative boundary reconstruction (LIBR) method based on linear elastic biomechanics, and use deep neural networks to learn its residual to the ground-truth deformation (LIBR+).

Abstract

The surgical environment imposes unique challenges to the intraoperative registration of organ shapes to their preoperatively-imaged geometry. Biomechanical model-based registration remains popular, while deep learning solutions remain limited due to the sparsity and variability of intraoperative measurements and the limited ground-truth deformation of an organ that can be obtained during the surgery. In this paper, we propose a novel \textit{hybrid} registration approach that leverage a linearized iterative boundary reconstruction (LIBR) method based on linear elastic biomechanics, and use deep neural networks to learn its residual to the ground-truth deformation (LIBR+). We further formulate a dual-branch spline-residual graph convolutional neural network (SR-GCN) to assimilate information from sparse and variable intraoperative measurements and effectively propagate it through the geometry of the 3D organ. Experiments on a large intraoperative liver registration dataset demonstrated the consistent improvements achieved by LIBR+ in comparison to existing rigid, biomechnical model-based non-rigid, and deep-learning based non-rigid approaches to intraoperative liver registration.
Paper Structure (18 sections, 6 equations, 6 figures, 3 tables)

This paper contains 18 sections, 6 equations, 6 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Background
  3. LIBR:
  4. Spline-GCNN:
  5. Methodology
  6. Construction of bipartite and liver-mesh graphs
  7. Liver-mesh graph:
  8. Bipartite graph:
  9. Spline Residual Graph Convolutional Networks (SR-GCN)
  10. Bipartite branch
  11. Liver-mesh branch:
  12. Optimization of LIBR+
  13. Experiments and Results
  14. Data:
  15. Baselines and metrics:
  16. ...and 3 more sections

Figures (6)

  • Figure 1: Overview of LIBR+
  • Figure 2: Outline of SR-GCN architecture.
  • Figure 3: A: TRE by categories of sparsae measurements used. S = Surface; USP = US Plane. B: Visual examples of vertex-wise error maps from LIBR vs. LIBR+.
  • Figure 4: Similar topology of preoperative (a), LIBR (b) and GT (c) deformed mesh after data coarsening. This gives a verified support of the assumption that a shared tetrahedron topology could be extracted from these three meshes after data coarsening.
  • Figure 5: Pipeline of data coarsening process: (a) Surface Down-Sample, (b) Surface Down-Sampling, and (c) Surface Reconstruction. We take preoperative mesh of liver model 1 as an example.
  • ...and 1 more figures