Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ACCURATE: Arbitrary-shaped Continuum Reconstruction Under Robust Adaptive Two-view Estimation

Yaozhi Zhang, Shun Yu, Yugang Zhang, Yang Liu

TL;DR

This work proposes ACCURATE, a 3D reconstruction framework integrating an image segmentation neural network with a geometry-constrained topology traversal and dynamic programming algorithm that enforces global biplanar geometric consistency, minimizes the cumulative point-to-epipolar-line distance, and remains robust to occlusions and epipolar ambiguities cases caused by noise and discretization.

Abstract

Accurate reconstruction of arbitrary-shaped long slender continuum bodies, such as guidewires, catheters and other soft continuum manipulators, is essential for accurate mechanical simulation. However, existing image-based reconstruction approaches often suffer from limited accuracy because they often underutilize camera geometry, or lack generality as they rely on rigid geometric assumptions that may fail for continuum robots with complex and highly deformable shapes. To address these limitations, we propose ACCURATE, a 3D reconstruction framework integrating an image segmentation neural network with a geometry-constrained topology traversal and dynamic programming algorithm that enforces global biplanar geometric consistency, minimizes the cumulative point-to-epipolar-line distance, and remains robust to occlusions and epipolar ambiguities cases caused by noise and discretization. Our method achieves high reconstruction accuracy on both simulated and real phantom datasets acquired using a clinical X-ray C-arm system, with mean absolute errors below 1.0 mm.

ACCURATE: Arbitrary-shaped Continuum Reconstruction Under Robust Adaptive Two-view Estimation

TL;DR

This work proposes ACCURATE, a 3D reconstruction framework integrating an image segmentation neural network with a geometry-constrained topology traversal and dynamic programming algorithm that enforces global biplanar geometric consistency, minimizes the cumulative point-to-epipolar-line distance, and remains robust to occlusions and epipolar ambiguities cases caused by noise and discretization.

Abstract

Accurate reconstruction of arbitrary-shaped long slender continuum bodies, such as guidewires, catheters and other soft continuum manipulators, is essential for accurate mechanical simulation. However, existing image-based reconstruction approaches often suffer from limited accuracy because they often underutilize camera geometry, or lack generality as they rely on rigid geometric assumptions that may fail for continuum robots with complex and highly deformable shapes. To address these limitations, we propose ACCURATE, a 3D reconstruction framework integrating an image segmentation neural network with a geometry-constrained topology traversal and dynamic programming algorithm that enforces global biplanar geometric consistency, minimizes the cumulative point-to-epipolar-line distance, and remains robust to occlusions and epipolar ambiguities cases caused by noise and discretization. Our method achieves high reconstruction accuracy on both simulated and real phantom datasets acquired using a clinical X-ray C-arm system, with mean absolute errors below 1.0 mm.
Paper Structure (7 sections, 5 equations, 5 figures, 1 table)

This paper contains 7 sections, 5 equations, 5 figures, 1 table.

Table of Contents

  1. Introduction
  2. Method
  3. Topology-aware Segmentation Network (TSN)
  4. Geometry-consistent Curve Topology Traversal (GCTT)
  5. Epipolar-constrained Dynamic Programming (ECDP)
  6. Experiment
  7. Conclusion

Figures (5)

  • Figure 1: Module (a) takes the original images as input and extracts centerline masks. Module (b) converts the centerlines into topology-ordered point sequences. Module (c) establishes cross-view point correspondences to reconstruct the 3D shape.
  • Figure 2: Illustration of the GCTT point selection strategy. From the current points $\mathcal{W}_t=\{\mathbf{s}_{t-m+1},\dots,\mathbf{s}_t\}$, multiple candidate successors may exist. Geometric loss $\mathcal{L}(\cdot)$ ensures smooth and consistent curve traversal.
  • Figure 3: Robust correspondence recovery by ECDP under challenging conditions.(a) Occlusion: ECDP interpolates consistent cross-view point pairs. (b) Epipolar degeneracy: ECDP avoids local mismatches.
  • Figure 4: GT and reconstruction results on simulation dataset.
  • Figure 5: Image, GT and reconstruction results on real phantom dataset.