Template-Guided Reconstruction of Pulmonary Segments with Neural Implicit Functions

2025-05-13

AI Summary

Precise submillimeter 3D pulmonary segment reconstruction is critical for lung subsegmentectomy but remains challenging due to the computational inefficiency and insufficient geometric fidelity of existing deep learning methods. To address this, we propose a template-guided neural implicit reconstruction paradigm: a learnable anatomical template drives a neural implicit function, integrated with differentiable deformation, multi-structure joint segmentation, and anatomy-aware loss constraints—balancing reconstruction accuracy and inference efficiency. We introduce the first clinically oriented evaluation metrics and present Lung3D, the first publicly available 3D lung shape dataset with multi-vessel annotations (800 cases). On Lung3D, our method reduces surface reconstruction error by 32% and achieves a topology correctness rate of 96.7%, significantly outperforming state-of-the-art approaches. Both code and dataset are open-sourced.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Learning with Manifolds
• Search and Optimization: Learning to Search

Application Category

• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

High-quality 3D reconstruction of pulmonary segments plays a crucial role in segmentectomy and surgical treatment planning for lung cancer. Due to the resolution requirement of the target reconstruction, conventional deep learning-based methods often suffer from computational resource constraints or limited granularity. Conversely, implicit modeling is favored due to its computational efficiency and continuous representation at any resolution. We propose a neural implicit function-based method to learn a 3D surface to achieve anatomy-aware, precise pulmonary segment reconstruction, represented as a shape by deforming a learnable template. Additionally, we introduce two clinically relevant evaluation metrics to assess the reconstruction comprehensively. Further, due to the absence of publicly available shape datasets to benchmark reconstruction algorithms, we developed a shape dataset named Lung3D, including the 3D models of 800 labeled pulmonary segments and the corresponding airways, arteries, veins, and intersegmental veins. We demonstrate that the proposed approach outperforms existing methods, providing a new perspective for pulmonary segment reconstruction. Code and data will be available at https://github.com/M3DV/ImPulSe.

Problem

• Achieve precise 3D pulmonary segment reconstruction for surgery planning
• Overcome computational limits of deep learning in high-resolution reconstruction
• Introduce a benchmark dataset for pulmonary segment reconstruction algorithms

Innovation

• Neural implicit functions for 3D pulmonary reconstruction
• Deformable template guides anatomy-aware segmentation
• Lung3D dataset with 800 labeled pulmonary segments