Label-Efficient Data Augmentation with Video Diffusion Models for Guidewire Segmentation in Cardiac Fluoroscopy

2024-12-20

AI Summary

To address the scarcity of annotated guidewire segmentation data in cardiac interventional procedures, this paper proposes the Segmentation-Guided Frame-Consistent Video Diffusion model (SF-VD), designed to synthesize high-fidelity, temporally coherent fluoroscopic videos with diverse guidewire visibility—thereby enhancing segmentation model generalization. Methodologically, SF-VD introduces a novel decoupled modeling framework that separately captures scene appearance distribution and motion dynamics; incorporates a segmentation-guided contrast modulation mechanism to enrich guidewire visibility diversity; and enforces inter-frame consistency via a dedicated temporal regularizer. Evaluated on real clinical fluoroscopy datasets, SF-VD achieves superior video fidelity over existing baselines. Guidewire segmentation accuracy improves by 6.2% in Dice score, and under low-data regimes, performance approaches that of full-supervision settings, demonstrating strong data-efficiency and generalizability.

Technology Category

• Computer Vision: Segmentation
• Machine Learning: Calibration & Uncertainty Quantification
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Data transparency and provenance
• Web Mining and Content Analysis: Content-based information diffusion

Abstract

The accurate segmentation of guidewires in interventional cardiac fluoroscopy videos is crucial for computer-aided navigation tasks. Although deep learning methods have demonstrated high accuracy and robustness in wire segmentation, they require substantial annotated datasets for generalizability, underscoring the need for extensive labeled data to enhance model performance. To address this challenge, we propose the Segmentation-guided Frame-consistency Video Diffusion Model (SF-VD) to generate large collections of labeled fluoroscopy videos, augmenting the training data for wire segmentation networks. SF-VD leverages videos with limited annotations by independently modeling scene distribution and motion distribution. It first samples the scene distribution by generating 2D fluoroscopy images with wires positioned according to a specified input mask, and then samples the motion distribution by progressively generating subsequent frames, ensuring frame-to-frame coherence through a frame-consistency strategy. A segmentation-guided mechanism further refines the process by adjusting wire contrast, ensuring a diverse range of visibility in the synthesized image. Evaluation on a fluoroscopy dataset confirms the superior quality of the generated videos and shows significant improvements in guidewire segmentation.

Problem

• Catheter_Detection
• X-ray_Videos
• Surgical_Assistance

Innovation

• SF-VD model
• limited annotated data
• guidewire localization