SAM-Assisted Ultrasound 3D Reconstruction Using Synthetic Data from CT Images

Le résumé fourni par la source

3D reconstruction of ultrasound images plays a vital role in surgical navigation by enabling accurate three-dimensional imaging of organs such as the liver. This not only facilitates early disease diagnosis but also provides essential support for surgical planning and treatment. To address the lack of standardized datasets, we propose a method for converting CT images into ultrasound (US) images, thereby constructing a synthetic dataset for training and evaluation. To overcome the issues of blurred vascular structures and unclear boundaries in the reconstruction results, we introduce a SAM-based vessel segmentation method, which provides critical structural information for 3D reconstruction. Experiments conducted on the synthetic dataset demonstrate that the average reconstruction errors in the X, Y, and Z directions are 6.56 mm, 2.67 mm, and 3.38 mm, achieving better performance than the feature point matching-based method.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.