Advances in Image-Driven Breathing Motion Modelsin Interventional Surgery: A Review

In thoracic and abdominal interventional procedures, respiratory motion often leads to changes in target position and image blurring, which in turn affects the accuracy of treatment. Compared with Respiratory Motion Management (RMM) techniques, respiratory motion modeling methods show more promise for preoperative and intraoperative applications. In this article, we review various respiratory motion modeling methods that combine 3D+t with the extraction of internal image substitution signals to reconstruct volumetric images with high temporal and spatial resolution, using image reconstruction techniques to support thoracic and abdominal interventional procedures. The article first introduces the method of extracting internal image substitution signals based on different imaging modalities (e.g., real-time ultrasound, real-time Cine MRI, real-time Xray fluoroscopy), and explains in detail the process of acquiring the Deformation Vector Field (DVF). Subsequently, the paper explores the application of statistical methods, adaptive methods, and deep learning models in respiratory motion modeling, supplemented by the introduction of related image reconstruction techniques. Ultimately, the article summarizes the latest progress in the field and provides new perspectives for future research directions.