Purpose: Techniques for three-dimensional (3D) co-registration (fusion) of x-ray fluoroscopy and echocardiography are under development to improve device visualization relative to critical anatomy during structural heart interventions. A phantom was constructed to perform multi-modality imaging in simulated transcatheter interventions and evaluate registration accuracy.
Methods: The phantom consists of a 20-cm diameter, 15-cm tall shell approximating the thoracic region of a small adult. Within, a columnar agar insert is surrounded by an ultrasound-compatible, tissue-mimicking fluid medium. An insert for evaluating x-ray/echo co-registration with embedded, custom-made dual-modality targets was designed. The targets were 5-mm agar spheres infused with barium sulfate and glass microbeads to yield x-ray and ultrasound contrast. Targets were arranged in a helix that encompassed a volume representative of the heart. The phantom was used to test a potential co-registration algorithm in which the 3D pose of a transthoracic echo (TTE) probe in x-ray coordinates is determined from an x-ray projection of small steel fiducials attached to the probe. An iterative probe pose-estimation algorithm was applied using a randomly chosen initial estimate within Â±3mm and Â±3Â° of the ground-truth pose. Registration accuracy was quantified with fiducial and target registration errors (FRE and TRE). Ground truth fiducial and target positions were measured from a C-arm CT scan.
Results: Pose estimation was performed on an x-ray projection image of the TTE probe with attached fiducials for five standard acoustic windows and three gantry orientations (10RAO, 0PA, 10LAO) per window. Mean 3D error (FRE/TRE) ranged from 0.38/2.83 mm up to 6.02/8.68 mm. The phantom was capable of identifying both good and poor TTE probe pose estimation by comparing echo-derived target positions to x-ray-derived target positions in a common x-ray coordinate space.
Conclusion: This customizable phantom can be used to quantify the accuracy of emerging x-ray/echo co-registration methods in 2D and 3D.
Funding Support, Disclosures, and Conflict of Interest: Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award Number T32GM008349, and funding received from Siemens Healthineers. The concepts presented in this paper are based on research and are not commercially available