Calibration of 3D ultrasound to an electromagnetic tracking system

Abstract

The use of electromagnetic (EM) tracking is an important guidance tool that can be used to aid procedures requiring accurate localization such as needle injections or catheter guidance. Using EM tracking, the information from different modalities can be easily combined using pre-procedural calibration information. These calibrations are performed individually, per modality, allowing different imaging systems to be mixed and matched according to the procedure at hand. In this work, a framework for the calibration of a 3D transesophageal echocardiography probe to EM tracking is developed. The complete calibration framework includes three required steps: data acquisition, needle segmentation, and calibration. Ultrasound (US) images of an EM tracked needle must be acquired with the position of the needles in each volume subsequently extracted by segmentation. The calibration transformation is determined through a registration between the segmented points and the recorded EM needle positions. Additionally, the speed of sound is compensated for since calibration is performed in water that has a different speed then is assumed by the US machine. A statistical validation framework has also been developed to provide further information related to the accuracy and consistency of the calibration. Further validation of the calibration showed an accuracy of 1.39 mm.

Locating Fetal Facial Surface, Oral Cavity and Airways by a 3D Ultrasound Calibration Using a Novel Cones' Phantom

“…The accuracy of our approach is better than the majority of the methods from the comparison in Table 3. Although Lang et al [8] obtained better accuracy than did our approach, a set of 30-50 volumes is required for each experiment to ensure the best accuracy, which made the calibration process long and tiresome. In conclusion, our system guarantees satisfactorily accuracy, and also only requires a simple experimental setup and two US volumes for 3D US calibration.…”

“…They obtained the root mean square (RMS) errors of the point accuracy measure of 2.15 mm (IXI-wire), 4.91 mm (cube), and 2.36 mm (stylus) through an optical tracking system. Lang et al [8] examined two similar solutions based on 2D US images [9], [10] extended to 3D using EM tracking, and obtained an average distance error of 1.39±0.54 mm. Nevertheless, most of them share the alignment and feature extraction problems.…”

Real-Time 3D Ultrasound Guided Interventional System for Cardiac Stem Cell Therapy with Motion Compensation

“…The fusion of the imaging modalities is first bootstrapped using electromagnetic (EM) tracking system and then an image based registration is used to compensate for respiratory motion intra-operatively. First, prior to the procedure, the live-3D US image is precalibrated to the tracking system by rigidly attaching 6-DOF EM sensors to the probe and calibrating using the method described in [6]. Second, the tip of the stem cell injection catheter is also precalibrated to the 5-DOF EM sensor mounted on it.…”

“…The one time calibration of the live 3D US image coordinate to the 6-DOF EM sensor was performed with a calibration accuracy of 1.94 mm [6]. MRI images of the phantom were acquired using Philips Panorama 1T system with fiducials mounted on the phantom.…”

Solving for free-hand and real-time 3D ultrasound calibration with anisotropic orthogonal Procrustes analysis

“…This transformation, a concatenation of scaling, followed by rotation, followed by translation, must be determined accurately. [2][3][4] Solving for the scaling factors between the voxel size to physical units remains a challenge. These scaling factors are anisotropic in nature.…”