Hierarchical CT to Ultrasound Registration of the Lumbar Spine: A Comparison with Other Registration Methods

Abstract

Three-dimensional (3D) measurement of the spine can provide important information for functional, developmental, diagnostic, and treatment-effect evaluations. However, existing measurement techniques are either 2-dimensional, highly invasive, or involve a high radiation dose, prohibiting their widespread and repeated use in both research and clinical settings. Non-invasive, non-ionizing, 3D measurement of the spine is still beyond the current state-of-the-art. Towards this goal, we developed an intensity-based hierarchical CT-ultrasound registration approach to quantify the 3D positions and orientations of lumbar vertebrae from 3D freehand ultrasound and one-time computed tomography. The method was validated using a human dry bone specimen (T12-L5) and a porcine cadaver (L2-L6) by comparing the registration results with a gold standard fiducial-based registration. Mean (SD) target registration error and percentage of successful registration were 1.2 (0.6) mm and 100% for the human dry bone specimen, and 2.18 (0.82) mm and 92% for the porcine cadaver, indicating that the method is accurate and robust under clinically realistic conditions. Given that the use of ultrasound eliminates ionizing radiation during pose measurements, we believe that the hierarchical CT-ultrasound registration method is an attractive option for quantifying 3D poses of individual vertebra and motion segment, and thus warrants further investigations.

A method of 3D‐3D multi‐stage non‐rigid registration of the spine based on binocular structured light

“…The registration accuracy of the spine obtained by this method was 1.7 mm. Compared with this method, Terry et al 31 . proposed to use CT images and ultrasonic images to complete the registration of lumbar images, and the accuracy of the registration reached 1.2 mm.…”

Actuator-Assisted Calibration of Freehand 3D Ultrasound System

“…In this study, actuator-assisted calibration was implemented on three point-based phantoms: (1) single cross wire phantom, (2) original collinear point target phantom, and (3) modified collinear point target phantom. To facilitate a direct comparison of the precision and accuracy among phantoms, calibrations of all phantoms were based on 40 point targets using the same freehand 3D ultrasound system [ 4 , 19 ] with depth and frequency settings of 90 mm and 4 MHz, respectively. The freehand 3D ultrasound system used in this study consisted of an ultrasound scanner (Ultramark 400c; ATL Ultrasound Inc., Bothell, WA) with a curvilinear probe (CLA 3.5/40), an optical tracking system (Optotrak 3020, Northern Digital Inc., Waterloo, Canada) with 5 noncoplanar infrared (IR) diodes attached on the ultrasound probe as well as on each phantom to keep track of the probe and phantom's pose with respect to the world coordinate system, and a personal computer with a frame grabber and a data acquisition card installed for capturing ultrasound images.…”

Ultrasound-based navigated pedicle screw insertion without intraoperative radiation: feasibility study on porcine cadavers

“…So far, most of the work emphasizes on evaluating the registration quality using plastic phantoms [8], [9], [10], ex vivo spine anatomy [11], or animal cadavers including sheep [12], [13], [14] and pigs [15], [16], [17], [18], [19]. The few papers evaluating feasibility on clinical data have focused on needle guidance applications [20], [22], [22], [23].…”