Open-source software for ultrasound-based guidance in spinal fusion surgery.

Spinal instrumentation and surgical manipulations may cause loss of navigation accuracy requiring an efficient re-alignment of the patient anatomy with pre-operative images during surgery. While intra-operative ultrasound (iUS) guidance has shown clear potential to reduce surgery time, compared with clinical computed tomography (CT) guidance, rapid registration aiming to correct for patient misalignment has not been addressed. In this article, we present an open-source platform for pedicle screw navigation using iUS imaging. The alignment method is based on rigid registration of CT to iUS vertebral images and has been designed for fast and fully automatic patient re-alignment in the operating room. Two steps are involved: first, we use the iUS probe's trajectory to achieve an initial coarse registration; then, the registration transform is refined by simultaneously optimizing gradient orientation alignment and mean of iUS intensities passing through the CT-defined posterior surface of the vertebra. We evaluated our approach on a lumbosacral section of a porcine cadaver with seven vertebral levels. We achieved a median target registration error of 1.47 mm (100% success rate, defined by a target registration error <2 mm) when applying the probe's trajectory initial alignment. The approach exhibited high robustness to partial visibility of the vertebra with success rates of 89.86% and 88.57% when missing either the left or right part of the vertebra and robustness to initial misalignments with a success rate of 83.14% for random starts within ±20° rotation and ±20 mm translation. Our graphics processing unit implementation achieves an efficient registration time under 8 s, which makes the approach suitable for clinical application.

Toward real-time rigid registration of intra-operative ultrasound with preoperative CT images for lumbar spinal fusion surgery.

PURPOSE: Accurate and effective registration of the vertebrae is crucial for spine surgical navigation procedures. Patient movement, surgical instrumentation or inadvertent contact with the tracked reference during the intervention may invalidate the registration, requiring a rapid correction of the misalignment. In this paper, we present a framework to rigidly align preoperative computed tomography (CT) with the intra-operative ultrasound (iUS) images of a single vertebra. METHODS: We use a single caudo-cranial axial sweep procedure to acquire iUS images, from which the scan trajectory is exploited to initialize the registration transform. To refine the transform, locations of the posterior vertebra surface are first extracted, then used to compute the CT-to-iUS image intensity gradient-based alignment. The approach was validated on a lumbosacral section of a porcine cadaver. RESULTS: We achieved an overall median accuracy of 1.48 mm (success rate of 84.42%) in [Formula: see text] 11 s of computation time, satisfying the clinically accepted accuracy threshold of 2 mm. CONCLUSION: Our approach using intra-operative ultrasound to register patient vertebral anatomy to preoperative images matches the clinical needs in terms of accuracy and computation time, facilitating its integration into the surgical workflow.

Ultrasound-CT registration of vertebrae without reconstruction.

PURPOSE: While robust and accurate, our previously developed volume-to-volume ultrasound-CT registration of vertebrae required that the 2D ultrasound slices be reconstructed into a 3D volume, a time-consuming step that increased the total registration time per vertebra. We have modified our registration technique to a slices-to-volume strategy to eliminate the ultrasound reconstruction step in order to make the total registration time more practical intraoperatively. METHODS: The slices-to-volume registration is achieved by performing backward scan line tracing on individual ultrasound slices as they are acquired, and then registering them as a group to the posterior vertebral surface extracted from the pre-operative CT image. The technique is validated using a lumbosacral Sawbones phantom and the lumbosacral section of three porcine cadavers. RESULTS: The slices-to-volume registration reduced the total registration time per vertebra from 8 to 4 min. The registration accuracy and robustness of the slices-to-volume registration were found to be equal or superior to those of our previous volume-to-volume registration. In addition, a trade-off was found between registration accuracy and registration speed by changing the number of ultrasound slices used in the registration. CONCLUSIONS: The slices-to-volume ultrasound-CT registration significantly reduces the total registration time per vertebra, making this automated technique more practical intraoperatively.

Validation of a hybrid Doppler ultrasound vessel-based registration algorithm for neurosurgery.

PURPOSE: We describe and validate a novel hybrid nonlinear vessel registration algorithm for intra-operative updating of preoperative magnetic resonance (MR) images using Doppler ultrasound (US) images acquired on the dura for the correction of brain-shift and registration inaccuracies. We also introduce an US vessel appearance simulator that generates vessel images similar in appearance to that acquired with US from MR angiography data. METHODS: Our registration uses the minimum amount of preprocessing to extract vessels from the raw volumetric images. This prevents the removal of important registration information and minimizes the introduction of artifacts that may affect robustness, while reducing the amount of extraneous information in the image to be processed, thus improving the convergence speed of the algorithm. We then completed 3 rounds of validation for our vessel registration method for robustness and accuracy using (i) a large number of synthetic trials generated with our US vessel simulator, (ii) US images acquired from a real physical phantom made from polyvinyl alcohol cryogel, and (iii) real clinical data gathered intra-operatively from 3 patients. RESULTS: Resulting target registration errors (TRE) of less than 2.5 mm are achieved in more than 90 % of the synthetic trials when the initial TREs are less than 20 mm. TREs of less than 2 mm were achieved when the technique was applied to the physical phantom, and TREs of less than 3 mm were achieved on clinical data. CONCLUSIONS: These test trials show that the proposed algorithm is not only accurate but also highly robust to noise and missing vessel segments when working with US images acquired in a wide range of real-world conditions.

Validation of automated ultrasound-CT registration of vertebrae.

PURPOSE: Image-guided spine surgery requires registration of the patient anatomy and preoperative computed tomography (CT) images. A technique for intraoperative ultrasound image registration to preoperative CT scans was developed and tested. Validation of the ultrasound-CT registration technique was performed using porcine cadavers. METHODS: An ultrasound-CT registration technique was evaluated using 18 thoracic and lumbar vertebrae of 3 porcine cadavers with 10 different sweep patterns for ultrasound acquisition. For each sweep pattern at each vertebra, 100 randomly simulated initial misalignments were introduced. Each misalignment was registered. The resulting registration transformations were compared to gold standard registrations based on implanted fiducials to assess accuracy and robustness of the technique. RESULTS: The orthogonal-sweep acquisition was found to perform best and yielded a registration accuracy of 1.65 mm across all vertebrae on all porcine cadavers, where 82.5% of the registrations resulted in target registration errors below the 2 mm threshold recommended by a joint report from the experts in the field. In addition, we found that registration accuracy varies by the sweep pattern and vertebral level, but neighboring vertebrae tend to result in statistically similar accuracy. Ultrasound-CT registration took an average of 2.5 min to run, and the total registration time per vertebra (also including time for ultrasound acquisition and reconstruction) is approximately 8 min. CONCLUSIONS: A previously described ultrasound-CT registration technique yields clinically acceptable accuracy and robustness on multiple vertebrae across multiple porcine cadavers. The total registration time is shorter than that of surface point-based manual registration.

New prototype neuronavigation system based on preoperative imaging and intraoperative freehand ultrasound: system description and validation.

PURPOSE: The aim of this report is to present IBIS (Interactive Brain Imaging System) NeuroNav, a new prototype neuronavigation system that has been developed in our research laboratory over the past decade that uses tracked intraoperative ultrasound to address surgical navigation issues related to brain shift. The unique feature of the system is its ability, when needed, to improve the initial patient-to-preoperative image alignment based on the intraoperative ultrasound data. Parts of IBIS Neuronav source code are now publicly available on-line. METHODS: Four aspects of the system are characterized in this paper: the ultrasound probe calibration, the temporal calibration, the patient-to-image registration and the MRI-ultrasound registration. In order to characterize its real clinical precision and accuracy, the system was tested in a series of adult brain tumor cases. RESULTS: Three metrics were computed to evaluate the precision and accuracy of the ultrasound calibration. 1) Reproducibility: 1.77 mm and 1.65 mm for the bottom corners of the ultrasound image, 2) point reconstruction precision 0.62-0.90 mm: and 3) point reconstruction accuracy: 0.49-0.74 mm. The temporal calibration error was estimated to be 0.82 ms. The mean fiducial registration error (FRE) of the homologous-point-based patient-to-MRI registration for our clinical data is 4.9 ± 1.1 mm. After the skin landmark-based registration, the mean misalignment between the ultrasound and MR images in the tumor region is 6.1 ± 3.4 mm. CONCLUSIONS: The components and functionality of a new prototype system are described and its precision and accuracy evaluated. It was found to have an accuracy similar to other comparable systems in the literature.

Towards accurate, robust and practical ultrasound-CT registration of vertebrae for image-guided spine surgery.

PURPOSE: Accurate registration of patient anatomy and preoperative computed tomography (CT) images is key to successful image-guided spine surgery. Current manual landmark and surface-based techniques are time-consuming and not always accurate. Intraoperative ultrasound imaging of the vertebrae, combined with automated registration, could improve surgery by improving accuracy, reducing operative time, and decreasing invasiveness. METHODS: We present a simple ultrasound-CT registration technique that is automated, accurate, and robust. Registration is achieved by aligning the posterior vertebral surface, extracted from both CT and ultrasound images, using a forward and a backward scan line tracing method, respectively. The registration technique is validated using a simple plastic phantom in a water bath and a more realistic porcine cadaver in a simulation of open back surgery. RESULTS: Clinically relevant accuracy was estimated by comparing automated registrations with gold standard imaging fiducial-based reference transformations, which yielded target registration errors of under 1 mm for the plastic phantom and under 1.6 mm for the porcine cadaver. CONCLUSIONS: Our registration technique demonstrates good accuracy and robustness under clinically realistic conditions and thus warrants further studies on its surgical application.

Gradient distortion correction for low frequency current density imaging.

Current density imaging (CDI) is a technique that uses magnetic resonance imaging (MRI) to measure the distribution of externally applied electric current inside tissues. However, GDI processing is rendered inaccurate by the distortion caused by the nonlinearity of MRI gradient fields. The distortion interferes with the proper registration and the curl operation required for correct computation of current density vectors. To address this problem, a calibration phantom was imaged to determine the distortion and to generate calibration maps to correct the distorted current density images. A validation experiment involving a cylindrical phantom was performed to verify this method. Comparison of the distorted and corrected images reveals that both the registration and the curl operation are successfully corrected by this method.