Fast and accurate vision-based stereo reconstruction and motion estimation for image-guided liver surgery.

Image-guided liver surgery aims to enhance the precision of resection and ablation by providing fast localisation of tumours and adjacent complex vasculature to improve oncologic outcome. This Letter presents a novel end-to-end solution for fast stereo reconstruction and motion estimation that demonstrates high accuracy with phantom and clinical data. The authors' computationally efficient coarse-to-fine (CTF) stereo approach facilitates liver imaging by accounting for low texture regions, enabling precise three-dimensional (3D) boundary recovery through the use of adaptive windows and utilising a robust 3D motion estimator to reject spurious data. To the best of their knowledge, theirs is the only adaptive CTF matching approach to reconstruction and motion estimation that registers time series of reconstructions to a single key frame for registration to a volumetric computed tomography scan. The system is evaluated empirically in controlled laboratory experiments with a liver phantom and motorised stages for precise quantitative evaluation. Additional evaluation is provided through testing with patient data during liver resection.

Hand-eye calibration using a target registration error model.

Surgical cameras are prevalent in modern operating theatres and are often used as a surrogate for direct vision. Visualisation techniques (e.g. image fusion) made possible by tracking the camera require accurate hand-eye calibration between the camera and the tracking system. The authors introduce the concept of 'guided hand-eye calibration', where calibration measurements are facilitated by a target registration error (TRE) model. They formulate hand-eye calibration as a registration problem between homologous point-line pairs. For each measurement, the position of a monochromatic ball-tip stylus (a point) and its projection onto the image (a line) is recorded, and the TRE of the resulting calibration is predicted using a TRE model. The TRE model is then used to guide the placement of the calibration tool, so that the subsequent measurement minimises the predicted TRE. Assessing TRE after each measurement produces accurate calibration using a minimal number of measurements. As a proof of principle, they evaluated guided calibration using a webcam and an endoscopic camera. Their endoscopic camera results suggest that millimetre TRE is achievable when at least 15 measurements are acquired with the tracker sensor ∼80 cm away on the laparoscope handle for a target ∼20 cm away from the camera.

Contact-less stylus for surgical navigation: registration without digitization.

PURPOSE: We present a laser-based, contact-less, stylus for the purpose of fiducial registration and digitization in the context of surgical navigation. METHODS: We augmented a laser pointer with a spatial measurement device and used the laser beam as a means to locate a fiducial in 3D space. We developed a method for calibrating the orientation of the laser beam with respect to its attached tracking target. Digitization of a fiducial was formulated as a line intersection problem, and registration was formulated as a point-to-line registration problem. RESULTS: We achieved an RMS fiducial localization error of 0.63 mm for 151 measurements of 12 fiducial markers. Mean TRE values of less than 1.5 mm over the entire surface of a lumbar vertebra were achievable using 4 fiducial markers. We found that contact-based rigid registration performed carefully under near-ideal conditions outperforms contact-less registration in terms of TRE. CONCLUSION: An inexpensive contact-less stylus can be used to obtain accurate fiducial registration, which can be performed without explicit fiducial digitization.

Guided ultrasound calibration: where, how, and how many calibration fiducials.

PURPOSE: Many image-guided interventions rely on tracked ultrasound where the transducer is augmented with a tracking device. The relationship between the ultrasound image coordinate system and the tracking sensor must be determined accurately via probe calibration. We introduce a novel calibration framework guided by the prediction of target registration error (TRE): Between successive measurements of the calibration phantom, our framework guides the user in choosing the pose of the calibration phantom by optimizing TRE. METHODS: We introduced an oriented line calibration phantom and modeled the ultrasound calibration process as a point-to-line registration problem. We then derived a spatial stiffness model of point-to-line registration for estimating TRE magnitude at any target. Assuming isotropic, identical localization error, we used the model to estimate TRE for each pixel using the current calibration estimate. We then searched through the calibration tool space to find the pose for the next fiducial which maximally minimized TRE. RESULTS: Both simulation and experimental results suggested that TRE decreases monotonically, reaching an asymptote when a sufficient number of measurements (typically around 12) are made. Independent point reconstruction accuracy assessment showed sub-millimeter accuracy of the calibration framework. CONCLUSION: We have introduced the first TRE-guided ultrasound calibration framework. Using a hollow straw as an oriented line phantom, we virtually constructed a rigid lines phantom and modeled the calibration process as a point-to-line registration. Highly accurate calibration was achieved with minimal measurements by using a spatial stiffness model of TRE to strategically choose the pose of the calibration phantom between successive measurements.

A laboratory comparison of computer navigation and individualized guides for distal radius osteotomy.

PURPOSE: This article presents the results of a multiuser, randomized laboratory trial comparing the accuracy and precision of image-based navigation against individualized guides for distal radius osteotomy (DRO). METHODS: Six surgeons each performed four DROs using image-based navigation and four DROs using individualized guides in a laboratory setting with plastic phantom replicas of radii from patients who had received DRO as treatment for radial deformity. Time required and correction errors of ulnar variance, radial inclination, and volar tilt were measured. RESULTS: There were no statistically significant differences in the average correction errors. There was a statistically significant difference in the standard deviation of ulnar variance error (2.0 mm for navigation vs. 0.6 mm for guides). There was a statistically significant difference in the standard deviation of radial inclination error ([Formula: see text] for navigation vs. [Formula: see text] for guides). There were statistically significant differences in the times required (705 s for navigation vs. 214 s for guides) and their standard deviations (144 s for navigation vs. 98 s for guides). CONCLUSIONS: Compared to navigated DRO, individualized guides were easier to use, faster, and produced more precise correction of ulnar variance and radial inclination. The combination of true three-dimensional planning, ease of use, and accurate and precise corrective guidance makes the individualized guide technique a promising approach for performing corrective osteotomy of the distal radius.

Computation and visualization of uncertainty in surgical navigation.

BACKGROUND: Surgical displays do not show uncertainty information with respect to the position and orientation of instruments. Data is presented as though it were perfect; surgeons unaware of this uncertainty could make critical navigational mistakes. METHODS: The propagation of uncertainty to the tip of a surgical instrument is described and a novel uncertainty visualization method is proposed. An extensive study with surgeons has examined the effect of uncertainty visualization on surgical performance with pedicle screw insertion, a procedure highly sensitive to uncertain data. RESULTS: It is shown that surgical performance (time to insert screw, degree of breach of pedicle, and rotation error) is not impeded by the additional cognitive burden imposed by uncertainty visualization. CONCLUSIONS: Uncertainty can be computed in real time and visualized without adversely affecting surgical performance, and the best method of uncertainty visualization may depend upon the type of navigation display.

Image-guided distal radius osteotomy using patient-specific instrument guides.

In this article, we describe a method for computer-assisted distal radius osteotomies in which computer-generated, patient-specific plastic guides are used for intraoperative guidance. Before surgery, the correction and plate location are planned using computed tomography scans for both radii and ulnae, and the planned locations of the distal and proximal drill holes for the plate are saved. A plastic, patient-specific instrument guide is created using a rapid prototyping machine into which a mirror image of intraoperative, accessible bone structure of the distal radius is integrated. This allows for unique positioning of the guide during surgery. For each planned drill location, a guidance hole is incorporated into the guide. During surgery, a conventional incision is made, and the guide is positioned on the radius. The surgeon drills the holes for the plate screws into the intact radius and performs the osteotomy using the conventional technique. Using the predrilled holes, the surgeon affixes the plate to the radius fragments. The guides are easy to integrate into the surgical workflow and minimize the need for intraoperative fluoroscopy for guidance of the procedure.

Comparison study of intraoperative surface acquisition methods for surgical navigation.

Soft-tissue image-guided interventions often require the digitization of organ surfaces for providing correspondence from medical images to the physical patient in the operating room. In this paper, the effect of several inexpensive surface acquisition techniques on target registration error and surface registration error (SRE) for soft tissue is investigated. A systematic approach is provided to compare image-to-physical registrations using three different methods of organ spatial digitization: 1) a tracked laser-range scanner (LRS), 2) a tracked pointer, and 3) a tracked conoscopic holography sensor (called a conoprobe). For each digitization method, surfaces of phantoms and biological tissues were acquired and registered to CT image volume counterparts. A comparison among these alignments demonstrated that registration errors were statistically smaller with the conoprobe than the tracked pointer and LRS (p<0.01). In all acquisitions, the conoprobe outperformed the LRS and tracked pointer: for example, the arithmetic means of the SRE over all data acquisitions with a porcine liver were 1.73 ± 0.77 mm, 3.25 ± 0.78 mm, and 4.44 ± 1.19 mm for the conoprobe, LRS, and tracked pointer, respectively. In a cadaveric kidney specimen, the arithmetic means of the SRE over all trials of the conoprobe and tracked pointer were 1.50 ± 0.50 mm and 3.51 ± 0.82 mm, respectively. Our results suggest that tissue displacements due to contact force and attempts to maintain contact with tissue, compromise registrations that are dependent on data acquired from a tracked surgical instrument and we provide an alternative method (tracked conoscopic holography) of digitizing surfaces for clinical usage. The tracked conoscopic holography device outperforms LRS acquisitions with respect to registration accuracy.

The effect of the cam deformity on the insertion of the femoral component in hip resurfacing.

Surface arthroplasty simulations were generated using 3-dimensional computed tomographic scans from 61 consecutive patients presenting with idiopathic osteoarthritis to evaluate the change in femoral component positioning that would allow optimal alignment when resurfacing a cam-type deformity. Anatomical parameters were measured to quantify the influence of the deformity on the insertion technique of the femoral implant. A modified femoral head ratio was initially calculated from plain radiographs to define the severity of cam deformity in these patients. A severe deformity required more superior translation of the entry point and greater reaming depth to allow safe insertion with optimal implant alignment. This could be achieved while preserving the leg length, minimizing the component size, and maximizing the amount of host bone contact, although the horizontal femoral offset was reduced. These findings suggest that the femoral component can be safely inserted by modifying the surgical technique despite progressive deformity of the femoral head.

Estimation of optimal fiducial target registration error in the presence of heteroscedastic noise.

We study the effect of point dependent (heteroscedastic) and identically distributed anisotropic fiducial localization noise on fiducial target registration error (TRE). We derive an analytic expression, based on the concept of mechanism spatial stiffness, for predicting TRE. The accuracy of the predicted TRE is compared to simulated values where the optimal registration transformation is computed using the heteroscedastic errors in variables algorithm. The predicted values are shown to be contained by the 95% confidence intervals of the root mean square TRE obtained from the simulations.

Computer-assisted FluoroGuide navigation of unicompartmental knee arthroplasty.

BACKGROUND: Most authorities recognize minimally invasive unicompartmental knee arthroplasty (UKA) as technically demanding with concerns regarding loss of implantation accuracy. We have previously reported on the potential inaccuracy of femoral intramedullary guides in UKA leading to poor component positioning. Our 3-dimensional analysis of alignment error showed that a short, narrow intramedullary rod inserted according to the manufacturer's specifications did not accurately find the direction of the anatomic axis, with errors occurring in both the coronal and sagittal planes. We sought to evaluate whether a fluoroscopic computer-assisted minimally invasive UKA procedure would improve the accuracy and precision in the placement of the femoral component in the coronal and sagittal planes compared with conventional surgery. METHODS: We performed a prospective study involving cohorts of 45 conventional versus 53 navigated UKAs. A single surgeon performed all surgeries over a 4-year period. RESULTS: Pain and knee function significantly improved in both surgical groups at 1 and 2 years after surgery. At a minimum of 1-year follow-up, radiographic evaluation revealed significant improvements in coronal alignment precision of the tibial component (p = 0.026) and sagittal alignment precision of the femoral component for the navigated group (p = 0.037). The use of a fluoroscopic computer-assisted technique did not significantly improve the accuracy of any of the alignment angles. CONCLUSION: We cannot justify the additional expense and complexity imposed by fluoroscopic navigation despite the observed improvements in alignment precision. Improved positioning precision may translate into a greater number of long-term functional results, but larger, longer-term studies are needed.

Quantifying degree of difficulty in hip - resurfacing of pistol-grip deformity.

This study used computer simulation to endeavor to quantify the relative degree of difficulty of resurfacing femora with pistol-grip deformities compared to relatively normal femora. Computer models of five pistol-grip femora and one relatively normal femur were computed from computed tomography (CT) scans of patients who had undergone computer-assisted hip resurfacing. A computer simulation of positioning the femoral resurfacing component on the femur was performed to count the number of acceptable configurations of the component on the femur. A high number of acceptable configurations implies that the surgeon has greater freedom, or greater margin for error, in implanting the component compared to a femur with a smaller number of acceptable configurations. We found that pistol-grip deformities dramatically reduce the number of acceptable configurations for valgus alignment, and that such configurations result in decreased femoral offset and increased depth of reaming.

A theoretical comparison of different target registration error estimators.

Estimation of target registration error (TRE), a common measure of the registration accuracy, is an important issue in computer assisted surgeries. Within the last decade, several new approaches have been developed to estimate either the mean squared value of TRE or the distribution of TRE under different noise conditions. In this paper, we theoretically demonstrate that all the proposed algorithms converge to a general Maximum Likelihood (ML) solution. Numerical simulations are performed to validate our derivations. Using experimentally measured fiducial localization error, we provide an example of TRE prediction in the presence of anisotropic noise.

Computation and validation of intra-operative camera uncertainty.

We present a simple method for computing uncertainty in an optical tracking system. Our significant contribution is that the covariance estimates produced by our tracking algorithm are shown to closely match lower bounds established by Ohta and Kanatani [13]. Our work addresses the existence of uncertainty in tracking and is a step toward a complete estimate of intra-operative uncertainty in computer-assisted surgery.

2D/3D registration of multiple bones.

The problem of 2D/3D registration is, given a 3D image of an object and one or more 2D images of the object in known poses, to recover the 3D pose of the object. We propose a solution for registering multiple bones in 2D radiographic images and 3D CT images using normalized correlation coefficient template matching. We performed tests using synthetic radiographs and CT volumes of two knees and one wrist. We obtained good registration results (less than 2 degrees and 2 mm registration error) for all of the larger bones, but were unable to successfully register the small carpal bones of the wrist with high accuracy.

On fiducial target registration error in the presence of anisotropic noise.

We study the effect of anisotropic noise on target registration error (TRE) by using a tracked and calibrated stylus tip as the fiducial registration application. We present a simple, efficient unscented Kalman filter algorithm that is suitable for fiducial registration even with a small number of fiducials. We also derive an equation that predicts TRE under anisotropic noise. The predicted TRE values are shown to closely match the simulated TRE values achieved using our UKF-based algorithm.

Proof of concept of a simple computer-assisted technique for correcting bone deformities.

We propose a computer-assisted technique for correcting bone deformities using the Ilizarov method. Our technique is an improvement over prior art in that it does not require a tracking system, navigation hardware and software, or intraoperative registration. Instead, we rely on a postoperative CT scan to obtain all of the information necessary to plan the correction and compute a correction schedule for the patient. Our laboratory experiments using plastic phantoms produced deformity corrections accurate to within 3.0 degrees of rotation and 1 mm of lengthening.

Three-dimensional analysis of alignment error in using femoral intramedullary guides in unicompartmental knee arthroplasty.

We used computerized simulations with 3-dimensional models of 20 cadaver femora, calculated from computed tomographic scans, and a model of a rod measuring 200 x 5 mm to study femoral alignment accuracy for unicompartmental knee arthroplasty via minimally invasive reconstruction. The anatomical axis and insertion site were identified on each femur. A simulation of all feasible flexion-extension and varus-valgus orientations was performed. The average rod orientation was 3.2 degrees flexion and 2.5 degrees valgus. The range of orientation was 3.2 degrees extension to 9.7 degrees flexion and 4.5 degrees varus to 8.9 degrees valgus. The study suggests that a short narrow intramedullary rod inserted according to the manufacturer's specifications does not accurately find the anatomical axis and may lead to poor alignment of the femoral prosthesis. Given our finding of consistent bias toward excessive flexion and valgus alignment, we recommend that the operating surgeon carefully plan the insertion point of the intramedullary rod during surgery to compensate for this bias.

Fast assessment of acetabular coverage using stereoscopic volume rendering.

Previous CT-based methods of measuring acetabular coverage of the femoral head have either been labor-intensive or have required extensive preprocessing of the data prior to visualization. We propose a method of measuring acetabular coverage using stereoscopic digitally reconstructed radiographs that required very little labor or image preprocessing time. Taking a craniocaudal view of the pelvis, we measured both preoperative and postoperative CTs of 10 patients treated with transtrochanteric periacetabular osteotomy. Measurements were then made in both monocular and stereoscopic rendering modes. Our method is fast, easy, and provides an intuitive means of visualizing an orthopedic parameter that is important in the progression of early hip arthritis.