Intensity-based 2D-3D Registration Research Articles

A comparative analysis of intensity-based 2D-3D registration for intraoperative use in pedicle screw insertion surgeries.

Although multiple algorithms have been reported that focus on improving the accuracy of 2D-3D registration techniques, there has been relatively little attention paid to quantifying their capture range. In this paper, we analyze the capture range for a number of variant formulations of the 2D-3D registration problem in the context of pedicle screw insertion surgery. We tested twelve 2D-3D registration techniques for capture range under different clinically realistic conditions. A registration was considered as successful if its error was less than 2mm and 2° in 95% of the cases. We assessed the sensitivity of capture range to a variety of clinically realistic parameters including: X-ray contrast, number and configuration of X-rays, presence or absence of implants in the image, inter-subject variability, intervertebral motion and single-level vs multi-level registration. Gradient correlation + Powell optimizer had the widest capture range and the least sensitivity to X-ray contrast. The combination of 4 AP + lateral X-rays had the widest capture range (725mm2). The presence of implant projections significantly reduced the registration capture range (up to 84%). Different spine shapes resulted in minor variations in registration capture range (SD 78mm2). Intervertebral angulations of less than 1.5° had modest effects on the capture range. This paper assessed capture range of a number of variants of intensity-based 2D-3D registration algorithms in clinically realistic situations (for the use in pedicle screw insertion surgery). We conclude that a registration approach based on the gradient correlation similarity and the Powell's optimization algorithm, using a minimum of two C-arm images, is likely sufficiently robust for the proposed application.

Intensity-based 2D-3D registration for an ACL reconstruction navigation system.

To improve the positioning accuracy of tunnels for anterior cruciate ligament (ACL) reconstruction, we proposed an intensity-based 2D-3D registration method for an ACL reconstruction navigation system. Methods for digitally reconstructed radiograph (DRR) generation, similarity measurement, and optimization are crucial to 2D-3D registration. We evaluated the accuracy, success rate, and processing time of different methods: (a) ray-casting and splating were compared for DRR generation; (b) normalized mutual information (NMI), Mattes mutual information (MMI), and Spearman's rank correlation coefficient (SRC) were assessed for similarity between registrations; and (c) gradient descent (GD) and downhill simplex (DS) were compared for optimization. The combination of splating, SRC, and GD provided the best composite performance and was applied in an augmented reality (AR) ACL reconstruction navigation system. The accuracy of the navigation system could fulfill the clinical needs of ACL reconstruction, with an end pose error of 2.50mm and an angle error of 2.74°.

Reproducibility of pelvic sagittal inclination while acquiring radiographs in supine and standing postures.

Pelvic position on the sagittal plane is usually evaluated with the pelvic sagittal inclination (PSI) angle from a single radiograph. However, the reproducibility of pelvic positioning has not been investigated, and thus, the validity of measuring the PSI from a single film/time point is not understood. Herein, the reproducibility of a patient's pelvic positions in supine and standing postures was analyzed. A total of 34 patients who underwent either a pelvic osteotomy or total hip arthroplasty were enrolled in this study. Preoperative radiographs in both supine and standing postures were acquired twice (first X-ray and second X-ray) within 6 months; preoperative computed tomography (CT) images of the full pelvis were also acquired in a supine posture (preop-CT). To eliminate measurement variability, each PSI was automatically measured from radiographs and CT images through the use of CT segmentation and landmark localization followed by intensity-based 2D-3D registration. The absolute difference of PSI among each image was calculated and the intra-class correlation coefficient (ICC) in each posture was also analyzed. The median absolute differences of PSI in the supine posture were 1.3° between the first and second X-rays, 1.2° between the first X-ray and preop-CT, and 1.3° between the second X-ray and preop-CT. The median absolute difference of PSI in the standing posture was 1.5°. The ICC was 0.965 (95% CI: 0.939-0.981) in supine and 0.977 (95% CI: 0.954-0.988) during standing. Pelvic positions in supine and standing postures are reproducible. Thus, measuring the PSI from a single radiograph is reliable.

Recovery of 3D rib motion from dynamic chest radiography and CT data using local contrast normalization and articular motion model.

Dynamic chest radiography (2D x-ray video) is a low-dose and cost-effective functional imaging method with high temporal resolution. While the analysis of rib-cage motion has been shown to be effective for evaluating respiratory function, it has been limited to 2D. We aim at 3D rib-motion analysis for high temporal resolution while keeping the radiation dose at a level comparable to conventional examination. To achieve this, we developed a method for automatically recovering 3D rib motion based on 2D-3D registration of x-ray video and single-time-phase computed tomography. We introduce the following two novel components into the conventional intensity-based 2D-3D registration pipeline: (1) a rib-motion model based on a uniaxial joint to constrain the search space and (2) local contrast normalization (LCN) as a pre-process of x-ray video to improve the cost function of the optimization parameters, which is often called the landscape. The effects of each component on the registration results were quantitatively evaluated through experiments using simulated images and real patients' x-ray videos obtained in a clinical setting. The rotation-angle error of the rib and the mean projection contour distance (mPCD) were used as the error metrics. The simulation experiments indicate that the proposed uniaxial joint model improved registration accuracy. By searching the rotation axis along with the rotation angle of the ribs, the rotation-angle error and mPCD significantly decreased from 2.246 ± 1.839° and 1.148 ± 0.743mm to 1.495 ± 0.993° and 0.742 ± 0.281mm, compared to simply applying De Troyer's model. The real-image experiments with eight patients demonstrated that LCN improved the cost function space; thus, robustness in optimization resulting in an average mPCD of 1.255 ± 0.615mm. We demonstrated that an anatomical-knowledge based constraint and an intensity normalization, LCN, significantly improved robustness and accuracy in rib-motion reconstruction using chest x-ray video.

Can Anatomic Measurements of Stem Anteversion Angle Be Considered as the Functional Anteversion Angle?

BackgroundStem anteversion angle is important in the combined anteversion theory to avoid implant impingement after total hip arthroplasty (THA). However, anatomic measurements of stem anteversion angle may not represent functional anteversion of the femur if the femur undergoes axial rotation. Herein, the femoral rotational angle (FRA) was measured in supine and standing positions before and after THA to evaluate the difference between anatomic and functional measurements. MethodsA total of 191 hips (174 patients) treated with THA for osteoarthritis were analyzed in this retrospective, case-controlled study. The FRA was measured as the angle between the posterior condylar line and the line through the bilateral anterior superior iliac spines (positive for external rotation) and was measured preoperatively and postoperatively in supine and standing positions with computed tomography segmentation and landmark localization of the pelvis and the femur followed by intensity-based 2D-3D registration. The number of cases in which the absolute FRA remained <15° in both positions was also calculated. ResultsThe average ± standard deviation preoperative FRA was 0.3° ± 8.3° in the supine position and −4.5° ± 8.8° during standing; the postoperative FRA was −3.8° ± 9.0° in supine and −14.3° ± 8.3° during standing. There were 134 cases (70%) in which the preoperative absolute FRA remained <15° in both positions while only 85 hips (45%) remained <15°, postoperatively. ConclusionSubstantial variability was seen in the FRA, especially during the postoperative period. These results suggest that the anatomic stem anteversion angle may not represent the functional anteversion of the femur.

Change in Pelvic Sagittal Inclination From Supine to Standing Position Before Hip Arthroplasty

BackgroundCup anteversion and inclination are important for avoiding implant impingement and dislocation in total hip arthroplasty. However, functional cup anteversion and cup inclination also change as the pelvic sagittal inclination (PSI) changes. Therefore, PSI in both supine and standing positions was measured in a large cohort in this study. MethodsA total of 422 patients (median age, 61; range, 15-87) who underwent total hip arthroplasty were the subjects of this study. There were 83 patients with primary osteoarthritis (OA), 274 patients with developmental dysplasia–derived secondary OA, 48 patients with osteonecrosis, and 17 patients with rapidly destructive coxopathy (RDC). Preoperative PSI in supine and standing positions was measured by automated computed topography segmentation and landmark localization of the pelvis followed by intensity-based 2D-3D registration, and the number of cases in which PSI changed more than 10° posteriorly was calculated. Hip disease, sex, and age were analyzed if they were related to a PSI change of more than 10°. ResultsThe median PSI was 5.1° (interquartile range, 0.4°-9.4°) in supine and −1.3° (interquartile range, −6.5° to 4.2°) in standing position. There were 79 cases (19%) in which the PSI changed more than 10° posteriorly from supine to standing. Elder age and patients with primary OA and RDC were revealed to be the related factors. ConclusionPSI changed more than 10° posteriorly from supine to standing in 19% of cases. Age and diagnosis of primary OA and RDC were the significant factors for the posterior rotation.

Intensity-based 2D 3D registration for lead localization in robot guided deep brain stimulation

Intraoperative assessment of lead localization has become a standard procedure during deep brain stimulation surgery in many centers, allowing immediate verification of targeting accuracy and, if necessary, adjustment of the trajectory. The most suitable imaging modality to determine lead positioning, however, remains controversially discussed. Current approaches entail the implementation of computed tomography and magnetic resonance imaging. In the present study, we adopted the technique of intensity-based 2D 3D registration that is commonly employed in stereotactic radiotherapy and spinal surgery. For this purpose, intraoperatively acquired 2D x-ray images were fused with preoperative 3D computed tomography (CT) data to verify lead placement during stereotactic robot assisted surgery. Accuracy of lead localization determined from 2D 3D registration was compared to conventional 3D 3D registration in a subsequent patient study. The mean Euclidian distance of lead coordinates estimated from intensity-based 2D 3D registration versus flat-panel detector CT 3D 3D registration was 0.7 mm ± 0.2 mm. Maximum values of these distances amounted to 1.2 mm. To further investigate 2D 3D registration a simulation study was conducted, challenging two observers to visually assess artificially generated 2D 3D registration errors. 95% of deviation simulations, which were visually assessed as sufficient, had a registration error below 0.7 mm.In conclusion, 2D 3D intensity-based registration revealed high accuracy and reliability during robot guided stereotactic neurosurgery and holds great potential as a low dose, cost effective means for intraoperative lead localization.

2D-3D radiograph to cone-beam computed tomography (CBCT) registration for C-arm image-guided robotic surgery.

C-arm radiographs are commonly used for intraoperative image guidance in surgical interventions. Fluoroscopy is a cost-effective real-time modality, although image quality can vary greatly depending on the target anatomy. Cone-beam computed tomography (CBCT) scans are sometimes available, so 2D-3D registration is needed for intra-procedural guidance. C-arm radiographs were registered to CBCT scans and used for 3D localization of peritumor fiducials during a minimally invasive thoracic intervention with a da Vinci Si robot. Intensity-based 2D-3D registration of intraoperative radiographs to CBCT was performed. The feasible range of X-ray projections achievable by a C-arm positioned around a da Vinci Si surgical robot, configured for robotic wedge resection, was determined using phantom models. Experiments were conducted on synthetic phantoms and animals imaged with an OEC 9600 and a Siemens Artis zeego, representing the spectrum of different C-arm systems currently available for clinical use. The image guidance workflow was feasible using either an optically tracked OEC 9600 or a Siemens Artis zeego C-arm, resulting in an angular difference of Δθ:∼ 30°. The two C-arm systems provided TRE mean ≤ 2.5 mm and TRE mean ≤ 2.0 mm, respectively (i.e., comparable to standard clinical intraoperative navigation systems). C-arm 3D localization from dual 2D-3D registered radiographs was feasible and applicable for intraoperative image guidance during da Vinci robotic thoracic interventions using the proposed workflow. Tissue deformation and in vivo experiments are required before clinical evaluation of this system.

Robust initialization of 2D-3D image registration using the projection-slice theorem and phase correlation

The image registration literature comprises many methods for 2D-3D registration for which accuracy has been established in a variety of applications. However, clinical application is limited by a small capture range. Initial offsets outside the capture range of a registration method will not converge to a successful registration. Previously reported capture ranges, defined as the 95% success range, are in the order of 4-11 mm mean target registration error. In this article, a relatively computationally inexpensive and robust estimation method is proposed with the objective to enlarge the capture range. The method uses the projection-slice theorem in combination with phase correlation in order to estimate the transform parameters, which provides an initialization of the subsequent registration procedure. The feasibility of the method was evaluated by experiments using digitally reconstructed radiographs generated from in vivo 3D-RX data. With these experiments it was shown that the projection-slice theorem provides successful estimates of the rotational transform parameters for perspective projections and in case of translational offsets. The method was further tested on ex vivo ovine x-ray data. In 95% of the cases, the method yielded successful estimates for initial mean target registration errors up to 19.5 mm. Finally, the method was evaluated as an initialization method for an intensity-based 2D-3D registration method. The uninitialized and initialized registration experiments had success rates of 28.8% and 68.6%, respectively. The authors have shown that the initialization method based on the projection-slice theorem and phase correlation yields adequate initializations for existing registration methods, thereby substantially enlarging the capture range of these methods.

Computer assisted determination of acetabular cup orientation using 2D–3D image registration

2D-3D image-based registration methods have been developed to measure acetabular cup orientation after total hip arthroplasty (THA). These methods require registration of both the prosthesis and the CT images to 2D radiographs and compute implant position with respect to a reference. The application of these methods is limited in clinical practice due to two limitations: (1) the requirement of a computer-aided design (CAD) model of the prosthesis, which may be unavailable due to the proprietary concerns of the manufacturer, and (2) the requirement of either multiple radiographs or radiograph-specific calibration, usually unavailable for retrospective studies. In this paper, we propose a new method to address these limitations. A new formulation for determination of post-operative cup orientation, which couples a radiographic measurement with 2D-3D image matching, was developed. In our formulation, the radiographic measurement can be obtained with known methods so that the challenge lies in the 2D-3D image matching. To solve this problem, a hybrid 2D-3D registration scheme combining a landmark-to-ray 2D-3D alignment with a robust intensity-based 2D-3D registration was used. The hybrid 2D-3D registration scheme allows computing both the post-operative cup orientation with respect to an anatomical reference and the pelvic tilt and rotation with respect to the X-ray imaging table/plate. The method was validated using 2D adult cadaver hips. Using the hybrid 2D-3D registration scheme, our method showed a mean accuracy of 1.0 +/- 0.7 degrees (range from 0.1 degrees to 2.0 degrees) for inclination and 1.7 +/- 1.2 degrees (range from 0.0 degrees to 3.9 degrees) for anteversion, taking the measurements from post-operative CT images as ground truths. Our new solution formulation and the hybrid 2D-3D registration scheme facilitate estimation of post-operative cup orientation and measurement of pelvic tilt and rotation.

Evaluation of similarity measures for use in the intensity‐based rigid 2D‐3D registration for patient positioning in radiotherapy

Rigid 2D-3D registration is an alternative to 3D-3D registration for cases where largely bony anatomy can be used for patient positioning in external beam radiation therapy. In this article, the authors evaluated seven similarity measures for use in the intensity-based rigid 2D-3D registration using a variation in Skerl's similarity measure evaluation protocol. The seven similarity measures are partitioned intensity uniformity, normalized mutual information (NMI), normalized cross correlation (NCC), entropy of the difference image, pattern intensity (PI), gradient correlation (GC), and gradient difference (GD). In contrast to traditional evaluation methods that rely on visual inspection or registration outcomes, the similarity measure evaluation protocol probes the transform parameter space and computes a number of similarity measure properties, which is objective and optimization method independent. The variation in protocol offers an improved property in the quantification of the capture range. The authors used this protocol to investigate the effects of the downsampling ratio, the region of interest, and the method of the digitally reconstructed radiograph (DRR) calculation [i.e., the incremental ray-tracing method implemented on a central processing unit (CPU) or the 3D texture rendering method implemented on a graphics processing unit (GPU)] on the performance of the similarity measures. The studies were carried out using both the kilovoltage (kV) and the megavoltage (MV) images of an anthropomorphic cranial phantom and the MV images of a head-and-neck cancer patient. Both the phantom and the patient studies showed the 2D-3D registration using the GPU-based DRR calculation yielded better robustness, while providing similar accuracy compared to the CPU-based calculation. The phantom study using kV imaging suggested that NCC has the best accuracy and robustness, but its slow function value change near the global maximum requires a stricter termination condition for an optimization method. The phantom study using MV imaging indicated that PI, GD, and GC have the best accuracy, while NCC and NMI have the best robustness. The clinical study using MV imaging showed that NCC and NMI have the best robustness. The authors evaluated the performance of seven similarity measures for use in 2D-3D image registration using the variation in Skerl's similarity measure evaluation protocol. The generalized methodology can be used to select the best similarity measures, determine the optimal or near optimal choice of parameter, and choose the appropriate registration strategy for the end user in his specific registration applications in medical imaging.

Fast generation of digitally reconstructed radiographs using attenuation fields with application to 2D-3D image registration

Generation of digitally reconstructed radiographs (DRRs) is computationally expensive and is typically the rate-limiting step in the execution time of intensity-based two-dimensional to three-dimensional (2D-3D) registration algorithms. We address this computational issue by extending the technique of light field rendering from the computer graphics community. The extension of light fields, which we call attenuation fields (AFs), allows most of the DRR computation to be performed in a preprocessing step; after this precomputation step, DRRs can be generated substantially faster than with conventional ray casting. We derive expressions for the physical sizes of the two planes of an AF necessary to generate DRRs for a given X-ray camera geometry and all possible object motion within a specified range. Because an AF is a ray-based data structure, it is substantially more memory efficient than a huge table of precomputed DRRs because it eliminates the redundancy of replicated rays. Nonetheless, an AF can require substantial memory, which we address by compressing it using vector quantization. We compare DRRs generated using AFs (AF-DRRs) to those generated using ray casting (RC-DRRs) for a typical C-arm geometry and computed tomography images of several anatomic regions. They are quantitatively very similar: the median peak signal-to-noise ratio of AF-DRRs versus RC-DRRs is greater than 43 dB in all cases. We perform intensity-based 2D-3D registration using AF-DRRs and RC-DRRs and evaluate registration accuracy using gold-standard clinical spine image data from four patients. The registration accuracy and robustness of the two methods is virtually identical whereas the execution speed using AF-DRRs is an order of magnitude faster.