Intraoperative cone-beam CT for image-guided tibial plateau fracture reduction

Abstract

Objectives: A mobile isocentric C-arm was modified in our laboratory in collaboration with Siemens Medical Solutions to include a large-area flat-panel detector providing multi-mode fluoroscopy and cone-beam CT (CBCT) imaging. This technology is an important advance over existing intraoperative imaging (e.g., Iso-C^3D), offering superior image quality, increased field of view, higher spatial resolution, and soft-tissue visibility. The aim of this study was to assess the system's performance and image quality in tibial plateau (TP) fracture reconstruction.

Methods: Three TP fractures were simulated in fresh-frozen cadaveric knees through combined axial loading and lateral impact. The fractures were reduced through a lateral approach and assessed by fluoroscopy. The reconstruction was then assessed using CBCT. If necessary, further reduction and localization of remaining displaced bone fragments was performed using CBCT images for guidance. CBCT image quality was assessed with respect to projection speed, dose and filtering technique.

Results: CBCT imaging provided exquisite visualization of articular details, subtle fragment detection and localization, and confirmation of reduction and implant placement. After fluoroscopic images indicated successful initial reduction, CBCT imaging revealed areas of malalignment and displaced fragments. CBCT facilitated fragment localization and improved anatomic reduction. CBCT image noise increased gradually with reduced dose, but little difference in images resulted from increased projections. High-resolution reconstruction provided better delineation of plateau depressions.

Conclusion: This study demonstrated a clear advantage of intraoperative CBCT over 2D fluoroscopy and Iso-C^3D in TP fracture fixation. CBCT imaging provided benefits in fracture type diagnosis, localization of fracture fragments, and intraoperative 3D confirmation of anatomic reduction.

• Cone-beam CT
• tibial plateau fracture
• intraoperative 3D imaging

Introduction

Tibial plateau fractures were originally termed fender or bumper fractures due to their association with automobile impacts. Lateral plateau fractures are the most common subgroup (75%), with trauma resulting from a valgus force to the knee. The Schatzker classification is used to categorize these fractures as depression, split, split depression, and condylar subtypes [1].

Tibial plateau fractures are common injuries that often affect patients during the most productive years of their lives with potentially devastating consequences. The goal of tibial plateau fracture fixation is anatomic restoration of the joint. As with any intra-articular fracture, inadequate treatment may result in joint instability and deformity coupled with a restricted range of motion [2] and eventual post-traumatic osteoarthritis. Secondary osteoarthritis following tibial plateau fracture was found in up to 44% of 131 cases at 7.6 years follow-up (range: 3.3–13.4 years). Associated ligamentous injuries, as well as postoperative infection, increased the incidence of secondary degeneration [3]. Increased articular step-off height can progressively increase valgus angulation and average contact area in the lateral compartment [4]. Although it is widely thought that anatomic reduction of the tibial plateau articular surface may prevent degenerative disease, there is no general agreement regarding the maximum acceptable step-off. Some clinical studies have suggested that maximum acceptable step-off of less than 5 mm is needed to obtain satisfactory results [5], [6], whereas others have suggested an upper limit of 10 mm [7]. Most authors agree that an acceptable range of intra-articular step-off is in the range of 2–10 mm [8–13]. Some authors also consider that another major indication for open reduction is more than 5 degrees of clinical valgus or varus instability [12], [14]. Long-term functional outcome studies have shown a clear benefit of tibial plateau fracture fixation, particularly in a younger patient population [15]. Therefore, anatomical reduction of the joint surface during open reduction and internal fixation of these fractures is crucial.

It is an ongoing challenge for surgeons to reduce these fractures with intraoperative certainty as to the quality of reduction. Intraoperative fluoroscopy is the main display modality for judging the quality and precision of reduction during tibial plateau fixation; however, the degree of articular depression is often under-appreciated on plain radiographs or fluoroscopy. While the fracture complexity and the spatial relation of fragments can be demonstrated with CT images and 3D reconstruction, this technique has to date been used solely in pre-planning for operative treatment and post-surgical assessment. Unsatisfactory reconstruction of the joint surface or incorrectly positioned screws are frequently discovered only after generating a post-operative CT scan. Such findings require an additional operative procedure or acceptance of an increased risk of future development of degenerative arthritis.

The use of intraoperative 3D imaging to help guide surgical instruments and check results of surgical procedures before closure has great potential for improving orthopedic outcomes. In the 1990s, CT guidance was used successfully in sacral and calcaneal fixation procedures [16], [17], though these procedures were performed within a CT suite rather than an operating theater. However, widespread use of intraoperative 3D imaging in orthopedic procedures requires that this capability be brought into the operating theater itself.

Recent developments have yielded new options for intraoperative 3D imaging. The SireMobil Iso-C^3D system (Siemens Medical Solutions, Erlangen, Germany), for example, combines the capabilities of routine intraoperative C-arm fluoroscopy with 3D reconstruction. The performance of the Iso-C^3D has been positively evaluated in several studies on the intraoperative display of high-contrast skeletal objects [18–20].

A new technology in 3D intraoperative imaging is offered by the application of flat-panel detectors (developed for radiographic/fluoroscopic imaging) to cone-beam CT. Cone-beam CT (CBCT) provides volumetric image reconstructions from 2D projections acquired across a given source-detector trajectory about the patient (e.g., a circular orbit). Development of flat-panel CBCT has progressed from benchtop prototypes[7], [13], [21], developed to test fundamental imaging performance, to pre-clinical systems under development for diagnostic [22], [23] and image-guided procedures [24–26]. Recently, this technology was adapted in the authors’ laboratory in collaboration with Siemens Medical Solution to a mobile isocentric C-arm platform (Siemens PowerMobil) as a proving ground for application in image-guided surgery and interventional radiology [24], [26–28]. Similar in some respects to other C-arm-based 3D imaging systems (e.g., Iso-C^3D)[6], [11], [29], there are important distinctions in this technology with respect to intraoperative guidance during orthopedic procedures. Intraoperative CBCT demonstrates isotropic, sub-millimetric 3D spatial resolution and soft-tissue detectability, and allows volumetric imaging of large anatomical sites within an open geometry.

This study aimed to determine the potential benefits of using intraoperative CBCT in the fixation of tibial plateau fractures. We hypothesize that intraoperative 3D CBCT imaging can improve the reduction of tibial plateau fractures by providing high-resolution images of the tibial plateau, including soft-tissue visualization and 3D localization for intraoperative guidance. In addition, the study evaluated the dose response and acquisition/reconstruction parameters required for visualization of the soft tissue and skeletal components around the knee, and compared the performance of the experimental prototype system for CBCT to other commercially available intraoperative techniques (viz., Iso-C^3D) for assessing the quality of tibial plateau fracture reduction.

Materials and methods

Imaging system

The 3D intraoperative imaging system used in this study was based on a Siemens PowerMobil mobile isocentric C-arm, modified to include a flat-panel detector in place of the X-ray image intensifier, a servo drive for orbital motion, a method for geometric calibration, and a computer control system for image acquisition and 3D reconstruction. The system operates in pulsed-fluoroscopic mode (1–6 p/s), with a magnification of 1.97× and a field of view at the isocenter of 20 cm × 15 cm. The PaxScan 4030CB flat-panel detector (Varian Imaging Products, Palo Alto, CA) is designed for real-time radiographic/fluoroscopic imaging, featuring a 2048 × 1536 pixel matrix (194 µm pixel pitch) in combination with a 0.6-mm-thick (∼270 mg/cm²) CsI:Tl X-ray converter (Figure 1). The panel can be read at frame rates up to 15 fps at full resolution, and up to 30 fps at half resolution (1024 × 768 pixels at 388 µm pitch). In this study the device was operated at half resolution and 6 fps (for fluoroscopy) and 3.3 fps (for CBCT acquisition).

Figure 1. (a) The C-arm imaging system: A host PC communicates with the the flat-panel detector (FPI) controller and the PowerMobil (via frame grabber and CANbus, respectively). The FPI controller in turn provides triggering of the X-ray generator in synchronization with FPI readout. Internal to the PowerMobil are a controller and microcomputer interfaced with the X-ray generator and orbital drive. (b) Operative set-up showing a knee specimen positioned near the isocenter of the C-arm with the assistance of back-pointing lasers mounted on the FPI. The magnified view shows one of the specimens following initial incision.

Cone-beam CT acquisition consisted of synchronized X-ray pulses and panel readout under continuous rotation of the C-arm. A typical acquisition required ∼60 seconds and consisted of 200 projections collected at 3.3 fps across the ∼178° orbit. Alternatively, “High-Speed” (100 projections acquired in ∼30 s) and “Slow” (500 projections in ∼150 s) acquisition modes were available. Volume reconstructions were formed by 3D filtered back-projection [30], with mechanical flex accommodated by a geometric calibration [31]. Image reconstruction required a total of ∼280 s for a (256 × 256 × 192) volume reconstructed from 200 projections on a Dell Precision 650 PC (Dual 2.0-GHz Xeon CPU with 3 GB RAM). Images were displayed on the host PC (immediately outside the operating room) and on an in-room display.

Fracture design

Three lower limbs were harvested from fresh-frozen cadaveric specimens (average age 82 years). Each tibia and fibula was osteotomized to retain the proximal 2/3 of their length. Each femur was osteotomized at the mid-portion of its length. The goal was to simulate a TP fracture without exposing the knee joint. This required the development of an appropriate TP fracture model. To create such a model, the proximal end of the femur and the distal end of the tibia and fibula were stripped of soft tissues and periosteum, potted in PMMA, and affixed to a servohydraulic testing machine (MTS Bionix 858) (Figure 2). An axial force was applied to the knee joint while maintaining full knee extension until the joint space was diminished on fluoroscopic images (indicating contact between the femur and tibial plateau). A valgus force was then applied laterally at the joint space level using a robust 5.5 kg steel cylinder. Lateral tibial plateau fractures were simulated in each of the three specimens and confirmed by fluoroscopy (Table I).

Figure 2. Fracture simulation set-up: Axial loading was applied to the extended knee on an MTS (Bionix 858 materials testing machine). Upon joint closure, confirmed by fluoroscopy, a lateral impact was applied to the specimen to create the tibial plateau fracture.

Table I.  Details of the simulated tibial plateau fracture patterns.

Specimen	Axial load (N)	Schatzker type	Description
1	4100	VI	Antero-lateral joint depression with medial involvement
2	2000	IV	Metaphyseal joint depression
3	960	III	Split joint depression

Performance in image-guided tibial plateau fracture fixation surgery

The specimens were fixed to the operating table with the knee joint aligned to the center of rotation of the C-arm (Figure 1b). A preoperative CBCT was acquired, confirming the type and severity of the tibial plateau fracture. The surgery was performed in the three specimens by a single orthopedic surgeon. Using the standard lateral tibial plateau approach, a hockey-stick incision was made extending proximally and laterally above the knee joint level, through Girdy's tubercle and distally towards the tibial shin. The tibialis anterior and the illiotibial band were stripped as a group and the fracture exposed and identified. The anterior horn of the lateral meniscus was detached from the tibial plateau by incising the coronary ligaments, the meniscus was tagged and retracted with a Vicryl 1 suture, and the lateral aspect of the plateau was exposed with the limitation of the approach.

The fracture was reduced using a bone punch to elevate fragments. The reduction was temporarily fixed with the use of 2-mm Kirschner wires placed using fluoroscopic guidance. Two fluoroscopy images, antero-posterior and lateral, were acquired. The images were analyzed by the surgeon to assess the continuity of the joint line. Based on these orthogonal fluoroscopic views, a decision was made as to whether tibial plateau alignment had been achieved or if further manipulation was required to improve the reduction. When it was determined from fluoroscopic imaging that a good reduction had been achieved, a CBCT scan was acquired. The information from the CBCT was then used to determine whether good alignment of the plateau had in fact been achieved, or if further reduction was needed based on visual clinical inspection of the scan.

Localization of remaining displaced bone fragments was made with the guidance of an aluminum Kirschner wire (KW) chosen to reduce metal artifact. This wire was localized based on fluoroscopic guidance close to the displaced fragment. A CBCT image was then obtained from which the exact placement of the KW was determined relative to the displaced fragments. Based on this information, a reduction tool was introduced relative to the placement of the KW and used to reorient the displaced fragments. At the end of the reduction a final set of fluoroscopic and CBCT images were obtained, confirming a satisfactory result.

Outcome parameters and images were collected to describe the ability of the system to diagnose fracture type preoperatively, assess the system's ability to deal with subtle fragments, and evaluate the differences between fluoroscopy and CBCT in assessing adequate fracture reduction.

Image quality assessment

Image quality was investigated as a function of dose, imaging speed, and choice of reconstruction filter. Image quality was assessed qualitatively in terms of the visibility of fine bony anatomy and soft-tissue contrast discrimination. Images of a single specimen (#1) were acquired at the conclusion of the fracture reduction, and image quality was evaluated in tri-planar views, with particular attention being paid to the region of the tibial plateau fracture.

Imaging dose was varied by nearly a factor of 50 through adjustment of tube current (mA), as in Table II (with X-ray energy fixed at 101 kVp and 200 projections acquired in each scan). Dose was computed from measurements of exposure using an ionization chamber [27]. CBCT images were equivalently windowed and leveled throughout for fair intercomparison.

Table II.  Image quality assessment parameters examining acquisition speed and dose.

Trial	Scan speed	Projections	X-ray technique	Dose (mGy)
1	Slow	500	101 kVp; 4.6 mA	11.7
2	Fast	200	101 kVp; 4.6 mA	4.7
3	Fast	200	101 kVp; 2.9 mA	3.0
4	Fast	200	101 kVp; 1.8 mA	1.8
5	Fast	200	101 kVp; 0.9 mA	0.9
6	Fast	200	101 kVp; 0.5 mA	0.5
7	Fast	200	101 kVp; 0.1 mA	0.1

Imaging speed is dependant on the number of projections acquired and the voxel size in 3D reconstructions. Imaging speed was varied according to the number of projections acquired in the CBCT scan: “Fast” corresponded to 200 projections in 60 s; “Slow” to 500 projections in 150 s. Each scan was acquired at 101 kVp and 4.6 mA. Voxel size in 3D reconstructions was varied from 0.4 to 0.8 mm, yielding high (512 × 512 × 384) and low (256 × 256 × 192) voxel reconstructions, respectively. Reconstruction times using a research system for 3D filtered back-projection (not optimized for speed) were ∼13.8 min and ∼4.7 min, respectively. Faster 3D reconstruction algorithms are commercially available (e.g., Exxim Cobra, San Mateo, CA), capable of reconstruction times of ∼80 s and ∼17 s, respectively; these are more consistent with the demands of clinical implementation.

The effect of reconstruction filter selection on image quality was examined for two choices of convolution kernel: “Smooth” (low-pass filter to reduce image noise and increase soft-tissue detectability) and “Sharp” (higher-pass filter to enhance image sharpness). Each reconstruction filter consists of a ramp filter in combination with a Gaussian apodization kernel [27].

Iso-C^3D comparison study

The image quality achieved using the flat-panel detector-based system (Figure 1) was compared to a commercially available X-ray image-intensifier-based system (SireMobil Iso-C^3D, Siemens Medical Systems, Erlangen, Germany). The same cadaver specimen was used as in the image-quality studies described above. Images were acquired with the Iso-C^3D at an equivalent kVp to the experimental C-arm, with the mA adjusted to give an equivalent dose. Image quality was compared in terms of spatial resolution, soft-tissue visibility, and field of view. Tri-planar views as well as volumetric renderings were compared, along with magnified views in the area of the tibial plateau fracture.

Results

Fracture design

Tibial plateau fracture patterns were simulated in all three specimens and confirmed using fluoroscopy and CBCT. Axial loading required to close the joint space ranged from 960 N to 4100 N (Table I). The tibial plateau fracture patterns generated ranged from Schatzker type III to type VI.

Performance in image-guided surgery

For each case presented below, the C-arm system was used to obtain both 2D fluoroscopic and 3D cone-beam CT images for purposes of intraoperative guidance of tibial plateau fracture reconstruction. CBCT was found to reveal subtle fragments normally beyond the limit of fluoroscopic visualization, and allowed a finer level of fracture reduction than was otherwise possible.

Case 1

The preoperative CBCT image acquired on the C-arm system just prior to reduction (Figure 3) illustrates a fairly comminuted Schatzker VI fracture with metaphyseal bone compression. The shaded surface rendering (Figure 3a) demonstrates the uniform quality and sub-millimetric 3D spatial resolution across a fairly large field of view. Coronal images (Figure 3b) show the tibial plateau fracture clearly. Following initial fracture reduction, a fluoroscopic image was acquired with a KW inserted for fixation and localization (Figures 4a and 4b). The fluoroscopic images demonstrated satisfactory reduction. However, acquisition of an intraoperative CBCT scan (Figure 4c) revealed a remaining displacement of the articular surface located just lateral to the tibial spine. At this stage, a 3D image was reconstructed and the location of the fixation KW in the 3D scan allowed for guided localized manipulation of the articular surface. A subsequent CBCT was obtained, showing an anatomical reduction of the joint surface (Figure 4d). Normally, in order to fill the gap created by the bone compression, a bone graft or a bone graft substitute, with subsequent fixation, would follow this operative stage.

Figure 3. Case 1. (a) Example image illustrating uniform volumetric image quality and sub-millimetric spatial resolution across a fairly large field of view. (b) Example CBCT image (coronal view) demonstrating the fracture type before reduction. Note the amount of compression and the irregularity of the articular surface.

Figure 4. After the first attempt at reduction, orthogonal fluoroscopic views (a and b) were obtained, and the reduction was judged to be acceptable. However, an intraoperative CBCT scan (c) acquired at the same point in the procedure demonstrates a subtle articular fragment that could be reduced further. This fragment and the articular surface were not observed on the fluoroscopy images. (d) CBCT image (coronal view) illustrating improved reduction of the articular surface after further manipulation based on the intraoperative image in (c).

Case 2

After initial reduction, fluoroscopic imaging showed room for improved reduction. This was confirmed by CBCT (Figure 5a). There was a need for localization of a small fragment, which is often very difficult to achieve with fluoroscopic guidance alone. Using CBCT, bone fragments could be localized in relation to an aluminum KW included in the scan, which indicated that the lesion was anterior and lateral to the tip of the pointer (Figure 5b). With the KW in place, a fine bone punch was introduced based on the relative placement of the KW and an additional reduction was made, demonstrating an excellent anatomical alignment of the articular cartilage (Figure 5c).

Figure 5. (a) Example coronal image obtained after fracture reduction and placement of a KW in case 2. (b) Localization of the small, displaced fragment was performed using an aluminum KW. The displaced fragment was localized antero-lateral to the KW tip, as shown in the example axial CBCT image. (c) The displaced fragment was manipulated in relation to the aluminum KW, and a near-perfect reduction of the articular surface was achieved under CBCT guidance.

Case 3

The fracture reduction in this case was straightforward and the first fluoroscopic image showed satisfactory reduction. The CBCT in this case confirmed the fluoroscopic finding and demonstrated fine details of the joint surface reduction.

In all cases, the final CBCT scan provided exquisite visualization of articular details in assessing the quality of reduction. The CBCT scan also provided assurance that there was no intra-articular debris or hardware, thus avoiding the need for a post-operative CT scan.

Image quality evaluation

Effect of dose

Visualization of the tibial plateau was examined in axial, coronal and sagittal views at dose levels spanning the operational range of the system. At the nominal dose (4.7 mGy) we observed fair delineation of fine, high-contrast structures as well as soft tissues (fat, muscle). The differentiation of soft tissues was best seen in the sagittal and magnified views. Image noise increased gradually with reduced dose, although the visibility of fine and low-contrast structures was maintained down to ∼0.5 mGy (Figure 6), below which image quality rapidly degraded (as compared to ∼10 mGy for a diagnostic CT scan).

Figure 6. Each column illustrates a CBCT sagittal view, with a magnified portion in the region of the fracture shown in the bottom row. At the nominal dose level (4.7 mGy), excellent delineation of fine, high-contrast structures as well as soft tissues is observed. Below ∼0.5 mGy, image quality rapidly degrades due to increased image noise.

Effect of acquisition and reconstruction parameters

The image acquisition time increased from 60 s for the Fast (200 projection) scan to 150 s for the Slow (500 projection) scan, with a resultant 2.5-fold increase in radiation dose. Image reconstruction time depends on both the number of projection images and the voxel size (0.8 mm for “Nominal”, and 0.4 mm for “High” resolution). The High-resolution reconstructions (512 × 512 × 384 voxels) (Figure 7b) provide excellent delineation of the joint depression, whereas voxelation in the Nominal case (256 × 256 × 192 voxels) (Figure 7a) somewhat limits the detectability of fine and subtle features. There is little appreciable difference between the Fast and Slow acquisition images (Figures 7b and 7c, respectively). Considering that the latter involves a longer acquisition time (and a possible 2.5-fold increase in radiation dose), there is little motivation to use the Slow acquisition for intraoperative knee imaging; the Fast mode was sufficient for visualizing bony and soft-tissue structures.

Figure 7. Example images acquired at various speeds and voxel sizes. (a) “Fast” acquisition (200 projections in 60 s) reconstructed at “Nominal” resolution (0.8 mm voxels). (b) “Fast” acquisition reconstructed at “High” resolution (0.4 mm voxels). (c) “Slow” acquisition (500 projections in [1]50 s) reconstructed at “High” resolution. Fracture details are seen in both the Nominal and High resolution reconstructions; however, fine and subtle features are not well delineated in the former. Little appreciable difference is seen between the Fast and Slow acquisition images. Therefore, “Fast/High” (b) gives a reasonable choice of acquisition/reconstruction parameters for the given imaging task.

Effect of reconstruction filter

Although the CBCT images reconstructed with the “Smooth” filter (Figure 8a) improved delineation of soft-tissue structures, image quality overall was judged superior for the “Sharp” filter (Figure 8b). Images reconstructed with the "Sharp" filter improved visualization of subtle fracture components and trabecular structure, and this was the nominal choice throughout this investigation, unless otherwise noted.

Figure 8. Example images reconstructed using (a) “Smooth” and (b) “Sharp” reconstruction filters. Overall image quality is superior for the Sharp filter, particularly for identification of subtle fracture components and trabecular structure. The Smooth filter provides somewhat improved delineation of soft-tissue structures.

Comparison with Iso-C^3D

A pronounced difference in all aspects of image quality was observed between the flat-panel CBCT experimental prototype and the Iso-C^3D (Figure 9). Coronal and sagittal views are shown in similar regions of the anatomy; slices through identical regions were difficult to obtain due to variations in positioning of the specimen. The experimental system exhibited higher spatial resolution and provided significantly finer visualization of bony detail. For example, trabecular structure that was invisible in Iso-C^3D images (Figure 9a) was clearly visualized in the CBCT images (Figure 9c). The system also exhibited significantly increased contrast and soft-tissue visualization. For example, clear delineation of muscle and fat was seen in the CBCT images (Figure 9c), but was indiscernible in the Iso-C^3D images. The increased field of view was an additional advantage – approximately 12 × 12 × 12 cm for the Iso-C^3D, compared to 20 × 20 × 15 cm for the experimental system. The full range of imaging techniques available with the Iso-C^3D was explored (e.g., 62–110 kVp, tube mA ranging from the lowest possible setting up through sensor saturation, Fast and Slow acquisition modes, etc.). The best Iso-C^3D image quality achieved is presented for comparison with the CBCT images (Figure 9a). Considering these results in relation to the image-quality study (Figure 6), we note consistently superior image quality for the experimental CBCT system with all available techniques.

Figure 9. Comparison of image quality between a commercially available system based on an X-ray image intensifier (Iso-C^3D) (a and b) and the experimental prototype incorporating a flat-panel detector for CBCT (c and d). Coronal and sagittal views are shown. X-ray technique was matched in terms of kVp, with imaging dose adjusted by variation of tube mA. For the Iso-C^3D, the best possible image quality was obtained using a “Slow” acquisition (doubling the number of projections and the imaging dose); imaging dose was estimated at 5 mGy. Images for the experimental system correspond to “Fast” acquisition (fewer projections) and “Nominal” reconstruction (0.8 mm voxels), with an imaging dose of 4.7 mGy.

Discussion

The development of novel imaging techniques and modalities for the guidance of interventional procedures represents an important area of medical research. Intraoperative visualization in orthopedic surgery is an ongoing challenge, particularly in complex fracture reduction. The tibial plateau fracture is a common and well-studied intra-articular fracture that requires anatomic reduction of the joint surface. The sequel of malreduction is osteoarthritis of the knee joint. The key for successful reduction is good visualization of the joint surface. In order to visualize the joint surface directly, the lateral meniscus is normally detached from the lateral tibial plateau prior to fracture reconstruction. This window allows the lateral surface of the joint to be partially visualized. In fractures involving central or medial aspects of the joint, however, this approach does not allow for sufficient direct visualization.

Indirect visualization is currently used intraoperatively in the reconstruction of tibial plateau fractures through 2D fluoroscopic imaging. However, using direct and indirect 2D visualization, a poor clinical outcome of post-traumatic osteoarthritis is still observed in up to 44% of patients [3]. This study has demonstrated a clear advantage of CBCT over the common practice of 2D fluoroscopy in the reconstruction of tibial plateau fractures. In addition, the image quality achieved with CBCT is superior to that of current commercially available systems (viz., Iso-C^3D), particularly in the improved field of view, articular detail, and soft-tissue delineation. The field of view becomes a significant concern in imaging large volumes, such as compound long bone or pelvic fractures. While soft-tissue visibility was not an important factor in fracture reduction in this study, it becomes more significant when the fracture occurs in proximity to complicated soft-tissue anatomical structures, such as blood vessels and nerves. Furthermore, the potential for soft-tissue visualization is consistent with the desire for techniques that are increasingly minimally invasive and spare soft tissues by precise intraoperative imaging and guidance.

Aside from fluoroscopy, a combination of knee arthroscopy together with a limited open approach has been used to aid in intraoperative visualization of the joint surface in the fixation of tibial plateau fractures [29], [32]. The use of this technique has proven to be successful in certain types of fractures; however, arthroscopy is relatively contraindicated in cases of capsular tears, and visualization is difficult with large metaphyseal involvement. Furthermore, the use of this modality requires specialized arthroscopic training and adds to both the cost and time of the operation. Use of high-resolution 3D imaging is rapid, less cumbersome, and does not require special training. CBCT offers visualization of all fracture components (not only those visible through the arthroscope), allows localization of fracture fragments, and can confirm placement of surgical implants.

The intraoperative decision as to when anatomical reduction has been achieved is affected by the imaging information available. 3D high-resolution imaging of the tibial plateau by CBCT indicated that additional improvements to the joint surface could be made, as compared to images acquired with fluoroscopy alone. CBCT also provided information on the localization of displaced fragments and guidance for improved reconstruction. These results, while partially qualitative, provide valuable insight regarding the strengths and advantages offered by CBCT for surgical guidance.

Image quality across the three cases described was highly illustrative and enabled visualization of osseous anatomy, joint surface, subtle fragments, surgical tools, and some of the soft tissue components. The choice of image acquisition mode (e.g., Fast or Slow acquisition) and voxel size in image reconstructions (e.g., Nominal or High resolution) depends on both the imaging task (i.e., what the surgeon is attempting to ascertain from the image data) and the procedural constraints (e.g., the amount of time available for acquisition/reconstruction of intraoperative images). For example, images at higher dose and spatial resolution (requiring more time) might be acquired immediately prior to and following the procedure, providing the surgeon with high-quality “preoperative” and “postoperative” images, while “intraoperative” images would more likely be acquired and reconstructed according to more stringent time constraints and a specific imaging task.

A new fracture model was used to simulate TP fracture, not previously described in the literature. The model was developed in order to generate a fracture at the tibial plateau in an indirect fashion without exposing the joint surface, allowing reduction of the fracture without direct visualization. The fractures created demonstrated high levels of metaphyseal compression but relatively small areas of joint depression, probably due to the reduced bone quality of the specimens associated with their age. In cases of larger or multiple areas of depression, the information added by CBCT may be even more valuable in achieving adequate joint surface reconstruction. The present model did demonstrate, however, the benefit of CBCT in reconciling small and displaced fracture fragments. Aside from joint depression, in all three cases CBCT identified displacement of additional fragments normally not visualized during fluoroscopy. Such benefits would not be reflected if the ability of CBCT were quantified using joint depression measurements alone. Further study is needed to correlate the effect of these reductions on clinical outcomes in patients.

The fully 3D nature of CBCT image reconstructions allows tri-planar and volumetric views that increase diagnostic certainty and verification, improve spatial understanding of the injury during surgery, and facilitate integration with navigational tools. During joint reconstructive surgeries, there is a frequent need for intraoperative localization of bone fragments. To assess localization using fluoroscopy requires multiple projections. Current fluoroscopy allows only single real-time projections (antero-posterior, lateral, or oblique), yet there is a frequent need to go back and forth from one projection to another to get sufficient localization. Using simultaneous multiple projections or a 3D model can provide a much more straightforward method for fragment localization. A simple aluminum pointer was used in conjunction with CBCT scanning to assist in the localization of displaced fragments. Localization of fragments relative to the placement of the fixed pointer was easily achieved in three dimensions and successful improvements in joint surface realignment were achieved using this technique. Integration of the imaging system with navigational tracking is an obvious and important next step. In fractures with a depressed surface in a more medial location, the localization potential of CBCT gains even more importance, as direct visualization is reduced and more reliance is placed on imaging data.

Previously, the only precise way to verify intra-articular fracture reduction outcome was through obtaining a postoperative CT to reveal inappropriate fracture reduction or intra-articular hardware penetration, which in the worst-case scenarios would require an additional operative procedure. Obtaining an intraoperative CBCT precludes this unfortunate scenario and allows a high level of confidence regarding the quality of reduction and placement of hardware.

Using the 3D localization modality of CBCT would also be extremely valuable in bone tumor surgeries. In cases of osteoid osteoma, lesions are often very small (< 1cm³), yet the success of surgery depends on precise localization of the center of the lesion (calcified nidus). Once localization is achieved, a percutaneous excision of the lesion can be done using a simple bone mill. Presently, due to the need for precise 3D localization of the lesion site, these surgeries are carried out within a CT suite [33]. CBCT technology would enable osteoid osteoma excision to be performed within an OR suite, providing a much less cumbersome alternative.

Intraoperative CBCT imaging has particular promise in improving the ability to use minimally invasive surgical procedures in orthopedics. Less direct visualization can be compensated for by high-resolution 3D images. This applies not only to implementation of CBCT alone, but also to its use in combination with computer-assisted navigation tools. CBCT allows registration of smaller fracture fragments that are not detected with 2D fluoroscopy systems, resulting in improved joint surface reconstruction.

Conclusion

Intraoperative CBCT is a useful modality for visualization of tibial plateau fractures. In this study we have demonstrated the benefits of CBCT in diagnosing, identifying, and manipulating fracture fragments. Intraoperative, rather than postoperative, confirmation of the reduction and fixation can ensure a high-quality reduction, without which long-term clinical outcome may be compromised or additional surgical intervention required. Further studies are needed to provide a statistically significant evaluation of CBCT versus conventional fluoroscopy; to determine the percentage of malalignments that are overlooked or unidentifiable on fluoroscopy; to evaluate the advantages of CBCT in other orthopedic surgery applications; and to relate them to clinical outcomes.

Acknowledgments

This work was supported by the National Institutes of Health (R01-CA89081-04) and conducted in collaboration with Siemens Medical Systems, Special Products Group (Erlangen, Germany). The technical expertise of Dr. Rainer Graumann and Dr. Dieter Ritter (Siemens Medical Systems) is gratefully acknowledged. Dr. S.M. Kim (Princess Margaret Hospital) assisted with the C-arm calibration and measurements. Mr. S. Ansell and Mr. G. Wilson (Princess Margaret Hospital) assisted with image acquisition and reconstruction. Image renderings were performed with software provided by Dr. B. Davey and Dr. V. Accomazzi (Cedara Software Corp.). The Iso-C^3D measurements were performed at the Department of Radiation Oncology, William Beaumont Hospital (Detroit, MI), with the expert assistance of Dr. E.P. Armour (William Beaumont Hospital).