Summary
Augmented reality technology was applied to core decompression for osteonecrosis of the femoral head to realize real-time visualization of this surgical procedure. This method can effectively improve the safety and precision of core decompression.
Abstract
Osteonecrosis of the femoral head (ONFH) is a common joint disease in young and middle-aged patients, which seriously burdens their lives and work. For early-stage ONFH, core decompression surgery is a classical and effective hip preservation therapy. In traditional procedures of core decompression with Kirschner wire, there are still many problems such as X-ray exposure, repeated puncture verification, and damage to normal bone tissue. The blindness of the puncture process and the inability to provide real-time visualization are crucial reasons for these problems.
To optimize this procedure, our team developed an intraoperative navigation system on the basis of augmented reality (AR) technology. This surgical system can intuitively display the anatomy of the surgical areas and render preoperative images and virtual needles to intraoperative video in real-time. With the guide of the navigation system, surgeons can accurately insert Kirschner wires into the targeted lesion area and minimize the collateral damage. We conducted 10 cases of core decompression surgery with this system. The efficiency of positioning and fluoroscopy is greatly improved compared to the traditional procedures, and the accuracy of puncture is also guaranteed.
Introduction
Osteonecrosis of the femoral head (ONFH) is a common disabling disease occurring in young adults1. Clinically, it is necessary to determine the staging of ONFH based on X-ray, CT, and MRI to decide the treatment strategy (Figure 1). For early-stage ONFH, hip preservation therapy is usually adopted2. Core decompression (CD) surgery is one of the most frequently used hip preservation methods for ONFH. Certain curative effects of core decompression with or without bone grafting in treating early-stage ONFH have been reported, which can avoid or delay subsequent total hip arthroplasty (THA) for a long time3,4,5. However, the success rate of CD with or without bone grafting was reported differently among previous studies, from 64% to 95%6,7,8,9. The surgical technique, especially the accuracy of drilling position, is important for the success of hip preservation10. Due to the blindness of the puncture and positioning procedure, the traditional techniques of CD have several problems, such as more fluoroscopy time, repeated puncture using Kirschner wire, and injury of normal bone tissue11,12.
In recent years, the augmented reality (AR)-assisted method has been introduced in orthopedic surgery13. The AR technique can visually show the anatomy of the surgical field, guide the surgeons in planning the operating procedure, and consequently reduce the difficulty of the operation. The applications of the AR technique in pedicle screw implantation and joint arthroplasty surgery have been reported earlier14,15,16,17. In this study, we aim to apply the AR technique to the CD procedure and verify its safety, accuracy, and feasibility in clinical practice.
System hardware components
The main components of the AR-based navigation surgical system include the following: (1) A depth camera (Figure 2A) installed directly above the surgical area; the video is shot from this and sent back to the workstation for registration and cooperation with the imaging data. (2) A puncture device (Figure 2B) and a non-invasive body surface marking frame (Figure 2C), both with passive infrared reflectors. A special reflective coating of marking balls (Figure 3) can be captured by infrared equipment to achieve accurate tracking of surgical equipment in the surgical area. (3) An infrared positioning device (Figure 2D) is responsible for tracking markers in the surgical area, matching the body surface marking frame and puncture device with high accuracy (Figure 4). (4) The host system (Figure 2E) is a 64-bit workstation, installed with the independently developed AR-assisted orthopedic surgery system. Augmented reality display of hip joint and femoral head puncture operation can be completed with its assistance.
Subscription Required. Please recommend JoVE to your librarian.
Protocol
This study was approved by the ethics committee of the China-Japan Friendship Hospital (approval number: 2021-12-K04). All of the following steps were performed according to standardized procedures to avoid injury to the patients and the surgeons. Informed patient consent was obtained for this study. The surgeon must be skilled in conventional core decompression procedures to ensure that the surgery can be performed in a traditional way in case of inaccurate navigation or other unexpected situations.
1. Preoperative diagnosis and grading of ONFH
- Identify patients with clinical symptoms of ONFH; typical symptoms such as persistent or intermittent pain in the groin region with ipsilateral hip or knee radiating pain. Physical examination showed deep tenderness in the groin region, Patrick's sign, a limited hip motion of internal rotation and abduction, or necrosis changes of the femoral head measured using X-ray, CT, and MRI.
- According to the association research circulation osseous (ARCO) staging, review patients' X-ray, CT, and MRI of the hip and determine the staging of ONFH. Two doctors conduct this work independently. If disagreements arise, ask a third expert to make the final decision.
- Record the preoperative visual analogue scale (VAS) and Harris hip score using a questionnaire.
- Include patients using the following criteria: 1) patients with ONFH; 2) stage I, IIA, and IIB of ONFH confirmed by imaging examination (X-ray, CT, and MRI); 3) the femoral head core decompression surgery is planned. Exclude patients when: 1) patients reject the CD surgery; 2) preoperative routine examination indicates surgical contradictions, such as infection or poor basic condition; 3) patients refuse to be enrolled in the group.
2. System registration and accuracy testing
- Run the AR-assisted orthopedic surgery system (due to commercialization issues software details cannot be provided) and click on Orthographic Video to activate the depth camera. An image of the surgical area will be displayed on the screen after activation (Figure 5A). Position the optical tracking device so that its tracking area can completely cover the surgical area (Figure 5B).
- Click on NDI Setting to select the device access port, COM4. Click the Virtual Needle Length Setting (generally a Kirschner needle is 180 mm long) and a virtual Kirschner needle image will be automatically generated in the surgical area in the video.
- Divide the planned surgical frontal area into upper and lower levels with each level 30 cm x 30 cm in size, and with a height difference of 15 cm between the levels. The system automatically inputs this spatial information of the surgical area in the software.
- Evenly allocate every level with 10 matching points; for every 30 cm x 30 cm area, divide it into three equal parts, with two parts having three points each, and one part (left part) having four points. Ask the assistant to place the non-invasive body surface marking frame (Figure 2C) according to the points. Once done, click on Match. The system's own special image for registration will be automatically superimposed on the marking frame (Figure 5C). Consider the registration of this point successful when the image and the marking frame coincide completely.
- Move the frame to the next registration point and repeat step 2.4. until all registration points are completed. As the shape of the marking frame equipped with the puncture device (Figure 3A2) is exactly the same as the non-invasive body surface marking frame, once the registration is completed, the former can also be tracked by the optical tracking device in the surgical area.
- Move the puncture device randomly in the surgical area to detect the matching degree of virtual needle and tracking delay (Figure 6). As the red-blue virtual Kirschner needle body automatically fits with the actual needle in the surgical area, the augmented reality display of the Kirschner needle is successful (Figure 5D).
NOTE: During the registration process, the position of the optical tracking device and depth camera should not be changed at will. If so, the spatial position relationship of virtual surgery will change, causing inaccurate matching between the virtual Kirschner needle and the physical one, and registration has to be re-conducted.
3. Patient and system preparation before puncture
- After entering the operating room, ask the patient to lie down in supine position and fix the lower limb of the affected side (Figure 7). Administer general anesthesia to all patients.
- Prepare the surgical site with iodine and 75% alcohol, and place the non-invasive body surface positioning device (sterilized using standard procedures) on the patient's affected hip.
- Move the C-ARM fluoroscope to the side of the operating table and position the source above the hip joint. Align the source with the depth camera and record the position of the surgical table as position 1.
- After the first fluoroscopy, export the BMP format radiograph to the system workstation, open it in Photo Editing, and adjust its gray scale by clicking on the Light Scale Option. Rotate clockwise and flip horizontally once by clicking the corresponding buttons to convert to BMP. Then, open it by clicking Painting 3D and save as the JPG format, which contained non-invasive body surface marking frame, and name it image 1 (Figure 8A).
NOTE: This conversion process is to promote the success of the system identification. Because of the special requirements of image conversion, it is necessary to adjust the gray scale of the X-ray image for rotation and inversion. - Slide the operating table directly below the depth camera to the operating area marked as position 2. Position 1 (in step 3.3) and position 2 are two points on the same horizontal plane, 30 cm apart.
- In the AR-assisted orthopedic surgery system, click File > Front X-ray image, and select image 1. the system automatically identifies the non-invasive body surface marking frame on the skin surface of the patient, and then superimposes this image to the hip joint in the surgical video (Figure 8B).
- Using the augmented reality display of the X-ray image and real-time video generated above, the surgeon plans the puncture path based on this.
4. Puncture assisted by surgical system
- The surgeon stands on the affected side and performs the following steps. Hold the puncture device and determine the best insertion angle. Mark the insertion point on the skin surface, guided by the virtual Kirschner wire and the X-ray image of the hip joint in the surgical video.
- Select a Kirschner wire with a diameter of 2.5 mm and pierce it from the insertion point. Observe the insertion depth and angle in the video and adjust it timely.
- When the virtual needle has reached the target area of necrosis, stop the puncture process and retain the screen shot as image 2 (Figure 9A) for subsequent puncture accuracy evaluation.
- Indwell the needle. Move the operating table to position 1 for the second fluoroscopy to verify the actual puncture condition of the Kirschner wire. Record the image file as image 3 (Figure 9B).
- Puncture is successful when the location of the Kirschner wire meets all of the surgeon's requirements. Then, use the lancet to cut the skin around the needle, and separate every level of soft tissue till exposing sub-trochanter bone, roughly to a depth of 3 cm. Drill into the necrotic area along the Kirschner wire with a 5 mm trephine to complete the subsequent operations (artificial bone or autologous bone implantation).
- After finishing all the procedures, close the skin with 3-0 silk thread and cover with sterile dressing (Figure 10). After returning to the ward, provide the patients accepted common orthopedic postoperative medication, like infection prevention, analgesia, and fluid infusion. If no complication occurs, discharge the patients 3 days after surgery.
5. Operation evaluation
- Import image 2 and image 3 into an image processing software and adjust the opacity to 52%.
- Click the Masking button to overlap the two images, then click the Rulers button to measure the distance (Lvirtual) between the virtual tip and the puncture point of the femoral cortex, and the distance (Lture) between the tip of the Kirschner needle and the puncture point of the femoral cortex. Calculate the difference between Lvirtual and Lture to assess puncture accuracy.
- During the puncture, measure the positioning time as follows: the positioning time starts from the time when the Kirschner wire pierces the skin, and stops when the X-ray confirms that the Kirschner wire has successfully reached the target area of the femoral head.
- Three months after surgery, take the hip X-ray (Figure 11) and record the visual analogue scale and Harris hip score.
Subscription Required. Please recommend JoVE to your librarian.
Representative Results
Operation characteristics
The surgical navigation system was applied in continuative 10 hips of nine patients. The average total positioning time of the surgery was 10.1 min (median 9.5 min, range 8.0-14.0 min). The mean C-ARM fluoroscopies was 5.5 times (median 5.5 times, range 4-8 times). The mean error of puncture accuracy was 1.61 mm (median 1.2 mm, range -5.76-19.73 mm; Table 1). The results show that the positioning time and fluoroscopy times are obviously shortened compared to traditional procedures.
Clinical outcome evaluation
The nine enrolled patients consisted of seven males and two females, with an average age of 41.6 ± 10.0 years.The mean BMI was 23.93 ± 3.08 kg/m2. For the hips evaluated, two hips were in ARCO I stage, four hips were in ARCO IIA stage, and four in ARCO IIB stage. Preoperative and postoperative visual analogue scale and Harris hip score were used to evaluate the outcome (Table 1). The mean preoperative VAS score was 6 and mean postoperative score was 3.75. The average preoperative Harris score was 77.5 and mean postoperative score was 85.5. Hip X-ray was examined 3 months after the surgery. All the patients returned to the ward safely. No postoperative complications such as infection, hematoma, or nerve damage were found. So far, no femoral head collapse occurred in any cases, and the long-term function and success rate of hip preservation are still being followed up. Surgical indicators and scores are shown in Table 2.
Figure 1: Imaging of early stage of femoral head necrosis. (A) The CT image. (B) The MRI image. Arrows indicate areas of necrosis. Please click here to view a larger version of this figure.
Figure 2: The main components of the AR-based navigation surgical system. (A) Depth camera. (B) The puncture device with a positioning frame. (C) Non-invasive body surface marking frame independently designed and developed. (D) Infrared positioning device. (E) The surgical system workstation. Please click here to view a larger version of this figure.
Figure 3: Installation of a passive infrared reflector. (A) Self-designed positioning frame mounted on puncture device. (B) The reflector is mounted at the four corners of the non-invasive body surface marking frame. (C) The specification of the passive infrared reflector is a spherical device with a diameter of 10 mm. Please click here to view a larger version of this figure.
Figure 4: Working principles of infrared positioning device. Infrared radiation emitted by the infrared positioning device is reflected by the passive infrared reflectors; the receivers in that device receive the reflected signal and transmit the movement data to the workstation. Please click here to view a larger version of this figure.
Figure 5: An overview of the preoperative registration process. (A) Operating interface of AR-assisted orthopedic surgery system. (B) The surgical area was planned using a non-invasive body surface marking frame. (C) Tips for successful registration of one of the matching points in the surgical video. (D) After all the matching points were successfully matched, the tracking of surgical instruments was tested. Please click here to view a larger version of this figure.
Figure 6: Superposition of virtual Kirschner wire on real Kirschner wire. (A-C) The images show that the virtual Kirschner needle is precisely superimposed on the physical one and moves with it in the screen. Please click here to view a larger version of this figure.
Figure 7: An overview of surgical scenarios. (A) The main components of the AR-based surgical system in the operating room. (B) A patient with necrosis of the femoral head is being treated with the aid of the surgical system. Please click here to view a larger version of this figure.
Figure 8: Hip joint imaging and augmented reality display. (A) Radiograph of the hip joint containing a non-invasive body surface marking frame. The black arrow indicates the passive infrared reflectors. (B) Radiograph is processed at the workstation and then superimposed by the surgical system on the surface of the affected hip on the screen. Please click here to view a larger version of this figure.
Figure 9: Puncture effect demonstration. (A) Image presents the screenshot after the puncture, the black-red-blue line is a virtual Kirschner wire in the system (step 2.6). (B) Image shows the hip radiograph after completion of puncture, the black line is an image of a real Kirschner wire in X-ray. Please click here to view a larger version of this figure.
Figure 10: Femoral head puncture guided by the AR-based surgical system. (A) The surgeon is adjusting the position of the puncture device according to the screen display. (B) Kirschner wire punctures the skin and points to the necrosis. (C) Drill into the necrotic area along the Kirschner wire with a 5 mm trephine to fill artificial bone or autologous bone implantation. (D) Close the wound. Please click here to view a larger version of this figure.
Figure 11: Postoperative radiograph of the hip joint. (A) From the front view. (B) The patient is in frog position. The black arrows indicate artificial bone implants in the femoral head. Please click here to view a larger version of this figure.
| Case | Sex | Age | BMI | Disease | ARCO |
| 1 | M | 22 | 28.40 | ONFH(left) | IIA |
| 2 | F | 21 | 22.40 | ONFH(right) | IIB |
| 3 | M | 42 | 19.56 | ONFH(left) | IIB |
| 4 | M | 51 | 22.10 | ONFH(left) | I |
| 5 | M | 31 | 24.34 | ONFH(bilateral) | L:IIB |
| 6 | R:IIA | ||||
| 7 | M | 46 | 27.24 | ONFH(right) | IIA |
| 8 | F | 41 | 21.20 | ONFH(left) | IIB |
| 9 | M | 56 | 22.83 | ONFH(right) | I |
| 10 | M | 38 | 27.30 | ONFH(left) | IIA |
Table 1: Basic patient information. The table provides the information for the 10 patients enrolled in this study.
| Case | Positioning time(min) | Fluoroscopy shots | Positioning error(mm) | Harris Hip Score | Visual Analogue Scale | ||
| Before | After | Before | After | ||||
| 1 | 13 | 6 | 2.83 | 82 | 89 | 6 | 4 |
| 2 | 9 | 6 | 0.35 | 86 | 85 | 4 | 3 |
| 3 | 9 | 4 | 2.05 | 88 | 89 | 5 | 3 |
| 4 | 10 | 5 | -5.01 | 73 | 85 | 7 | 4 |
| 5 | 8 | 6 | -1.52 | L:84 | L:88 | L:4 | L:3 |
| 6 | 14 | 4 | -4.13 | R:68 | R:82 | R:6 | R:4 |
| 7 | 11 | 7 | 3.97 | 74 | 84 | 7 | 4 |
| 8 | 10 | 5 | 3.55 | 81 | 89 | 5 | 3 |
| 9 | 9 | 8 | 19.73 | 74 | 82 | 6 | 4 |
| 10 | 8 | 4 | -5.76 | 62 | 81 | 8 | 5 |
Table 2: Surgical indicators and scores. The positioning time, fluoroscopy time, and puncture accuracy were calculated and are shown. The pre- and post-operative VAS score and Harris score are also shown in this table.
Subscription Required. Please recommend JoVE to your librarian.
Discussion
Although THA has developed rapidly in recent years and become an effective ultimate method for ONFH, hip preservation therapy still plays an important role in treating early-stage ONFH18,19. CD is a basic and effective hip preservation surgery, which can release hip pain and delay the development of femoral head collapse20. The puncture positioning of the focal necrosis is the crucial procedure of CD, as it determines the success or failure of the surgery. However, the traditional puncture positioning method still contains some blind spots that may lead to repeated puncture, increase in exposure to fluoroscopy, and increased operation time10,11. Many scholars have also made efforts to improve this aspect, such as using 3D printing, a combination of hip arthroscopy, and the use of a robot-assisted navigation system12,21,22,23. These methods certainly improve the efficiency and accuracy of puncture positioning, however they also have some deficiencies in other aspects, such as adding operative complexity, causing subsidiary injury, and increasing medical cost.
The system shown here can divide the virtual surgical area in the preoperative registering process. In the virtual surgical area, a high-precision trace of the electro-optical target tracking equipment and the virtual display of the Kirschner wire can be achieved. As required, the second film and superposition can also be conducted when adjusting the angle of the hip. The mean registering time is only 10.1 min. When performing other operations in the same areas, repeated registration is not required. The whole process of registering and positioning is non-invasive, thus ensuring a high level of safety and fitting with a less invasive surgical principle.
AR technique superimposes the imperceptible information into the real-time video frame, which achieves the combination of virtuality and reality24. The AR technique has been combined into many orthopedic surgeries, such as fracture reduction, bone tumor resection, etc.25,26,27. To our knowledge, this is the first study applying AR in CD surgery. The greatest advantage of our system is real-time visualization, which can reduce the difficulty of surgery and shorten the learning curve of the surgeons.
There are also some limitations in this study. Firstly, the sample size of this study is very small and therefore, the result is not convincing enough. Secondly, we only report the early clinical outcomes; further follow-up is also required to evaluate the real benefit for the patients. Certainly, there is still some room for development in this system. With the improvement of performance, it will better serve the clinical practice.
Subscription Required. Please recommend JoVE to your librarian.
Disclosures
The authors declare that they have no competing interests.
Acknowledgments
This work was supported by Beijing Natural Science Foundation(7202183), National Natural Science Foundation of China(81972107), and Beijing Municipal Science and Technology Commission(D171100003217001).
Materials
| Name | Company | Catalog Number | Comments |
| AR-assisted Orthopedic Surgery System | Self development | None | An operating software that implements AR for orthopedic surgery |
| Depth camera | Stereolabs | ZED depth camera(ZED mini) | shoot video and sent back to the workstation. |
| Image processing software | Adobe Systems Incorporated | Adobe Photoshop CS6 | Image processing software |
| Infrared positioning device | Northern Digital Inc. | NDI Polaris Spectra optical tracking device | Tracking markers in the surgical area. |
| Puncture device | Stryker | Stryker System 7 Cordless driver and Sabo | Insert kirschner wire into the necrotic area. |
References
- Cohen-Rosenblum, A., Cui, Q.
- Migliorini, F., et al. Prognostic factors in the management of osteonecrosis of the femoral head: A systematic review. The Surgeon: journal of the Royal Colleges of surgeons of Edinburgh and Ireland. (21), 00199 (2022).
- Mont, M. A., Jones, L. C., Hungerford, D. S. Nontraumatic osteonecrosis of the femoral head: ten years later. The Journal of Bone and Joint Surgery. American Volume. 88 (5), 1117-1132 (2006).
- Wang, L., Tian, X., Li, K., Liu, C. Combination use of core decompression for osteonecrosis of the femoral head: A systematic review and meta-analysis using Forest and Funnel Plots. Computational and Mathematical Methods in Medicine. , 1284149 (2021).
- Hua, K. C., et al. The efficacy and safety of core decompression for the treatment of femoral head necrosis: a systematic review and meta-analysis. Journal of Orthopaedic Surgery and Research. 14 (1), 306 (2019).
- Ganz, R., Krushell, R. J., Jakob, R. P., Küffer, J.
- Yoshikawa, K., et al. Training with hybrid assistive limb for walking function after total knee arthroplasty. Journal of Orthopaedic Surgery and Research. 13 (1), 163 (2018).
- Wu, C. T., Yen, S. H., Lin, P. C., Wang, J. W. Long-term outcomes of Phemister bone grafting for patients with non-traumatic osteonecrosis of the femoral head. International Orthopaedics. 43 (3), 579-587 (2019).
- Mont, M. A., Marulanda, G. A., Seyler, T. M., Plate, J. F., Delanois, R. E. Core decompression and nonvascularized bone grafting for the treatment of early stage osteonecrosis of the femoral head. Instructional Course Lectures. 56, 213-220 (2007).
- Wang, W., et al. Patient-specific core decompression surgery for early-stage ischemic necrosis of the femoral head. PLoS One. 12 (5), 0175366 (2017).
- Hoffmann, M. F., Khoriaty, J. D., Sietsema, D. L., Jones, C. B. Outcome of intramedullary nailing treatment for intertrochanteric femoral fractures. Journal of Orthopaedic Surgery and Research. 14 (1), 360 (2019).
- Dennler, C., et al. Augmented reality-based navigation increases precision of pedicle screw insertion. Journal of Orthopaedic Surgery and Research. 15 (1), 174 (2020).
- Yonezawa, H., et al. Low-grade myofibroblastic sarcoma of the levator scapulae muscle: a case report and literature review. BMC Musculoskeletal Disorders. 21 (1), 836 (2020).
- Tsukada, S., et al. Augmented reality- vs accelerometer-based portable navigation system to improve the accuracy of acetabular cup placement during total hip arthroplasty in the lateral decubitus position. The Journal of Arthroplasty. 37 (3), 488-494 (2021).
- Raymond, J., et al. Pharmacogenetics of direct oral anticoagulants: a systematic review. Journal of Personalized Medicine. 11 (1), 37 (2021).
- Bhatt, F. R., et al. Augmented reality-assisted spine surgery: an early experience demonstrating safety and accuracy with 218 screws. Global Spine Journal. , 21925682211069321 Advance online (2022).
- Weiss, H. R., Nan, X., Potts, M. A. Is there an indication for surgery in patients with spinal deformities? - A critical appraisal. The South African Journal of Physiotherapy. 77 (2), 1569 (2021).
- Boontanapibul, K., Amanatullah, D. F., Huddleston, J. I., Maloney, W. J., Goodman, S. B. Outcomes of cemented total knee arthroplasty for secondary osteonecrosis of the knee. The Journal of Arthroplasty. 36 (2), 550-559 (2021).
- Bakircioglu, S., Atilla, B. Hip preserving procedures for osteonecrosis of the femoral head after collapse. J Clin Orthop Trauma. 23, 101636 (2021).
- Ma, H. Y., et al. Core decompression with local administration of zoledronate and enriched bone marrow mononuclear cells for treatment of non-traumatic osteonecrosis of femoral head. Orthopaedic Surgery. 13 (6), 1843-1852 (2021).
- Hu, L., et al. Comparison of intramedullary nailing and plate fixation in distal tibial fractures with metaphyseal damage: a meta-analysis of randomized controlled trials. Journal of Orthopaedic Surgery and Research. 14 (1), 30 (2019).
- Pierannunzii, L. Endoscopic and arthroscopic assistance in femoral head core decompression. Arthroscopy Techniques. 1 (2), 225-230 (2012).
- Salas, A. P., et al. Hip arthroscopy and core decompression for avascular necrosis of the femoral head using a specific aiming guide: a step-by-step surgical technique. Arthroscopy Techniques. 10 (12), 2775-2782 (2021).
- Beer, A. J., Dijkgraaf, I. Editorial European journal of nuclear medicine and molecular imaging. European Journal of Nuclear Medicine and Molecular Imaging. 44 (2), 284-285 (2017).
- Negrillo-Cárdenas, J., Jiménez-Pérez, J. R., Feito, F. R. The role of virtual and augmented reality in orthopedic trauma surgery: From diagnosis to rehabilitation. Computer Methods and Programs in Biomedicine. 191, 105407 (2020).
- Brookes, M. J., et al.
- Cho, H. S., et al. Can augmented reality be helpful in pelvic bone cancer surgery? an in vitro study. Clinical Orthopaedics and Related Research. 476 (9), 1719-1725 (2018).
