Knee Implant Identification by Fine-Tuning Deep Learning Models.

BACKGROUND: Identification of implant model from primary knee arthroplasty in pre-op planning of revision surgery is a challenging task with added delay. The direct impact of this inability to identify the implants in time leads to the increase in complexity in surgery. Deep learning in the medical field for diagnosis has shown promising results in getting better with every iteration. This study aims to find an optimal solution for the problem of identification of make and model of knee arthroplasty prosthesis using automated deep learning models. METHODS: Deep learning algorithms were used to classify knee arthroplasty implant models. The training, validation and test comprised of 1078 radiographs with a total of 6 knee arthroplasty implant models with anterior-posterior (AP) and lateral views. The performance of the model was calculated using accuracy, sensitivity, and area under the receiver-operating characteristic curve (AUC), which were compared against multiple models trained for comparative in-depth analysis with saliency maps for visualization. RESULTS: After training for a total of 30 epochs on all 6 models, the model performing the best obtained an accuracy of 96.38%, the sensitivity of 97.2% and AUC of 0.985 on an external testing dataset consisting of 162 radiographs. The best performing model correctly and uniquely identified the implants which could be visualized using saliency maps. CONCLUSION: Deep learning models can be used to differentiate between 6 knee arthroplasty implant models. Saliency maps give us a better understanding of which regions the model is focusing on while predicting the results.

Automated Identification of Orthopedic Implants on Radiographs Using Deep Learning.

Accurate identification of metallic orthopedic implant design is important for preoperative planning of revision arthroplasty. Surgical records of implant models are frequently unavailable. The aim of this study was to develop and evaluate a convolutional neural network for identifying orthopedic implant models using radiographs. In this retrospective study, 427 knee and 922 hip unilateral anteroposterior radiographs, including 12 implant models from 650 patients, were collated from an orthopedic center between March 2015 and November 2019 to develop classification networks. A total of 198 images paired with autogenerated image masks were used to develop a U-Net segmentation network to automatically zero-mask around the implants on the radiographs. Classification networks processing original radiographs, and two-channel conjoined original and zero-masked radiographs, were ensembled to provide a consensus prediction. Accuracies of five senior orthopedic specialists assisted by a reference radiographic gallery were compared with network accuracy using McNemar exact test. When evaluated on a balanced unseen dataset of 180 radiographs, the final network achieved a 98.9% accuracy (178 of 180) and 100% top-three accuracy (180 of 180). The network performed superiorly to all five specialists (76.1% [137 of 180] median accuracy and 85.6% [154 of 180] best accuracy; both P < .001), with robustness to scan quality variation and difficult to distinguish implants. A neural network model was developed that outperformed senior orthopedic specialists at identifying implant models on radiographs; real-world application can now be readily realized through training on a broader range of implants and joints, supported by all code and radiographs being made freely available. Supplemental material is available for this article. Keywords: Neural Networks, Skeletal-Appendicular, Knee, Hip, Computer Applications-General (Informatics), Prostheses, Technology Assess-ment, Observer Performance © RSNA, 2021.

Orthopaedic Surgeon Brain Radiation During Fluoroscopy: A Cadaver Model.

BACKGROUND: The aims of this study were to quantify exposure of the surgeon's brain to radiation during short cephalomedullary (SC) nailing, to extrapolate lifetime dose, and to determine the effects of personal protective equipment (PPE) on brain dose. METHODS: Two cadaveric specimens were used: (1) a whole cadaveric body representing the patient, with a left nail inserted to act as the scatter medium, and (2) an isolated head-and-neck cadaveric specimen representing a surgeon, with radiation dosimeters placed in specific locations in the brain. The "patient" cadaver's left hip was exposed in posteroanterior and lateral radiographic planes. Measurements were performed without shielding of the head-and-neck specimen and then repeated sequentially with different PPE configurations. An average surgeon career was estimated to be 40 years (ages 25 to 65 years) with the caseload obtained from the department's billing data. RESULTS: The mean radiation dose to the surgeon brain without PPE was 3.35 µGy (95% confidence interval [CI]: 2.4 to 4.3) per nail procedure. This was significantly reduced with use of a thyroid collar (2.94 µGy [95% CI: 1.91 to 3.91], p = 0.04). Compared with use of the thyroid collar in isolation, there was no significant additional reduction in radiation when the collar was used with leaded glasses (2.96 µGy [95% CI: 2.15 to 3.76], p = 0.97), with a lead cap (3.22 µGy [95% CI: 2.31 to 4.13], p = 0.55), or with both (2.31 µGy [95% CI: 1.61 to 3.01], p = 0.15). The extrapolated lifetime dose over 40 working years for SC nailing without PPE was 2,146 µGy (95% CI: 1,539 to 2,753), with an effective dose of 21.5 µSv. CONCLUSIONS: The extrapolated cumulative lifetime radiation to a surgeon's brain from SC nailing based on our institution's workload and technology is low and comparable with radiation during a one-way flight from London to New York. Of note, we studied only one of many fluoroscopy-aided procedures and likely underestimated total lifetime exposure if exposures from other procedures are included. This study also demonstrates that thyroid collars significantly reduce brain dose for this procedure whereas other head/neck PPE such as lead caps appear to have minimal additional effect. This study provides a methodology for future studies to quantify brain dose for other common orthopaedic procedures. CLINICAL RELEVANCE: This study, based on our institutional data, demonstrates that although the lifetime brain dose from SC nailing is low, thyroid collars significantly reduce this dose further. As such, in accordance with the "as low as reasonably achievable" radiation exposure principle, radiation safety programs and individual surgeons should consider use of thyroid collars in this setting.

Computerised tomography vs magnetic resonance imaging for modeling of patient-specific instrumentation in total knee arthroplasty.

AIM: To summarise and compare currently available evidence regarding accuracy of pre-operative imaging, which is one of the key choices for surgeons contemplating patient-specific instrumentation (PSI) surgery. METHODS: The MEDLINE and EMBASE medical literature databases were searched, from January 1990 to December 2013, to identify relevant studies. The data from several clinical studies was assimilated to allow appreciation and comparison of the accuracy of each modality. The overall accuracy of each modality was calculated as proportion of outliers > 3% in the coronal plane of both computerised tomography (CT) or magnetic resonance imaging (MRI). RESULTS: Seven clinical studies matched our inclusion criteria for comparison and were included in our study for statistical analysis. Three of these reported series using MRI and four with CT. Overall percentage of outliers > 3% in patients with CT-based PSI systems was 12.5% vs 16.9% for MRI-based systems. These results were not statistically significant. CONCLUSION: Although many studies have been undertaken to determine the ideal pre-operative imaging modality, conclusions remain speculative in the absence of long term data. Ultimately, information regarding accuracy of CT and MRI will be the main determining factor. Increased accuracy of pre-operative imaging could result in longer-term savings, and reduced accumulated dose of radiation by eliminating the need for post-operative imaging and revision surgery.

Comparison of intra-operative regimes of tranexamic acid administration in primary total hip replacement.

The administration of tranexamic acid to decrease blood loss in primary total hip replacement is established. In this observational study three different regimes of tranexamic acid were used to investigate the effects of combined intravenous and topical administration of tranexamic acid to a single intravenous bolus given at induction or closure. Group 1 (n = 50) received 1 g tranexamic acid intravenously at induction and 500 mg tranexamic acid topically during closure. Group 2 (n = 50) received 1 g tranexamic acid intravenously at induction. Group 3 (n = 50) received 1 g tranexamic acid at closure. The mean haemoglobin loss was 2.83 g/dL (95% Confidence interval [CI] 2.51 to 3.15 g/dL) in Group 1, 2.92 g/dL (95% CI 2.65 to 3.19 g/dL) in Group 2 and 3.36 g/dL (95% CI 2.94 to 3.77 g/dL) in Group 3. No significant difference in mean haemoglobin loss was found (p = 0.123). In this observational, non-randomised study we found no additional advantage to giving topical tranexamic acid at closure in addition to intravenous tranexamic acid given at induction.

Long-term survival of cemented distal femoral endoprostheses with a hydroxyapatite-coated collar: a histological study and a radiographic follow-up.

BACKGROUND: The objective of this study was to examine the degree of osteointegration into a hydroxyapatite-coated collar and relate this finding to aseptic loosening in patients with a distal femoral replacement used to treat primary bone cancer. Our hypothesis was that the implant collar would increase osteointegration and reduce the rate of aseptic implant loosening. METHODS: Sixty-one patients treated with a primary cemented distal femoral prosthesis between 1992 and 2001 were included in this study. The mean duration of follow-up was 8.5 years (range, two to eighteen years). Extracortical bone growth into the grooved hydroxyapatite-coated collar was quantified radiographically. Histological sections through four hydroxyapatite-coated collars and four implants with no collar, retrieved following amputation due to local recurrence or at autopsy at a mean of 3.5 years (range, 1.4 to 6.1 years) after implantation, were evaluated as well. RESULTS: Five (8%) of the implants were revised because of aseptic loosening, 3% of the implants fractured, and 3% were revised because of infection. Six limbs (10%) required amputation because of local tumor recurrence. On radiographs, osteointegration into the collar was seen to have occurred in 70% of the patients and did not correlate with sex, age, diagnosis, or length of time postoperatively. Histological analysis showed mature lamellar bone within the grooves of the hydroxyapatite-coated collar, and bone was observed in direct contact with the hydroxyapatite coating. Extracortical bone failed to make direct contact with the surface of the implants manufactured without a collar. CONCLUSIONS: The use of cemented distal femoral massive bone tumor prostheses with a hydroxyapatite-coated collar located at the shoulder of the implant was followed by a low (8%) rate of revision due to aseptic loosening. The use of hydroxyapatite grooved collars may lead to osteointegration of the implant shoulder (collar) and may reduce the rate of aseptic loosening.

No difference in the long term final functional outcome after nailing or cast bracing of high energy displaced tibial shaft fractures.

BACKGROUND: Cast bracing (CB) has been a well established method of treating tibial shaft fractures. Majority of the recent literature on treatment of tibial shaft fractures have upheld intramedullary nailing (IMN) as the treatment of choice. Most of these studies are from the west, in public funded health set ups and in hospitals with very low rates of infection. This has lead to bewilderment in the minds of surgeons wishing to opt for conservative treatment in countries with scarcity of health resources. We therefore undertook this study to compare the two modalities in the scenario of the developing world. MATERIAL AND METHODS: Sixty-eight consecutive patients were treated alternately with CB and IMN for high energy, displaced, closed and Gustilo Grade 1 open fractures of the tibial shaft, between 1995 and 2001. RESULTS: An average follow up at 4.3 years revealed no statistical difference in the final functional outcome as per Johner and Wruhs' criteria with modification to Indian lifestyle. IMN group had a) slightly shorter time to fracture union (mean 21.3 weeks versus 23.1 weeks for CB, p > 0.05), (b) lesser time off work (mean 17.6 weeks versus 25.6 weeks for CB, p <0.01), (c) fewer outpatient visits (mean 6.2 versus 9.7 for CB, p < 0.05), (d) less limb length discrepancy (mean 4.3 mm versus 6.6 mm for CB, p < 0.05). The difference in residual antero-posterior angulation (mean 3.2 degrees for IMN versus 4.9 degrees for CB, p = 0.14) and varus-valgus angulation (mean 3.7 degrees for IMN versus 5.1 degrees for CB, p = 0.7) were not statistically significant. However CB group had no deep infections as compared to two in the IMN group. The average cost of hospital treatment of CB group was less than half incurred by the IMN group (average USD 831 versus USD 2071 for nailed group, p < 0.05). CONCLUSION: Treating tibial shaft fracture either with IMN or CB provided equally gratifying results with no statistical difference in final functional outcome. The economic cost to the patient in Indian conditions is significantly less with CB and therefore stands as an equally reliable treatment option, especially in countries with fewer resources.

Familial crossed polysyndactyly in four generations of an Indian family.

BACKGROUND: Polydactyly is the most common malformation of the limbs. "Crossed" polydactyly of hands and feet, i.e., preaxial in one and postaxial in the other, is extremely rare. It has not been included in the standard classification of hand and feet anomalies. METHODS: We report an Indian family with 7 affected members across 4 generations who had "crossed polysyndactyly". All but one affected member had involvement of all four limbs. There were no other congenital anomalies in any of the family members. RESULTS: Familial crossed polysyndactyly appeared to follow an autosomal dominant transmission. This is probably the first case of familial crossed polysyndactyly without any associated anomalies. CONCLUSION: Familial crossed polysyndactyly is a rare malformation and all family members should be screened for other congenital malformations.