External validation of a deep learning model for predicting bone mineral density on chest radiographs.

We developed a new model for predicting bone mineral density on chest radiographs and externally validated it using images captured at facilities other than the development environment. The model performed well and showed potential for clinical use. PURPOSE: In this study, we performed external validation (EV) of a developed deep learning model for predicting bone mineral density (BMD) of femoral neck on chest radiographs to verify the usefulness of this model in clinical practice. METHODS: This study included patients who visited any of the collaborating facilities from 2010 to 2020 and underwent chest radiography and dual-energy X-ray absorptiometry (DXA) at the femoral neck in the year before and after their visit. A total of 50,114 chest radiographs were obtained, and BMD was measured using DXA. We developed the model with 47,150 images from 17 facilities and performed EV with 2914 images from three other facilities (EV dataset). We trained the deep learning model via ensemble learning based on chest radiographs, age, and sex to predict BMD using regression. The outcomes were the correlation of the predicted BMD and measured BMD with diagnoses of osteoporosis and osteopenia using the T-score estimated from the predicted BMD. RESULTS: The mean BMD was 0.64±0.14 g/cm2 in the EV dataset. The BMD predicted by the model averaged 0.61±0.08 g/cm2, with a correlation coefficient of 0.68 (p<0.01) when compared with the BMD measured using DXA. The accuracy, sensitivity, and specificity of the model were 79.0%, 96.6%, and 34.1% for T-score < -1 and 79.7%, 77.1%, and 80.4% for T-score ≤ -2.5, respectively. CONCLUSION: Our model, which was externally validated using data obtained at facilities other than the development environment, predicted BMD of femoral neck on chest radiographs. The model performed well and showed potential for clinical use.

Deep Learning for Bone Mineral Density and T-Score Prediction from Chest X-rays: A Multicenter Study.

Although the number of patients with osteoporosis is increasing worldwide, diagnosis and treatment are presently inadequate. In this study, we developed a deep learning model to predict bone mineral density (BMD) and T-score from chest X-rays, which are one of the most common, easily accessible, and low-cost medical imaging examination methods. The dataset used in this study contained patients who underwent dual-energy X-ray absorptiometry (DXA) and chest radiography at six hospitals between 2010 and 2021. We trained the deep learning model through ensemble learning of chest X-rays, age, and sex to predict BMD using regression and T-score for multiclass classification. We assessed the following two metrics to evaluate the performance of the deep learning model: (1) correlation between the predicted and true BMDs and (2) consistency in the T-score between the predicted class and true class. The correlation coefficients for BMD prediction were hip = 0.75 and lumbar spine = 0.63. The areas under the curves for the T-score predictions of normal, osteopenia, and osteoporosis diagnoses were 0.89, 0.70, and 0.84, respectively. These results suggest that the proposed deep learning model may be suitable for screening patients with osteoporosis by predicting BMD and T-score from chest X-rays.

Detection of Sacral Fractures on Radiographs Using Artificial Intelligence.

Sacral fractures are often difficult to diagnose on radiographs. Computed tomography (CT) and magnetic resonance imaging (MRI) can improve the detection rate but cannot always be performed. The accuracy of artificial intelligence (AI) in detecting orthopaedic fractures is now comparable with that of orthopaedic specialists. However, the ability of AI to detect sacral fractures has not been investigated, to our knowledge. We hypothesized that the ability to detect sacral fractures on radiographs could be improved by using AI, and aimed to develop an AI model to detect sacral fractures accurately on radiographs with better accuracy than that of orthopaedic surgeons. Methods: Subjects were patients with suspected pelvic fractures for whom radiographs and CT scans had been obtained. The radiographs were labeled according to sacral fracture status based on CT results. The data set was divided into a training set (2,038 images) and a test set (200 images). Eight convolutional neural network (CNN) models were trained using the training set. Post-trained models were used to evaluate their discrimination ability. The detection ability of 4 experienced orthopaedic surgeons was also measured using the same test set. The results of fracture assessment by the orthopaedic surgeons were compared with those of the 3 CNNs with the greatest area under the receiver operating characteristic curve. Results: Among the 8 trained models, the highest areas under the curve were for InceptionV3 (0.989), Xception (0.987), and Inception ResNetV2 (0.984). The detection rate was significantly higher for these 3 CNNs than for the orthopaedic surgeons. Conclusions: By enhancing the processing of probabilistic tasks and the communication of their results, AI may be better able to detect sacral fractures than orthopaedic surgeons. Level of Evidence: Diagnostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Distribution of hounsfield unit values in the pelvic bones: a comparison between young men and women with traumatic fractures and older men and women with fragility fractures: a retrospective cohort study.

BACKGROUND: The fixation strength of bone screws depends on bone mineral density (BMD), so it is important to evaluate bone strength at fracture sites. Few studies have investigated BMD in the pelvis. The aims of this study were to measure the regional Hounsfield unit (HU) values in the cancellous bone of the acetabulum and pelvic ring and to compare these values between young and older patients. METHODS: This study enrolled young patients with high-energy trauma (aged 20-44 years; young group) and older patients with low-energy trauma (aged 65-89 years; older group). Patients without pelvic computed tomography (CT) scans, those with pelvic bone implants, and those who died were excluded. The HU values on the contralateral (non-fractured) side of the pelvis were measured on CT scans. The CT data were divided into 7 areas: the pubic bone, the anterior and posterior walls and roof of the acetabulum, the ischial tuberosity, the body of the ilium, and the third lumbar vertebra. The HU values in each area were compared between the young and older groups. RESULTS: Sixty-one young patients and 154 older patients were included in the study. The highest HU value was in the roof of the acetabulum regardless of age and sex. HU values were significantly higher in the ischial tuberosity and body of the ilium and lower in the pubic bone and anterior wall. The HU values in all pelvic areas were significantly lower in the older group than in the young group, especially in the anterior area. CONCLUSIONS: HU values in the 6 pelvic areas were not uniform and were strongly related to load distribution. The HU distribution and age-related differences could explain the characteristic causes and patterns of acetabular fractures in the older and may help in surgical treatment.

Factors affecting spine-femur discordance in the percentage of young adult mean on dual-energy X-ray absorptiometry in the elderly population: a retrospective study.

BACKGROUND: Several retrospective studies have reported spine-femur discordance in bone mineral density (BMD) values. However, the average age of individuals in these studies was the mid-50s, which is younger than the typical age of individuals requiring treatment for primary osteoporosis. Therefore, we aimed to investigate factors associated with discordance in the percentage of young adult mean (YAM) between the lumbar spine and femoral neck in the elderly population. METHODS: We evaluated 4549 dual-energy X-ray absorptiometry (DXA) measurements obtained from 2161 patients (269 men and 1892 women) between January 2014 and December 2017 at our hospital. For individuals with more than one eligible set of measurements, the first record was used. We investigated each patient's age, sex, body mass index, current smoking status, alcohol consumption, use of steroids, presence of diabetes mellitus, and presence of rheumatoid arthritis. RESULTS: The mean age of the patients was 76.4 ± 8.9 years. Older age (p <  0.001), male sex (p <  0.001), and diabetes mellitus (p = 0.007) were significantly associated with spine-femur discordance in the percentage of YAM. CONCLUSION: The frequency and magnitude of spine-femur discordance in the percentage of YAM from DXA scans increased with age. Notably, more than 77.4% of patients in their 90s had spine-femur discordance > 10% of YAM. Furthermore, the frequency of spine-femur discordance was higher in men and in patients with diabetes mellitus, suggesting that the percentage of YAM at the lumbar spine may not be reliable for diagnosis of osteoporosis in patients with these factors.

Acute Compartment Syndrome of the Upper Extremity in Acquired Hemophilia A: A Case Report and Literature Review.

CASE: Acute compartment syndrome (ACS) with acquired hemophilia A (AHA) is rare and has no established treatment strategy. A 64-year-old woman presented with a giant hematoma in the rectus abdominis. Laboratory findings included decreased hemoglobin and increased activated partial thromboplastin time. Arterial embolization was performed for hemostasis. After catheter removal, she developed severe arm pain and numbness with blistering. Fasciotomy was performed to decrease intracompartmental pressure. Laboratory investigations revealed decreased factor VIII (FVIII) activity and increased FVIII inhibitor. AHA was diagnosed and treated with immunosuppressive and FVIII inhibitor-bypassing agents. CONCLUSIONS: Fasciotomy should be performed promptly if ACS with AHA is suspected.

Local application of alendronate controls bone formation and ß-tricalcium phosphate resorption induced by recombinant human bone morphogenetic protein-2.

This study examined the ability of local alendronate (ALN) administration to control ß-tricalcium phosphate (ß-TCP) resorption as well as the induction of bone formation by recombinant human bone morphogenetic protein-2 (rhBMP-2). A 15-mm critical-sized bone defect was created in the diaphysis of rabbit ulnae. Nine female rabbits (4 to 5 months-old) were divided into 3 groups. Group 1 (n = 6 ulnae) animals received implants consisting of ß-TCP granules and 25 µg of rhBMP-2 in 6.5% collagen gel. Group 2 (6 ulnae) and Group 3 (6 ulnae) animals received the same implants, but with 10-6 M and 10-3 M ALN-treated TCP granules, respectively. Two weeks postsurgery, tartrate-resistant acid phosphatase-positive cell counts, new bone formation, and residual ß-TCP were evaluated. This study showed that a high dose of ALN strongly reduced osteoclastic resorption of ß-TCP induced by rhBMP-2, resulting in decreased bone formation. In contrast, a low dose of ALN slightly reduced the bone resorptive effect but increased bone formation. These results suggest that osteoclast-mediated resorption plays an important role in bone formation and a coupling-like phenomenon could occur in the ß-TCP-implanted area, and that administration of a low dose of ALN may solve clinical bone resorptive problems induced by rhBMP-2.

Effects of micro-porosity and local BMP-2 administration on bioresorption of ß-TCP and new bone formation.

BACKGROUND: It has been reported that the microporous structure of calcium phosphate (CaP) ceramics is important to osteoconduction. Bone morphogenetic protein-2 (BMP-2) has been shown to be a promising alternative to bone grafting and a therapeutic agent promoting bone regeneration when delivered locally. The aim of this study was to evaluate the effects of micro-porosity within beta-tricalcium phosphate (ß-TCP) cylinders and local BMP-2 administration on ß-TCP resorption and new bone formation. METHODS: Bilateral cylindrical bone defects were created in rabbit distal femora, and the defects were filled with ß-TCP. Rabbits were divided into 3 groups; defects were filled with a ß-TCP cylinder with a total of approximately 60% porosity (Group A: 13.4% micro- and 46.9% macropore, Group B: 38.5% micro- and 20.3% macropore, Group C: the same micro- and macro-porosity as in group B supplemented with BMP-2). Rabbits were sacrificed 4, 8, 12, and 24 weeks postoperatively. RESULTS: The number of TRAP-positive cells and new bone formation in group B were significantly greater than those in group A at every period. The amount of residual ß-TCP in group C was less than that in group B at all time periods, resulting in significantly more new bone formation in group C at 8 and 12 weeks. The number of TRAP-positive cells in group C was maximum at 4 weeks. CONCLUSIONS: These results suggest that the amount of submicron microporous structure and local BMP-2 administration accelerated both osteoclastic resorption of ß-TCP and new bone formation, probably through a coupling-like phenomenon between resorption and new bone formation.