Proximal Femur Dual-energy X-ray Absorptiometry Research Articles

Prediction of osteoporosis from simple hip radiography using deep learning algorithm

Despite being the gold standard for diagnosis of osteoporosis, dual-energy X-ray absorptiometry (DXA) could not be widely used as a screening tool for osteoporosis. This study aimed to predict osteoporosis via simple hip radiography using deep learning algorithm. A total of 1001 datasets of proximal femur DXA with matched same-side cropped simple hip bone radiographic images of female patients aged ≥ 55 years were collected. Of these, 504 patients had osteoporosis (T-score ≤ − 2.5), and 497 patients did not have osteoporosis. The 1001 images were randomly divided into three sets: 800 images for the training, 100 images for the validation, and 101 images for the test. Based on VGG16 equipped with nonlocal neural network, we developed a deep neural network (DNN) model. We calculated the confusion matrix and evaluated the accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV). We drew the receiver operating characteristic (ROC) curve. A gradient-based class activation map (Grad-CAM) overlapping the original image was also used to visualize the model performance. Additionally, we performed external validation using 117 datasets. Our final DNN model showed an overall accuracy of 81.2%, sensitivity of 91.1%, and specificity of 68.9%. The PPV was 78.5%, and the NPV was 86.1%. The area under the ROC curve value was 0.867, indicating a reasonable performance for screening osteoporosis by simple hip radiography. The external validation set confirmed a model performance with an overall accuracy of 71.8% and an AUC value of 0.700. All Grad-CAM results from both internal and external validation sets appropriately matched the proximal femur cortex and trabecular patterns of the radiographs. The DNN model could be considered as one of the useful screening tools for easy prediction of osteoporosis in the real-world clinical setting.

A Hip Analysis Protocol for Pediatric Bone Densitometry: The Iowa Bone Development Study

Pediatric proximal femur dual-energy X-ray absorptiometry (DXA) scans present analytic challenges because of the lack of standard points of reference in the growing skeleton. The Iowa Bone Development Study (IBDS) developed a modified pediatric-specific proximal femur analysis protocol using Hologic software. Serial DXA measurements were obtained for 214 children at approximate ages 5, 8, 11, and 13 yr. Standard analysis procedures as described by the manufacturer (Hologic default) were compared with the IBDS protocol. The IBDS protocol yielded lower but more stable results for bone area, bone mineral content (BMC), and bone mineral density for total hip, femoral neck, trochanter, and intertrochanter as a result of more precisely controlling the regions of interest. Linear regression models with body size, age, and gender as predictors were developed to examine variation in measurements. Coefficients of determination ( R 2) with the IBDS protocol were greater for each time point, demonstrating that the modified protocol was better aligned with body size. Similarly, Spearman correlation coefficients between total hip and hip subregions were consistently higher for BMC and bone area with the IBDS protocol with differences more notable among younger children. The IBDS protocol provides a reproducible method for evaluating pediatric proximal femur DXA scans during growth.

A quantitative comparison of a bone remodeling model with dual-energy X-ray absorptiometry and analysis of the inter-individual biological variability of femoral neck T-score

The development of consistent procedures with the inclusion of patient-specific data is essential in the computational modeling of biological processes, in order to achieve clinical relevant data. In this work, these issues are addressed with the development of a methodology that combines the gold standard technique for bone mineral density measurement and osteoporosis diagnosis, Dual energy X-ray absorptiometry (DXA), with a computational model for bone remodeling simulation. The DXA results were divided in three samples constituted from proximal femur DXA exams of patients in different stages of bone mineral density (normal, osteopenia and osteoporosis). These results were quantitatively compared with computational model results. A correlation study was performed between femoral neck T-score and a parameter from the model to ascertain the hypothesis of adjusting the model accordingly to biological variables. The results evidenced the predictive ability of the computational model in the estimation of femoral neck bone mineral content (BMC), with a maximum relative error of 3.92%. On the other hand, a strong correlation ( R=−0.862) was found between the variables in study and a mathematical relationship was obtained to estimate the range of values for a model parameter that leads to biological relevant results. The methodology developed and the results obtained represent a solid and reliable basis to further studies on bone quality, ensuring the validity of the computational model in the simulation of bone remodeling process.

A Novel DXA-Based Hip Failure Index Captures Hip Fragility Independent of BMD

Capability of a novel dual-energy X-ray absorptiometry (DXA)-based hip failure index (HiFI) to discriminate between hip fracture cases and controls was evaluated. Given the constraints of planar DXA, the femoral neck was assumed a foam-filled (∼trabecular bone), thin-walled (∼cortical bone) sandwich structure, while HiFI estimated the critical force sufficient to buckle the wall of such a structure. Proximal femur DXA data from 1379 women aged 65 yr and older, 268 with prior hip fracture were used. Comparison between standard areal bone mineral density (BMD), femur strength index (FSI), and HiFI was based on areas under receiver operatoring characteristic curves (AUC). The mean femoral neck BMD (SD) was 0.689 (0.109) g/cm 2 among the cases and 0.768 (0.119) g/cm 2 among the controls; the mean FSI 1.33 (0.36) and 1.54 (0.41), and the mean HiFI −0.28 (0.14) and −0.18 (0.15), respectively; all intergroup differences were highly significant ( p < 0.001). The intergroup difference for HiFI remained significant ( p < 0.002) after adjusting for age and BMD or FSI. The AUCs were 0.696 (95% confidence interval [CI]: 0.661–0.730) for BMD, 0.665 (0.630–0.700) for FSI, and 0.701 (0.666–0.736) for HiFI. In conclusion, HiFI may capture structural traits that account for femoral neck fragility independently of BMD or FSI. Obviously, the use of actual geometric and structural information from three-dimensional imaging of the femoral neck would help diminish the crude assumptions of the present DXA approach and reveal the true potential of the HiFI approach to gauge hip fragility and identify at-risk individuals for hip fractures.

Discordance in DXA male reference ranges.

Previous studies have reported discordance in female lumbar spine and proximal femur dual-energy X-ray absorptiometry (DXA) reference ranges. Although the NHANES III reference range is recommended for the proximal femur in males and females, there are no published data in men on the concordance or otherwise of the different manufacturer-specific lumbar spine bone mineral density (BMD) reference ranges. Potentially, the use of different reference populations by different manufacturers could result in inconsistencies in the diagnosis of osteopenia or osteoporosis. We compared lumbar spine BMD, as well as T-scores and Z-scores, in 45 men scanned using Lunar DPXL and Norland Excel densitometers. The BMD measured by the two instruments was highly correlated (lumbar spine: r = 0.99, p < 0.0001). However, the two instruments assigned significantly different BMD T-scores. These differences relate primarily to the different standard deviations employed in the calculations. There were also significant differences when BMD was expressed with respect to age-matched values (Z-scores). This study shows that in men, as previously demonstrated in women, two commonly used DXA instruments provide comparable lumbar spine standardized BMD, but there are significant differences in derived T-scores because of differences in the manufacturer-specific reference ranges. Standardization of lumbar spine reference ranges in men should be a high priority.

Application of magnetic resonance imaging to evaluation of femoral neck structure in growing girls.

Conventional density measures by dual-energy X-ray absorptiometry (DXA) are confounded by increases in bone size and do not assess bone geometry. We assessed precision of magnetic resonance imaging (MRI) and used MRI, DXA, and hip structure analysis (HSA) to assess 7-mo changes in bone structure at the femoral neck in 18 prepubertal girls. At baseline, girls were 10.4 (0.5) yr, 144.0 (8.2) cm, and 35.2 (7.0) kg, on average. Total bone and cortical cross-sectional area (ToA and CoA) were calculated from high-resolution T1-weighted MRI oblique axial images of the femoral neck. We used proximal femur DXA scans (Hologic QDR-4500) and the HSA program to estimate bone cross-sectional area (CSA), and calculate section modulus. MRI precision was determined by scanning 10 volunteers (13-46 yr old) three times with and without repositioning. Precision (CVrms) was 2% for ToA and 7% for CoA. Significant correlations were observed between FN area and MRI-derived ToA (r = 0.57, p = 0.013) and CoA (r = 0.47, p = 0.050). There were significant positive changes over 7 mo by both methods. In conclusion, MRI provides useful information on femoral neck bone area in children. The reproducibility of cortical dimensions at the femoral neck needs improvement through technical modifications and appropriate analysis software.

A prospective study of discordance in diagnosis of osteoporosis using spine and proximal femur bone densitometry.

The T-score definition of osteoporosis, originally intended for epidemiologic research in this condition, is frequently used in making treatment decisions for individual patients. Discordance in classification depending on the site and type of measurement has been reported in retrospective and cross-sectional studies, but the impact of such discordance on clinical practice is unknown. This is potentially important in view of a recent proposal to confine osteoporosis diagnosis to densitometry at the hip. Having excluded those with degenerative changes in the lumbar spine, we compared the T-score classification of a prospective cohort of patients referred for their first dual-energy X-ray absorptiometry (DXA) scan, analyzing data for men and women in 10-year age groups. Total hip and neck of femur DXA identified significantly fewer osteoporotic patients than spine DXA, and this reduced sensitivity could not be improved by adjusting the T-score threshold without an unacceptable increase in non-osteoporotic cases. The majority of patients undetected by proximal femur DXA were at significantly increased risk of vertebral fracture. DXA at the lumbar spine had only moderate sensitivity and specificity for osteoporosis redefined by total hip densitometry, indicating differential rates of bone loss at the proximal femur and spine. We conclude that, as the most usual indication for bone densitometry is to aid the determination of an individual's fracture risk, both proximal femur and lumbar spine should continue to be assessed.