Can we improve the prediction of hip fracture by assessing bone structure using shape and appearance modelling? (original) (raw)
Related papers
Fracture Risk Predictions Based on Statistical Shape and Density Modeling of the Proximal Femur
Journal of Bone and Mineral Research, 2014
Increased risk of skeletal fractures due to bone mass loss is a major public health problem resulting in significant morbidity and mortality, particularly in the case of hip fractures. Current clinical methods based on two-dimensional measures of bone mineral density (areal BMD or aBMD) are often unable to identify individuals at risk of fracture. We investigated predictions of fracture risk based on statistical shape and density modeling (SSDM) methods using a case-cohort sample of individuals from the Osteoporotic Fractures in Men (MrOS) study. Baseline quantitative computed tomography (QCT) data of the right femur were obtained for 513 individuals, including 45 who fractured a hip during follow-up (mean 6.9 year observation, validated by physician review). QCT data were processed for 450 individuals (including 40 fracture cases) to develop individual models describing three-dimensional bone geometry and density distribution. Comparison of mean fracture and non-case models indicated complex structural differences that appear to be responsible for resistance to hip fracture. Logistic regressions were used to model the relation of baseline hip BMD and SSDM weighting factors to the occurrence of hip fracture. Area under the receiver operating characteristic (ROC) curve (AUC) for a prediction model based on weighting factors and adjusted by age was significantly greater than AUC for a prediction model based on aBMD and age (0.94 versus 0.83, respectively). The SSDM-based prediction model adjusted by age correctly identified 55% of the fracture cases (and 94.7% of the non-cases), whereas the clinical standard aBMD correctly identified 10% of the fracture cases (and 91.3% of the non-cases). SSDM identifies subtle changes in combinations of structural bone traits (eg, geometric and BMD distribution traits) that appear to indicate fracture risk. Investigation of important structural differences in the proximal femur between fracture and no-fracture cases may lead to improved prediction of those at risk for future hip fracture.
Prediction of Hip Fracture Can Be Significantly Improved by a Single Biomedical Indicator
Annals of Biomedical Engineering, 2000
Femoral neck fractures are a relevant clinical and social problem. The aim of this study was to improve the prediction of patients at-risk of femoral neck fracture with respect to the current densitometric-based methods. In particular, finite element models were used to assess the prediction accuracy obtained by combining together data from the bone density distribution, the proximal femur anatomy, and the fallrelated loading conditions. Two-dimensional finite element models were developed based on dual energy x-ray absorptiometry data. A population of 93 elder Caucasian women ͑half of them reporting a femoral neck fracture͒ were retrospectively classified both using the standard clinical protocol and Bayes' linear classifiers. This study showed that the bone mineral density in the femoral neck region dominated the fracture event ͑65% accuracy͒. Adding the subject's height and the neck-shaft angle to the bone density increased the accuracy to 77%. The classification accuracy was further improved to 82% by including the peak principal tensile strain obtained from the finite element analyses. This research demonstrated that adding one single biomechanical indicator to the standard clinical measurements improves the identification of patients at-risk of femoral neck fracture.
Osteoporosis International, 2004
The shape of the proximal femur has been demonstrated to be important in the occurrence of fractures of the femoral neck. Unfortunately, multiple geometric measurements frequently used to describe this shape are highly correlated. A new method, active shape modeling (ASM) has been developed to quantify the morphology of the femur. This describes the shape in terms of orthogonal modes of variation that, consequently, are all independent. To test this method, digitized standard pelvic radiographs were obtained from 26 women who had suffered a hip fracture and compared with images from 24 age-matched controls with no fracture. All subjects also had their bone mineral density (BMD) measured at five sites using dual-energy X-ray absorptiometry. An ASM was developed and principal components analysis used to identify the modes which best described the shape. Discriminant analysis was used to determine which variable, or combination of variables, was best able to discriminate between the groups. ASM alone correctly identified 74% of the individuals and placed them in the appropriate group. Only one of the BMD values (WardÕs triangle) achieved a higher value (82%). A combination of WardÕs triangle BMD and ASM improved the accuracy to 90%. Geometric variables used in this study were weaker, correctly classifying less than 60% of the study group. Logistic regression showed that after adjustment for age, body mass index, and BMD, the ASM data was still independently associated with hip fracture (odds ratio (OR)=1.83, 95% confidence interval 1.08 to 3.11). The odds ratio was calculated relative to a 10% increase in the probability of belonging to the fracture group. Though these initial results were obtained from a limited data set, this study shows that ASM may be a powerful method to help identify individuals at risk of a hip fracture in the future.
Bone Shape, Structure, and Density as Determinants of Osteoporotic Hip Fracture
Investigative Radiology, 2005
Objectives: This article compares and combines methods for examining the external shape and the internal structure of the proximal femur with bone mineral density (BMD) to provide a classifier for hip fracture. Materials and Methods: Fifty standard pelvic radiographs were available from age-matched fracture and control groups of postmenopausal women. Femoral shape was measured using an active shape model, the trabecular structure by means of a Fourier transform. Results: Both the shape and various structure measures were independent of BMD (P ϭ 0.16 and Ͼ0.50, respectively). Calculating the area under the receiver operator characteristic (ROC) curve (A z ), each of shape (A z ϭ 0.81), the best structure measure (A z ϭ 0.79 -0.93), and BMD (A z ϭ 0.79), could partially classify the fracture and control groups. However, the combination achieved almost perfect separation (A z ϭ 0.99).
Prediction of hip and other osteoporotic fractures from hip geometry in a large clinical cohort
Osteoporosis International, 2009
Incident hip fractures and non-hip osteoporotic fractures were studied in 30,953 women during mean 3.7 years of observation. Hip axis length (HAL) and strength index (SI) made a small but statistically significant contribution to hip fracture prediction that was independent of age and hip bone density. Introduction It is uncertain whether bone geometric measures improve fracture prediction independent of conventional areal bone mineral density (BMD).
Investigative Radiology, 2005
Objectives: This article compares and combines methods for examining the external shape and the internal structure of the proximal femur with bone mineral density (BMD) to provide a classifier for hip fracture. Materials and Methods: Fifty standard pelvic radiographs were available from age-matched fracture and control groups of postmenopausal women. Femoral shape was measured using an active shape model, the trabecular structure by means of a Fourier transform. Results: Both the shape and various structure measures were independent of BMD (P ϭ 0.16 and Ͼ0.50, respectively). Calculating the area under the receiver operator characteristic (ROC) curve (A z ), each of shape (A z ϭ 0.81), the best structure measure (A z ϭ 0.79 -0.93), and BMD (A z ϭ 0.79), could partially classify the fracture and control groups. However, the combination achieved almost perfect separation (A z ϭ 0.99).
Osteoporosis International, 2002
Hip geometry and bone mineral density (BMD) have previously been shown to relate independently to hip fracture risk. Our objective was to determine by how much hip geometric data improved the identification of hip fracture. Lunar pencil beam scans of the proximal femur were obtained. Geometric and densitometric values from 800 female controls aged 60 years or more (from population samples which were participants in the European Prospective Osteoporosis Study, EPOS) were compared with data from 68 female hip fracture patients aged over 60 years who were scanned within 4 weeks of a contralateral hip fracture. We used Lunar DPX ‘beta’ versions of hip strength analysis (HSA) and hip axis length (HAL) applied to DPX(L) data. Compressive stress (Cstress), calculated by the HSA software to occur as a result of a typical fall on the greater trochanter, HAL, body mass index (BMI: weight/(height)2) and age were considered alongside femoral neck BMD (FN-BMD, g/cm2) as potential predictors of fracture. Logistic regression was used to generate predictors of fracture initially from FN-BMD. Next age, Cstress (as the most discriminating HSA-derived parameter), HAL and BMI were added to the model as potentially independent predictors. It was not necessary to include both HAL and Cstress in the logistic models, so the entire data set was examined without excluding the subjects missing HAL measurements. Cstress combined with age and BMI provided significantly better prediction of fracture than FN-BMD used alone as is current practice, judged by comparing areas under receiver operating characteristic (ROC) curves (p<0.001, deLong’s test). At a specificity of 80%, sensitivity in identification was improved from 66% to 81%. Identifying women at high risk of hip fracture is thus likely to be substantially enhanced by combining bone density with age, simple anthropometry and data on the structural geometry of the hip. HSA might prove to be a valuable enhancement of DXA densitometry in clinical practice and its use could justify a more pro-active approach to identifying women at high risk of hip fracture in the community.
Journal of Bone and Mineral Research, 2008
The role of bone tissue's geometric distribution in hip fracture risk requires full evaluation in large population-based datasets. We tested whether section modulus, a geometric index of bending strength, predicted hip fracture better than BMD. Among 7474 women from the Study of Osteoporotic Fractures (SOF) with hip DXA scans at baseline, there were 635 incident hip fractures recorded over 13 yr. Hip structural analysis software was used to derive variables from the DXA scans at the narrow neck (NN), intertrochanter (IT), and shaft (S) regions. Associations of derived structural variables with hip fracture were assessed using Cox proportional hazard modeling. Hip fracture prediction was assessed using the C-index concordance statistic. Incident hip fracture cases had larger neck-shaft angles, larger subperiosteal and estimated endosteal diameters, greater distances from lateral cortical margin to center of mass (lateral distance), and higher estimated buckling ratios (p < 0.0001 for each). Areal BMD, cross-sectional area, cross-sectional moment of inertia, section modulus, estimated cortical thickness, and centroid position were all lower in hip fracture cases (p < 0.044). In hip fracture prediction using NN region parameters, estimated cortical thickness, areal BMD, and estimated buckling ratio were equivalent (C-index ס 0.72; 95% CI, 0.70, 0.74), but section modulus performed less well (C-index ס 0.61; 95% CI, 0.58, 0.63; p < 0.0001 for difference). In multivariable models combining hip structural analysis variables and age, effects of bone dimensions (i.e., lateral distance, subperiosteal diameter, and estimated endosteal width) were interchangeable, whereas age and neck-shaft angle were independent predictors. Several parsimonious multivariable models that were prognostically equivalent for the NN region were obtained combining a measure of width, a measure of mass, age, and neck-shaft angle (BMD is a ratio of mass to width in the NN region; C-index ס 0.77; 95% CI, 0.75, 0.79). Trochanteric fractures were best predicted by analysis of the IT region. Because section modulus failed to predict hip fracture risk as well as areal BMD, the thinner cortices and wider bones among those who fractured may imply that simple failure in bending is not the usual event in fracture. Fracture might require initiation (e.g., by localized crushing or buckling of the lateral cortex).
Journal of Bone and Mineral Research, 1997
We used an experimental software measuring the hip axis length (HAL) and bone mineral density (BMD) in specific regions of the lower and upper part of the femoral neck on dual-energy X-ray absorptiometry scans. To determine whether these parameters were significant predictors of the type of hip fracture, we measured 167 healthy women (controls), 24 women with trochanteric, and 42 women with cervical hip fractures within the EPIDOS prospective cohort. EPIDOS is a multicenter prospective study on risk factors for hip fracture performed in 7575 elderly women living at home, aged 75-95 and conducted in five French centers (Amiens, Lyon, Montpellier, Paris, Toulouse). Measurements were performed on data acquired at baseline before the occurrence of fracture. In the cervical fracture group, HAL was significantly longer than in controls (94.2 vs. 92.3, p ؍ 0.03), and the associated odds ratio (OR) adjusted for age, weight, and total femoral neck BMD was significant (OR ؍ 1.64, 95% confidence interval [CI] 1.06 -2.55). In contrast, HAL was not significantly different from controls in the trochanteric fracture group. Femoral neck diameter was not a predictor of fracture. The upper and lower femoral neck BMD was lower in the trochanteric fracture group than in controls, and both measurements predicted trochanteric femoral neck fracture. In contrast, the prediction of cervical femoral neck fracture was enhanced by measuring only the upper part of the femoral neck (OR ؍ 2.79 vs. 1.97 for the total femoral neck) while BMD of the lower part was not different from controls. Hip axis length is a predictor of femoral neck fracture. Femoral neck BMD distribution is different between cervical and trochanteric fractures. These results support the hypothesis of a different pathophysiological mechanism between the two types of hip fractures
Medical image computing and computer-assisted intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention, 2011
This work presents a statistical model of both the shape and Bone Mineral Density (BMD) distribution of the proximal femur for fracture risk assessment. The shape and density model was built from a dataset of Quantitative Computed Tomography scans of fracture patients and a control group. Principal Component Analysis and Horn's parallel analysis were used to reduce the dimensionality of the shape and density model to the main modes of variation. The input data was then used to analyze the model parameters for the optimal separation between the fracture and control group. Feature selection using the Fisher criterion determined the parameters with the best class separation, which were used in Fisher Linear Discriminant Analysis to find the direction in the parameter space that best separates the fracture and control group. This resulted in a Fisher criterion value of 6.70, while analyzing the Dual-energy X-ray Absorptiometry derived femur neck areal BMD of the same subjects result...