Ludovic Humbert

Reconstructing the 3D Shape and Bone Mineral Density Distribution of the Proximal Femur From Dual-Energy X-Ray Absorptiometry

The accurate diagnosis of osteoporosis has gained increasing importance due to the aging of our society. Areal bone mineral density (BMD) measured by dual-energy X-ray absorptiometry (DXA) is an established criterion in the diagnosis of osteoporosis. This measure, however, is limited by its two-dimensionality. This work presents a method to reconstruct both the 3D bone shape and 3D BMD distribution of the proximal femur from a single DXA image used in clinical routine. A statistical model of the combined shape and BMD distribution is presented, together with a method for its construction from a set of quantitative computed tomography (QCT) scans. A reconstruction is acquired in an intensity based 3D-2D registration process whereby an instance of the model is found that maximizes the similarity between its projection and the DXA image. Reconstruction experiments were performed on the DXA images of 30 subjects, with a model constructed from a database of QCT scans of 85 subjects. The accuracy was evaluated by comparing the reconstructions with the same subject QCT scans. The method presented here can potentially improve the diagnosis of osteoporosis and fracture risk assessment from the low radiation dose and low cost DXA devices currently used in clinical routine.

Hip fracture discrimination from dual-energy X-ray absorptiometry by statistical model registration

Although the areal Bone Mineral Density (BMD) measurements from dual-energy X-ray absorptiometry (DXA) are able to discriminate between hip fracture cases and controls, the femoral strength is largely determined by the 3D bone structure. In a previous work a statistical model was presented which parameterizes the 3D shape and BMD distribution of the proximal femur. In this study the parameter values resulting from the registration of the model onto DXA images are evaluated for their hip fracture discrimination ability with respect to regular DXA derived areal BMD measurements. The statistical model was constructed from a large database of QCT scans of females with an average age of 67.8 ± 17.0 years. This model was subsequently registered onto the DXA images of a fracture and control group. The fracture group consisted of 175 female patients with an average age of 66.4 ± 9.9 years who suffered a fracture on the contra lateral femur. The control group consisted of 175 female subjects with an average age of 65.3 ± 10.0 years and no fracture history. The discrimination ability of the resulting model parameter values, as well as the areal BMD measurements extracted from the DXA images were evaluated using a logistic regression analysis. The area under the receiver operating curve (AUC) of the combined model parameters and areal BMD values was 0.840 (95% CI 0.799-0.881), whilst using only the areal BMD values resulted in an AUC of 0.802 (95% CI 0.757-0.848). These results indicate that the discrimination ability of the areal BMD values is improved by supplementing them with the model parameter values, which give a more complete representation of the subject specific shape and internal bone distribution. Thus, the presented method potentially allows for an improved hip fracture risk estimation whilst maintaining DXA as the current standard modality.

A statistical model of shape and bone mineral density distribution of the proximal femur for fracture risk assessment

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 Horns 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 resulted in a Fisher criterion value of 0.98. This indicates that a fracture risk estimation approach based on the presented model might improve upon the current standard clinical practice.

3D reconstruction of both shape and Bone Mineral Density distribution of the femur from DXA images

The diagnosis of osteoporosis and the prevention of femur fractures is a major challenge for our society. However, the diagnosis performed in clinical routine from Dual Energy X-ray Absorptiometry (DXA) images is limited. This paper proposes a 3D reconstruction method of both the shape and the Bone Mineral Density (BMD) distribution of the proximal femur from routinely used DXA images. The reconstruction accuracy that can be obtained from single-view and multi-view DXA devices was assessed. This evaluation, from 20 bone specimens and simulated DXA images, highlighted a mean shape accuracy of 1.3mm and a BMD accuracy of 4.4% from a single-view DXA image. A multi-view configuration with 2 views (frontal-sagittal) appeared as a good compromise (mean shape accuracy of 0.9mm and BMD accuracy of 3.2%). We are currently using this method for in vivo clinical studies in order to improve the diagnosis of osteoporosis and the prevention of femur fractures.

3D bone mineral density distribution and shape reconstruction of the proximal femur from a single simulated DXA image: an in vitro study

Area Bone Mineral Density (aBMD) measured by Dual-energy X-ray Absorptiometry (DXA) is an established criterion in the evaluation of hip fracture risk. The evaluation from these planar images, however, is limited to 2D while it has been shown that proper 3D assessment of both the shape and the Bone Mineral Density (BMD) distribution improves the fracture risk estimation. In this work we present a method to reconstruct both the 3D bone shape and 3D BMD distribution of the proximal femur from a single DXA image. A statistical model of shape and a separate statistical model of the BMD distribution were automatically constructed from a set of Quantitative Computed Tomography (QCT) scans. The reconstruction method incorporates a fully automatic intensity based 3D-2D registration process, maximizing the similarity between the DXA and a digitally reconstructed radiograph of the combined model. For the construction of the models, an in vitro dataset of QCT scans of 60 anatomical specimens was used. To evaluate the reconstruction accuracy, experiments were performed on simulated DXA images from the QCT scans of 30 anatomical specimens. Comparisons between the reconstructions and the same subject QCT scans showed a mean shape accuracy of 1.2mm, and a mean density error of 81mg/cm3. The results show that this method is capable of accurately reconstructing both the 3D shape and 3D BMD distribution of the proximal femur from DXA images used in clinical routine, potentially improving the diagnosis of osteoporosis and fracture risk assessments at a low radiation dose and low cost.

3D-DXA: Assessing the Femoral Shape, the Trabecular Macrostructure and the Cortex in 3D from DXA images

The 3D distribution of the cortical and trabecular bone mass in the proximal femur is a critical component in determining fracture resistance that is not taken into account in clinical routine Dual-energy X-ray Absorptiometry (DXA) examination. In this paper, a statistical shape and appearance model together with a 3D-2D registration approach are used to model the femoral shape and bone density distribution in 3D from an anteroposterior DXA projection. A model-based algorithm is subsequently used to segment the cortex and build a 3D map of the cortical thickness and density. Measurements characterising the geometry and density distribution were computed for various regions of interest in both cortical and trabecular compartments. Models and measurements provided by the “3D-DXA” software algorithm were evaluated using a database of 157 study subjects, by comparing 3D-DXA analyses (using DXA scanners from three manufacturers) with measurements performed by Quantitative Computed Tomography (QCT). The mean point-to-surface distance between 3D-DXA and QCT femoral shapes was 0.93 mm. The mean absolute error between cortical thickness and density estimates measured by 3D-DXA and QCT was 0.33 mm and 72 mg/cm3. Correlation coefficients (R) between the 3D-DXA and QCT measurements were 0.86, 0.93, and 0.95 for the volumetric bone mineral density at the trabecular, cortical, and integral compartments respectively, and 0.91 for the mean cortical thickness. 3D-DXA provides a detailed analysis of the proximal femur, including a separate assessment of the cortical layer and trabecular macrostructure, which could potentially improve osteoporosis management while maintaining DXA as the standard routine modality.

Technical note: cortical thickness and density estimation from clinical CT using a prior thickness-density relationship

PURPOSE Cortical thickness and density are critical components in determining the strength of bony structures. Computed tomography (CT) is one possible modality for analyzing the cortex in 3D. In this paper, a model-based approach for measuring the cortical bone thickness and density from clinical CT images is proposed. METHODS Density variations across the cortex were modeled as a function of the cortical thickness and density, location of the cortex, density of surrounding tissues, and imaging blur. High resolution micro-CT data of cadaver proximal femurs were analyzed to determine a relationship between cortical thickness and density. This thickness-density relationship was used as prior information to be incorporated in the model to obtain accurate measurements of cortical thickness and density from clinical CT volumes. The method was validated using micro-CT scans of 23 cadaver proximal femurs. Simulated clinical CT images with different voxel sizes were generated from the micro-CT data. Cortical thickness and density were estimated from the simulated images using the proposed method and compared with measurements obtained using the micro-CT images to evaluate the effect of voxel size on the accuracy of the method. Then, 19 of the 23 specimens were imaged using a clinical CT scanner. Cortical thickness and density were estimated from the clinical CT images using the proposed method and compared with the micro-CT measurements. Finally, a case-control study including 20 patients with osteoporosis and 20 age-matched controls with normal bone density was performed to evaluate the proposed method in a clinical context. RESULTS Cortical thickness (density) estimation errors were 0.07 ± 0.19 mm (-18 ± 92 mg/cm(3)) using the simulated clinical CT volumes with the smallest voxel size (0.33 × 0.33 × 0.5 mm(3)), and 0.10 ± 0.24 mm (-10 ± 115 mg/cm(3)) using the volumes with the largest voxel size (1.0 × 1.0 × 3.0 mm(3)). A trend for the cortical thickness and density estimation errors to increase with voxel size was observed and was more pronounced for thin cortices. Using clinical CT data for 19 of the 23 samples, mean errors of 0.18 ± 0.24 mm for the cortical thickness and 15 ± 106 mg/cm(3) for the density were found. The case-control study showed that osteoporotic patients had a thinner cortex and a lower cortical density, with average differences of -0.8 mm and -58.6 mg/cm(3) at the proximal femur in comparison with age-matched controls (p-value < 0.001). CONCLUSIONS This method might be a promising approach for the quantification of cortical bone thickness and density using clinical routine imaging techniques. Future work will concentrate on investigating how this approach can improve the estimation of mechanical strength of bony structures, the prevention of fracture, and the management of osteoporosis.

3D reconstruction of the lumbar vertebrae from anteroposterior and lateral dual-energy X-ray absorptiometry

Current vertebral fracture prevention measures use Dual-energy X-ray Absorptiometry (DXA) to quantify the density of the vertebrae and subsequently determine the risk of fracture. This modality however only provides information about the projected Bone Mineral Density (BMD) while the shape and spatial distribution of the bone determines the strength of the vertebrae. Quantitative Computed Tomography (QCT) allows for the measurement of the vertebral dimensions and volumetric densities, which have been shown to be able to determine the fracture risk more reliably than DXA. However, for the high cost and high radiation dose, QCT is not used in clinical routine for fracture risk assessment. In this work, we therefore propose a method to reconstruct the 3D shape and density volume of lumbar vertebrae from an anteroposterior (AP) and lateral DXA image used in clinical routine. The method is evaluated for the L2, L3 and L4 vertebra. Of these vertebrae a statistical model of the vertebral shape and density distribution is first constructed from a large dataset of QCT scans. All three models are then simultaneously registered onto both AP and lateral DXA image. The shape and volumetric BMD at several regions of the reconstructed vertebrae is then evaluated with respect to the ground truth QCT volumes. For the L2, L3 and L4 vertebrae respectively the shape was reconstructed with a mean (2RMS) point-to-surface distance of 1.00 (2.64) mm, 0.93(2.52) mm and 1.34(3.72) mm and a strong correlation (r > 0.82) was found between the trabecular volumetric BMD extracted from the reconstructions and from the same subject QCT scans. These results indicate that the proposed method is able to accurately reconstruct the 3D shape and density volume of the lumbar vertebrae from AP and lateral DXA, which can potentially improve the fracture risk estimation accuracy with respect to the currently used DXA derived areal BMD measurements.

Technical Note: Comparison between single and multiview simulated DXA configurations for reconstructing the 3D shape and bone mineral density distribution of the proximal femur

PURPOSE Dual-energy x-ray absorptiometry (DXA) is used in clinical routine to provide a two-dimensional (2D) analysis of the bone mineral density (BMD). 3D reconstruction methods from 2D DXA images could improve the BMD analysis. To find the optimal configuration that should be used in clinical routine, this paper relies on a 3D reconstruction method from DXA images to compare the accuracy that can be obtained from one single-view and from multiview DXA images (two to four projections). METHODS The 3D reconstruction method uses a statistical model and a nonrigid registration technique to recover in 3D the shape and the BMD distribution of the proximal femur. The accuracy was evaluated in vivo by comparing 3D reconstructions obtained from simulated DXA images of 30 patients (using between one and four DXA views) with quantitative computed tomography reconstructions. RESULTS This comparison showed that the use of one single DXA provides accurate 3D reconstructions (mean shape accuracy of 1.0 mm and BMD distribution errors of 7.0%). Among the multiview configurations, the use of two views (0° and 45°) was the best compromise, increasing the accuracy of pose (mean accuracy of 0.7°/1.2°/0.9° against 1.0°/3.5°/3.3° for the single view), reducing slightly the BMD errors (5.7%) while maintaining the same shape accuracy. CONCLUSIONS The use of two views constitutes an interesting configuration when multiview DXA devices are available in clinical routine. However, the use of only one single view remains an accurate solution to recover the shape and the BMD distribution in 3D, with the advantage of a higher potential for clinical translation.

Age-Related changes in vertebral morphometry by statistical shape analysis

The morphological changes of the vertebrae associated with normal aging are still subject of debate, whereas this knowledge is important in detecting vertebral fractures and degenerative shape changes. The aim of this study is to present a method to statistically analyze the vertebral shape and determining the morphometric changes related to normal aging. The analysis is performed on the L2 lumbar vertebrae from a large dataset of Computed Tomography scans. The surface meshes of all vertebrae, with a groupwise vertex correspondence between them, are first acquired by an intensity based registration process onto a segmented reference. Principal component analysis then reduces the dimensionality to the main modes of variation which were subsequently analyzed by multiple linear regression to acquire the global shape variations with respect to the age of the subjects. In addition, the correlation with age of the deformation at each mesh vertex is analyzed, giving a significance map of the age related changes. This analysis shows several shape changes which are in agreement with previous studies while also giving a more detailed global shape analysis. Understanding the normal shape changes allows for a better diagnosis of vertebral fractures and spinal pathologies.