Abstract
Studies using quantitative computed tomography (QCT) and data-driven image analysis techniques have shown that trabecular and cortical volumetric bone mineral density (vBMD) can improve the hip fracture prediction of dual-energy X-ray absorptiometry areal BMD (aBMD). Here, we hypothesize that (1) QCT imaging features of shape, density and structure derived from data-driven image analysis techniques can improve the hip fracture discrimination of classification models based on mean femoral neck aBMD (Neck.aBMD), and (2) that data-driven cortical bone thickness (Ct.Th) features can improve the hip fracture discrimination of vBMD models. We tested our hypotheses using statistical multi-parametric modeling (SMPM) in a QCT study of acute hip fracture of 50 controls and 93 fragility fracture cases. SMPM was used to extract features of shape, vBMD, Ct.Th, cortical vBMD, and vBMD in a layer adjacent to the endosteal surface to develop hip fracture classification models with machine learning logistic LASSO. The performance of these classification models was evaluated in two aspects: (1) their hip fracture classification capability without Neck.aBMD, and (2) their capability to improve the hip fracture classification of the Neck.aBMD model. Assessments were done with 10-fold cross-validation, areas under the receiver operating characteristic curve (AUCs), differences of AUCs, and the integrated discrimination improvement (IDI) index. All LASSO models including SMPM-vBMD features, and the majority of models including SMPM-Ct.Th features performed significantly better than the Neck.aBMD model; and all SMPM features significantly improved the hip fracture discrimination of the Neck.aBMD model (Hypothesis 1). An interesting finding was that SMPM-features of vBMD also captured Ct.Th patterns, potentially explaining the superior classification performance of models based on SMPM-vBMD features (Hypothesis 2). Age, height and weight had a small impact on model performances, and the model of shape, vBMD and Ct.Th consistently yielded better performances than the Neck.aBMD models. Results of this study clearly support the relevance of bone density and quality on the assessment of hip fracture, and demonstrate their potential on patient and healthcare cost benefits.
Similar content being viewed by others
References
Allison, S. J., K. E. S. Poole, G. M. Treece, A. H. Gee, C. Tonkin, W. J. Rennie, J. P. Folland, G. D. Summers, and K. Brooke-Wavell. The influence of high-impact exercise on cortical and trabecular bone mineral content and 3D distribution across the proximal femur in older men: a randomized controlled unilateral intervention. J. Bone Miner. Res. 30(9):1709–1716, 2015.
Baker-LePain, J. C., K. R. Luker, J. A. Lynch, N. Parimi, M. C. Nevitt, and N. E. Lane. Active shape modeling of the hip in the prediction of incident hip fracture. J. Bone Miner. Res. 26(3):468–474, 2011.
Bauer, D. C., P. Garnero, J. P. Bilezikian, S. L. Greenspan, K. E. Ensrud, C. J. Rosen, L. Palermo, and D. M. Black. Short-term changes in bone turnover markers and bone mineral density response to parathyroid hormone in postmenopausal women with osteoporosis. J. Clin. Endocrinol. Metab. 91(4):1370–1375, 2006.
Berry, S. D., E. J. Samelson, M. J. Pencina, R. R. McLean, L. A. Cupples, K. E. Broe, and D. P. Kiel. Repeat bone mineral density screening and prediction of hip and major osteoporotic fracture. JAMA 310(12):1256–1262, 2013.
Black, D. M., M. L. Bouxsein, L. M. Marshall, S. R. Cummings, T. F. Lang, J. A. Cauley, K. E. Ensrud, C. M. Nielson, E. S. Orwoll, and G. Osteoporotic Fractures in Men Research. Proximal femoral structure and the prediction of hip fracture in men: a large prospective study using QCT. J. Bone Miner. Res. 23(8):1326–1333, 2008.
Blank, J. B., P. M. Cawthon, M. L. Carrion-Petersen, L. Harper, J. P. Johnson, E. Mitson, and R. R. Delay. Overview of recruitment for the osteoporotic fractures in men study (MrOS). Contemp. Clin. Trials 26(5):557–568, 2005.
Bousson, V. D., J. Adams, K. Engelke, M. Aout, M. Cohen-Solal, C. Bergot, D. Haguenauer, D. Goldberg, K. Champion, R. Aksouh, E. Vicaut, and J. D. Laredo. In vivo discrimination of hip fracture with quantitative computed tomography: results from the prospective European Femur Fracture Study (EFFECT). J. Bone Miner. Res. 26(4):881–893, 2011.
Bredbenner, T. L., R. L. Mason, L. M. Havill, E. S. Orwoll, D. P. Nicolella, and S. Osteoporotic Fractures in Men. Fracture risk predictions based on statistical shape and density modeling of the proximal femur. J. Bone Miner. Res. 29(9):2090–2100, 2014.
Burge, R., B. Dawson-Hughes, D. H. Solomon, J. B. Wong, A. King, and A. Tosteson. Incidence and economic burden of osteoporosis-related fractures in the United States, 2005-2025. J. Bone Miner. Res. 22(3):465–475, 2007.
Carballido-Gamio, J., S. Bonaretti, I. Saeed, R. Harnish, R. Recker, A. J. Burghardt, J. H. Keyak, T. Harris, S. Khosla, and T. F. Lang. Automatic multi-parametric quantification of the proximal femur with quantitative computed tomography. Quant. Imaging Med. Surg. 5(4):552–568, 2015.
Carballido-Gamio, J., R. Harnish, I. Saeed, T. Streeper, S. Sigurdsson, S. Amin, E. J. Atkinson, T. M. Therneau, K. Siggeirsdottir, X. Cheng, L. J. Melton, 3rd, J. Keyak, V. Gudnason, S. Khosla, T. B. Harris, and T. F. Lang. Proximal femoral density distribution and structure in relation to age and hip fracture risk in women. J. Bone Miner. Res. 28(3):537–546, 2013.
Carballido-Gamio, J., R. Harnish, I. Saeed, T. Streeper, S. Sigurdsson, S. Amin, E. J. Atkinson, T. M. Therneau, K. Siggeirsdottir, X. Cheng, L. J. Melton, 3rd, J. H. Keyak, V. Gudnason, S. Khosla, T. B. Harris, and T. F. Lang. Structural patterns of the proximal femur in relation to age and hip fracture risk in women. Bone 57(1):290–299, 2013.
Carballido-Gamio, J., and D. P. Nicolella. Computational anatomy in the study of bone structure. Curr. Osteoporos Rep. 11(3):237–245, 2013.
Carballido-Gamio, J., A. Yu, L. Wang, S. Yongbin, T. F. Lang, and X. Cheng. Fracture risk estimation with statistical multi-parametric modeling. ASBMR Annual Meeting, 2016.
Cootes, T. F. and C. J. Taylor. Statistical models of appearance for medical image analysis and computer vision. Medical Imaging: 2001: Image Processing, Pts 1–3 2(27):236–248, 2001.
Cootes, T. F., C. J. Taylor, D. H. Cooper, and J. Graham. Active shape models—their training and application. Comput. Vis. Image Underst. 61(1):38–59, 1995.
Crabtree, N. J., H. Kroger, A. Martin, H. A. Pols, R. Lorenc, J. Nijs, J. J. Stepan, J. A. Falch, T. Miazgowski, S. Grazio, P. Raptou, J. Adams, A. Collings, K. T. Khaw, N. Rushton, M. Lunt, A. K. Dixon and J. Reeve. Improving risk assessment: hip geometry, bone mineral distribution and bone strength in hip fracture cases and controls. The EPOS study. European Prospective Osteoporosis Study. Osteoporos Int. 13(1):48–54, 2002.
Dong, X. L. N., R. Pinninti, T. Lowe, P. Cussen, J. E. Ballard, D. Di Paolo, and M. Shirvaikar. Random field assessment of inhomogeneous bone mineral density from DXA scans can enhance the differentiation between postmenopausal women with and without hip fractures. J. Biomech. 48(6):1043–1051, 2015.
Eastell, R., T. Lang, S. Boonen, S. Cummings, P. D. Delmas, J. A. Cauley, Z. Horowitz, E. Kerzberg, G. Bianchi, D. Kendler, P. Leung, Z. Man, P. Mesenbrink, E. F. Eriksen, D. M. Black, and H. P. F. Trial. Effect of once-yearly zoledronic acid on the spine and hip as measured by quantitative computed tomography: results of the HORIZON Pivotal Fracture Trial. Osteoporos Int. 21(7):1277–1285, 2010.
Engelke, K., T. Fuerst, B. Dardzinski, J. Kornak, S. Ather, H. K. Genant, and A. de Papp. Odanacatib treatment affects trabecular and cortical bone in the femur of postmenopausal women: results of a two-year placebo-controlled trial. J. Bone Miner. Res. 30(1):30–38, 2015.
Engelke, K., T. Fuerst, G. Dasic, R. Y. Davies, and H. K. Genant. Regional distribution of spine and hip QCT BMD responses after one year of once-monthly ibandronate in postmenopausal osteoporosis. Bone 46(6):1626–1632, 2010.
Engelke, K., T. Lang, S. Khosla, L. Qin, P. Zysset, W. D. Leslie, J. A. Shepherd, and J. T. Schousboe. Clinical Use of Quantitative Computed Tomography (QCT) of the Hip in the Management of Osteoporosis in Adults: the 2015 ISCD Official Positions-Part I. J. Clin. Densitom. 18(3):338–358, 2015.
Genant, H. K., C. Libanati, K. Engelke, J. R. Zanchetta, A. Hoiseth, C. K. Yuen, S. Stonkus, M. A. Bolognese, E. Franek, T. Fuerst, H. S. Radcliffe, and M. R. McClung. Improvements in hip trabecular, subcortical, and cortical density and mass in postmenopausal women with osteoporosis treated with denosumab. Bone 56(2):482–488, 2013.
Goodyear, S. R., R. J. Barr, E. McCloskey, S. Alesci, R. M. Aspden, D. M. Reid, and J. S. Gregory. Can we improve the prediction of hip fracture by assessing bone structure using shape and appearance modelling? Bone 53(1):188–193, 2013.
Gregory, J. S., A. Stewart, P. E. Undrill, D. M. Reid, and R. M. Aspden. Bone shape, structure, and density as determinants of osteoporotic hip fracture: a pilot study investigating the combination of risk factors. Invest. Radiol. 40(9):591–597, 2005.
Gregory, J. S., D. Testi, A. Stewart, P. E. Undrill, D. M. Reid, and R. M. Aspden. A method for assessment of the shape of the proximal femur and its relationship to osteoporotic hip fracture. Osteoporos. Int. 15(1):5–11, 2004.
Harris, T. B., L. J. Launer, G. Eiriksdottir, O. Kjartansson, P. V. Jonsson, G. Sigurdsson, G. Thorgeirsson, T. Aspelund, M. E. Garcia, M. F. Cotch, H. J. Hoffman, and V. Gudnason. Age, Gene/Environment Susceptibility-Reykjavik Study: multidisciplinary applied phenomics. Am. J. Epidemiol. 165(9):1076–1087, 2007.
Johannesdottir, F., T. Turmezei, and K. E. Poole. Cortical bone assessed with clinical computed tomography at the proximal femur. J. Bone Miner. Res. 29(4):771–783, 2014.
Keaveny, T. M., P. F. Hoffmann, M. Singh, L. Palermo, J. P. Bilezikian, S. L. Greenspan, and D. M. Black. Femoral bone strength and its relation to cortical and trabecular changes after treatment with PTH, alendronate, and their combination as assessed by finite element analysis of quantitative CT scans. J. Bone Miner. Res. 23(12):1974–1982, 2008.
Keyak, J. H., S. Sigurdsson, G. Karlsdottir, D. Oskarsdottir, A. Sigmarsdottir, S. Zhao, J. Kornak, T. B. Harris, G. Sigurdsson, B. Y. Jonsson, K. Siggeirsdottir, G. Eiriksdottir, V. Gudnason, and T. F. Lang. Male-female differences in the association between incident hip fracture and proximal femoral strength: a finite element analysis study. Bone 48(6):1239–1245, 2011.
Keyak, J. H., S. Sigurdsson, G. S. Karlsdottir, D. Oskarsdottir, A. Sigmarsdottir, J. Kornak, T. B. Harris, G. Sigurdsson, B. Y. Jonsson, K. Siggeirsdottir, G. Eiriksdottir, V. Gudnason, and T. F. Lang. Effect of finite element model loading condition on fracture risk assessment in men and women: the AGES-Reykjavik study. Bone 57(1):18–29, 2013.
Lane, N. E., S. Sanchez, G. W. Modin, H. K. Genant, E. Pierini, and C. D. Arnaud. Bone mass continues to increase at the hip after parathyroid hormone treatment is discontinued in glucocorticoid-induced osteoporosis: results of a randomized controlled clinical trial. J. Bone Miner. Res. 15(5):944–951, 2000.
Lang, T. F., I. H. Saeed, T. Streeper, J. Carballido-Gamio, R. J. Harnish, L. A. Frassetto, S. M. Lee, J. D. Sibonga, J. H. Keyak, B. A. Spiering, C. M. Grodsinsky, J. J. Bloomberg, and P. R. Cavanagh. Spatial heterogeneity in the response of the proximal femur to two lower-body resistance exercise regimens. J. Bone Miner. Res. 29(6):1337–1345, 2014.
Leslie, W. D., P. S. Pahlavan, J. F. Tsang, L. M. Lix, and P. Manitoba Bone Density. Prediction of hip and other osteoporotic fractures from hip geometry in a large clinical cohort. Osteoporos. Int 20(10):1767–1774, 2009.
Lewiecki, E. M., T. M. Keaveny, D. L. Kopperdahl, H. K. Genant, K. Engelke, T. Fuerst, A. Kivitz, R. Y. Davies, and L. A. Fitzpatrick. Once-monthly oral ibandronate improves biomechanical determinants of bone strength in women with postmenopausal osteoporosis. J. Clin. Endocrinol. Metab. 94(1):171–180, 2009.
Li, G. W., S. X. Chang, Z. Xu, Y. Chen, H. Bao, and X. Shi. Prediction of hip osteoporotic fractures from composite indices of femoral neck strength. Skelet. Radiol. 42(2):195–201, 2013.
Li, W., J. Kornak, T. Harris, J. Keyak, C. Li, Y. Lu, X. Cheng, and T. Lang. Identify fracture-critical regions inside the proximal femur using statistical parametric mapping. Bone 44(4):596–602, 2009.
Orwoll, E., J. B. Blank, E. Barrett-Connor, J. Cauley, S. Cummings, K. Ensrud, C. Lewis, P. M. Cawthon, R. Marcus, L. M. Marshall, J. McGowan, K. Phipps, S. Sherman, M. L. Stefanick, and K. Stone. Design and baseline characteristics of the osteoporotic fractures in men (MrOS) study—a large observational study of the determinants of fracture in older men. Contemp. Clin. Trials 26(5):569–585, 2005.
Poole, K. E., G. M. Treece, P. M. Mayhew, J. Vaculik, P. Dungl, M. Horak, J. J. Stepan, and A. H. Gee. Cortical thickness mapping to identify focal osteoporosis in patients with hip fracture. PLoS ONE 7(6):e38466, 2012.
Qian, J., T. Hastie, J. Friedman, R. Tibshirani and N. Simon. Glmnet for Matlab, 2013. http://www.stanford.edu/~hastie/glmnet_matlab/.
Schuler, B., K. D. Fritscher, V. Kuhn, F. Eckstein, T. M. Link, and R. Schubert. Assessment of the individual fracture risk of the proximal femur by using statistical appearance models. Med. Phys. 37(6):2560–2571, 2010.
Sellmeyer, D. E., D. M. Black, L. Palermo, S. Greenspan, K. Ensrud, J. Bilezikian, and C. J. Rosen. Hetereogeneity in skeletal response to full-length parathyroid hormone in the treatment of osteoporosis. Osteoporos. Int. 18(7):973–979, 2007.
Tibshirani, R. Regression shrinkage and selection via the lasso. J. R. Stat. Soc B. 58(1):267–288, 1996.
Treece, G. M., A. H. Gee, C. Tonkin, S. K. Ewing, P. M. Cawthon, D. M. Black, K. E. Poole, and S. Osteoporotic Fractures in Men. Predicting hip fracture type with cortical bone mapping (CBM) in the Osteoporotic Fractures in Men (MrOS) Study. J. Bone Miner. Res. 30(11):2067–2077, 2015.
Walker, M. D., I. Saeed, D. J. McMahon, J. Udesky, G. Liu, T. Lang, and J. P. Bilezikian. Volumetric bone mineral density at the spine and hip in Chinese American and White women. Osteoporos. Int. 23(10):2499–2506, 2012.
Whitmarsh, T., K. D. Fritscher, L. Humbert, L. M. Del Rio Barquero, T. Roth, C. Kammerlander, M. Blauth, R. Schubert, and A. F. Frangi. A statistical model of shape and bone mineral density distribution of the proximal femur for fracture risk assessment. Med. Image Comput. Comput. Assist. Interv. 14(Pt 2):393–400, 2011.
Wiener, J. M., and J. Tilly. Population ageing in the United States of America: implications for public programmes. Int. J. Epidemiol. 31(4):776–781, 2002.
Yang, L., W. J. M. Udall, E. V. McCloskey, and R. Eastell. Distribution of bone density and cortical thickness in the proximal femur and their association with hip fracture in postmenopausal women: a quantitative computed tomography study. Osteoporos. Int. 25(1):251–263, 2014.
Acknowledgments
This work was supported by the NIH/NIAMS under grants R01AR068456 and R01AR064140. This study was also supported by grants from the National Natural Science Foundation of China (81071131), the Beijing Bureau of Health 215 Program (2013-3-033; 2009-2-03), Beijing Technology Foundation for Selected Overseas Chinese Scholar and Beijing Talents Fund (2015000021467), Capital Characteristic Clinic Project (Z141107002514072).
Author information
Authors and Affiliations
Corresponding author
Additional information
Associate Editor Stefan M Duma oversaw the review of this article.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
This section has the purpose of supporting the necessity of a third PCA step to address potential correlations between shape and feature principal component scores.
In this study, for the first 10 principal components, out of 100 correlations of principal component scores: (1) 37% between shape and vBMD, (2) 43% between shape and Ct.Th, (3) 29% between shape and Ct.vBMD, and (4) 40% between shape and EndoTb.vBMD, were significant.
Rights and permissions
About this article
Cite this article
Carballido-Gamio, J., Yu, A., Wang, L. et al. Hip Fracture Discrimination Based on Statistical Multi-parametric Modeling (SMPM). Ann Biomed Eng 47, 2199–2212 (2019). https://doi.org/10.1007/s10439-019-02298-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10439-019-02298-x