Ultrasound-Based Estimates of Cortical Bone Thickness and Porosity Are Associated With Nontraumatic Fractures in Postmenopausal Women: A Pilot Study

Abstract

Recent ultrasound (US) axial transmission techniques exploit the multimode waveguide response of long bones to yield estimates of cortical bone structure characteristics. This pilot cross‐sectional study aimed to evaluate the performance at the one‐third distal radius of a bidirectional axial transmission technique (BDAT) to discriminate between fractured and nonfractured postmenopausal women. Cortical thickness (Ct.Th) and porosity (Ct.Po) estimates were obtained for 201 postmenopausal women: 109 were nonfractured (62.6 ± 7.8 years), 92 with one or more nontraumatic fractures (68.8 ± 9.2 years), 17 with hip fractures (66.1 ± 10.3 years), 32 with vertebral fractures (72.4 ± 7.9 years), and 17 with wrist fractures (67.8 ± 9.6 years). The areal bone mineral density (aBMD) was obtained using DXA at the femur and spine. Femoral aBMD correlated weakly, but significantly with Ct.Th (R = 0.23, p < 0.001) and Ct.Po (R = ‐0.15, p < 0.05). Femoral aBMD and both US parameters were significantly different between the subgroup of all nontraumatic fractures combined and the control group (p < 0.05). The main findings were that (1) Ct.Po was discriminant for all nontraumatic fractures combined (OR = 1.39; area under the receiver operating characteristic curve [AUC] equal to 0.71), for vertebral (OR = 1.96; AUC = 0.84) and wrist fractures (OR = 1.80; AUC = 0.71), whereas Ct.Th was discriminant for hip fractures only (OR = 2.01; AUC = 0.72); there was a significant association (2) between increased Ct.Po and vertebral and wrist fractures when these fractures were not associated with any measured aBMD variables; (3) between increased Ct.Po and all nontraumatic fractures combined independently of aBMD neck; and (4) between decreased Ct.Th and hip fractures independently of aBMD femur. BDAT variables showed comparable performance to that of aBMD neck with all types of fractures (OR = 1.48; AUC = 0.72) and that of aBMD femur with hip fractures (OR = 2.21; AUC = 0.70). If these results are confirmed in prospective studies, cortical BDAT measurements may be considered useful for assessing fracture risk in postmenopausal women. © 2019 American Society for Bone and Mineral Research.

Automatic Classifying of Patients With Non-Traumatic Fractures Based on Ultrasonic Guided Wave Spectrum Image Using a Dynamic Support Vector Machine

“…It can be noted that this approach was then extended to an inverse procedure providing estimates of two cortical parameters: Ct.Th and Ct.Po [29]. Inverse problem results obtained in [28] will be used as a baseline in the following.…”

“…Recently, estimates of Ct.Th and Ct.Po, derived from guided wave spectrum measured by bi-directional axial transmission (BDAT), have been tested in a pilot clinical study [28]. In this, study, Ct.Po discrimination between fractured and non-fractured patients, was found similar to DXA, with an AUC, Area Under the Receiver Operator Characteristic (ROC) Curve, equal to 71%, after adjustement performed according to age, Body Mass Index (BMI) and cortisone treatment.…”

“…Indeed, the success of an inverse problem, i.e., its ability to deliver a robust and non ambiguous solution, depends on the measurement quality and the ability of the model to represent realistic cortical bone layer. In particular, some BDAT measurement failures have been associated with large soft tissue thickness [28] or irregular inner interface [29]. Thus, the objective of this study is to explore if those two limitations of the BDAT approach, moderate failure rate and fracture discrimination similar to DXA, could be overcome by automatic classification tools.…”

“…Thus in this paper, automatic classification tools have been tested on the same clinical data set of guided wave spectrum images obtained by BDAT in a previous crosssectional study [28]. To achieve our aim, we build a dynamic SVM model that allows us to classify non-fractured patients and patients who suffered non-traumatic fractures.…”

Influence of guided waves in bone on pulse‐inversion contrast‐enhanced ultrasound

“…[21][22][23] Frequency-dispersive guided waves have been extensively used for cortical bone characterization. [28][29][30] However, only few studies notice and discuss the influence of guided waves on CEUS, even if the guided waves have been used to detect and trap microdroplets and gas bubbles in acoustic manipulation. [31][32][33] In our previous studies, we observed the microbubbles' nonlinear scattering altered by the guided waves under a fixed phase transmission, which is verified by using the cross-validations of simulated and measured results from bone cortex and plate model.…”