Osteoporosis Prediction Using Machine-Learned Optical Bone Densitometry Data.

Optical bone densitometry (OBD) has been developed for the early detection of osteoporosis. In recent years, machine learning (ML) techniques have been actively implemented for the areas of medical diagnosis and screening with the goal of improving diagnostic accuracy. The purpose of this study was to verify the feasibility of using the combination of OBD and ML techniques as a screening tool for osteoporosis. Dual energy X-ray absorptiometry (DXA) and OBD measurements were performed on 203 Thai subjects. From the OBD measurements and readily available demographic data, machine learning techniques were used to predict the T-score measured by the DXA. The T-score predicted using the Ridge regressor had a correlation of r = 0.512 with respect to the reference value. The predicted T-score also showed an AUC of 0.853 for discriminating individuals with osteoporosis. The results obtained suggest that the developed model is reliable enough to be used for screening for osteoporosis.

Optical bone densitometry insensitive to skin thickness.

Skin thickness, including the adipose layer, which varies from individual to individual, affects the bone density measurement using light. In this study, we proposed a method to measure skin thickness using light and to correct the bias caused by differences in skin thickness and verified the proposed method by experiments using a phantom. We measured simulated skin of different thicknesses and bovine trabecular bone of different bone mineral densities (BMDs) using an optical system consisting of lasers of 850 and 515 nm wavelengths, lenses, and slits. Although the slope of the light intensity distribution formed on the surface of the material when irradiated by the 850 nm laser is affected by the thickness of the skin phantom. The difference of the intensity distribution peaks (δy) between the 850 and 515 nm lasers was strongly correlated with the thickness of the skin phantom. The coefficient of determination between the measurements and the BMD was improved by correcting the 850 nm laser measurements with δy. This result suggests that the method is applicable to optical bone densitometry, which is insensitive to differences in skin thickness.

Optical bone densitometry robust to variation of soft tissue using machine learning techniques: validation by Monte Carlo simulation.

SIGNIFICANCE: To achieve early detection of osteoporosis, a simple bone densitometry method using optics was proposed. However, individual differences in soft tissue structure and optical properties can cause errors in quantitative bone densitometry. Therefore, developing optical bone densitometry that is robust to soft tissue variations is important for the early detection of osteoporosis. AIM: The purpose of this study was to develop an optical bone densitometer that is insensitive to soft tissue, using Monte Carlo simulation and machine learning techniques, and to verify its feasibility. APPROACH: We propose a method to measure spatially resolved diffuse light from three directions of the biological tissue model and used machine learning techniques to predict bone density from these data. The three directions are backward, forward, and lateral to the direction of ballistic light irradiation. The method was validated using Monte Carlo simulations using synthetic biological tissue models with 1211 different random structural and optical properties. RESULTS: The results were computed after a 10-fold cross-validation. From the simulated optical data, the machine learning model predicted bone density with a coefficient of determination of 0.760. CONCLUSIONS: The optical bone densitometry method proposed in this study was found to be robust against individual differences in soft tissue.

Random Electromyostimulation Promotes Osteogenesis and the Mechanical Properties of Rat Bones.

Exercise is often recommended as a promising non-pharmacologic countermeasure to prevent osteoporosis. However, elderly osteoporotic patients generally have physical fitness difficulties preventing them from performing effective and sustainable exercise. Electromyostimulation should be one effective modality for non-pharmacological prevention of osteoporosis without any voluntary physical movements. However, successful stimulation patterns remain controversial. As suggested by our previous in vitro studies, randomized timing of stimulation could be a candidate to maximize the osteogenic effect of electromyostimulation. In this study, the effects of random stimulation to the quadriceps on osteogenesis in the femurs were investigated using rats, in comparison with a periodic stimulation pattern. In histomorphometric assessments, both stimulation patterns demonstrated increases in bone formation rate either in cortical bone at the midshaft or in trabecular bone at the femoral neck on the stimulated side. However, maximum load and strain energy to failure were enhanced only by the random stimulation, on either the stimulated or non-stimulated side. It is concluded that randomized muscle stimulation has effective osteogenic capability at the stimulation site, similar to periodic stimulation; however, its effectiveness on mechanical properties is expandable to other non-stimulated sites.

Frequency-Dependence of Mechanically Stimulated Osteoblastic Calcification in Tissue-Engineered Bone In Vitro.

The effect of mechanical stimulation on osteogenesis remains controversial, especially with respect to the loading frequency that maximizes osteogenesis. Mechanical stimulation at an optimized frequency may be beneficial for the bone tissue regeneration to promote osteoblastic calcification. The objective of this study was to investigate the frequency-dependent effect of mechanical loading on osteoblastic calcification in the tissue-engineered bones in vitro. Tissue-engineered bones were constructed by seeding rat osteoblasts into a type I collagen sponge scaffold at a cell density of 1600 or 24,000 cells/mm(3). Sinusoidal compressive deformation at the peak of 0.2% was applied to the tissue-engineered bones at 0.2, 2, 10, 20, 40, and 60 Hz for 3 min/day for 14 consecutive days. Optically-monitored calcium content started to increase on days 5-7 and reached the highest value at 2 Hz on day 14; however, no increase was observed at 0.2 Hz and in the control. Ash content measured after the mechanical stimulation also showed the highest at 2 Hz despite the differences in cell seeding density. It was concluded that mechanical stimulation at 2 Hz showed the highest promotional effect for osteogenesis in vitro among the frequencies selected in this study.

Frequency-dependent enhancement of bone formation in murine tibiae and femora with knee loading.

Knee loading is a relatively new loading modality in which dynamic loads are laterally applied to the knee to induce bone formation in the tibia and the femur. The specific aim of the current study was to evaluate the effects of loading frequencies (in Hz) on bone formation at the site away from the loading site on the knee. The left knee of C57/BL/6 mice was loaded with 0.5 N force at 5, 10, or 15 Hz for 3 min/day for 3 consecutive days, and bone histomorphometry was conducted at the site 75% away from the loading site along the length of tibiae and femora. The results revealed frequency-dependent induction of bone formation, in which the dependence was different in the tibia and the femur. Compared with the sham-loading control, for instance, the cross-sectional cortical area was elevated maximally at 5 Hz in the tibia, whereas the most significant increase was observed at 15 Hz in the femur. Furthermore, mineralizing surface, mineral apposition rate, and bone formation rate were the highest at 5 Hz in the tibia (2.0-, 1.4-, and 2.7 fold, respectively) and 15 Hz in the femur (1.5-, 1.2-, and 1.8 fold, respectively). We observed that the tibia had a lower bone mineral density with more porous microstructures than the femur. Those differences may contribute to the observed differential dependence on loading frequencies.

Knee loading stimulates cortical bone formation in murine femurs.

BACKGROUND: Bone alters its architecture and mass in response to the mechanical environment, and thus varying loading modalities have been examined for studying load-driven bone formation. The current study aimed to evaluate the anabolic effects of knee loading on diaphyseal cortical bone in the femur. METHODS: Using a custom-made piezoelectric loader, 0.5-N loads were laterally applied to the left knee of C57/BL/6 mice at 5, 10, 15, and 20 Hz for 3 minutes per day for 3 consecutive days. Animals were sacrificed for examination 13 days after the last loading. The contralateral femur was used as a non-loading control, and the statistical significance of loading effects was evaluated with p < 0.05. RESULTS: Although diaphyseal strains were measured as small as 12 mustrains, bone histomorphometry clearly demonstrated frequency-dependent enhancement of bone formation. Compared to a non-loading control, bone formation on the periosteal surface was significantly enhanced. The loading at 15 Hz was most effective in elevating the mineralizing surface (1.7 x; p < 0.05), mineral apposition rate (1.4 x; p < 0.001), and bone formation rate (2.4 x; p < 0.01). The loading at 10 Hz elevated the mineralizing surface (1.4 x; p < 0.05), mineral apposition rate (1.3 x; p < 0.01), and bone formation rate (1.8 x; p < 0.05). The cross-sectional cortical area and the cortical thickness in the femoral diaphysis were significantly increased by loading at 10 Hz (both 9%) and 15 Hz (12% and 13%, respectively). CONCLUSION: The results support the anabolic effects of knee loading on diaphyseal cortical bone in the femur with small in situ strain, and they extend our knowledge on the interplay between bone and joints. Strengthening the femur contributes to preventing femoral fractures, and the discovery about the described knee loading might provide a novel strategy to strengthen osteoporotic bones. Further analyses are required to understand the biophysical and molecular mechanism behind knee loading.

Low-amplitude, broad-frequency vibration effects on cortical bone formation in mice.

Mechanical loading of the skeleton is necessary to maintain bone structure and strength. Large amplitude strains associated with vigorous activity typically result in the greatest osteogenic response; however, data suggest that low-amplitude, broad-frequency vibration results in new bone formation and may enhance adaptation through a stochastic resonance (SR) phenomenon. That is, random noise may maximally enhance bone formation to a known osteogenic stimulus. The aims of this study were to (1) assess the ability of different vibration signals to enhance cortical bone formation during short- and long-term loading and (2) determine whether vibration could effect SR in bone. Two studies were completed wherein several osteogenic loading waveforms, with or without an additive low-amplitude, broad-frequency (0-50 Hz) vibration signal, were applied to the mouse ulna in axial compression. In study 1, mice were loaded short-term (30 s/day, 2 days) with either a carrier signal alone (1 or 2 N sine waveform), vibration signal alone [0.1 N or 0.3 N root mean square (RMS)] or combined carrier and vibration signal. In study 2, mice were loaded long-term (30 s/day, 3 days/week, 4 weeks) with a carrier signal alone (static or sine waveform), vibration signal alone (0.02 N, 0.04 N, 0.08 N or 0.25 N RMS) or combined carrier and vibration signal. Sequential calcein bone labels were administered at 2 and 4 days and at 4 and 29 days after the first day of loading in study 1 and 2, respectively; bone formation parameters and changes in geometry were measured. Combined application of the carrier and vibration signals in study 1 resulted in significantly greater bone formation than with either signal alone (P < 0.001); however, this increase was independently explained by increased strain levels associated with additive vibration. When load and strain levels were similar across loading groups in study 2, cortical bone formation and changes in geometry were not significantly altered by vibration. Vibration alone did not result in any new bone formation. Our data suggest that low-amplitude, broad-frequency vibration superimposed onto an osteogenic waveform or vibration alone does not enhance cortical bone adaptation at the frequencies, amplitudes and loading periods tested.

Diaphyseal bone formation in murine tibiae in response to knee loading.

Mechanical stimulation is critical for bone architecture and bone mass. The aim of this study was to examine the effects of mechanical loads applied to the knee. The specific question was whether loads applied to the tibial epiphysis would enhance bone formation in the tibial diaphysis. In C57/BL/6 mice, loads of 0.5 N were applied for 3 min per day for 3 days at 5, 10, or 15 Hz. Bone samples were harvested 13 days after the last loading. The strains were measured 13 +/- 2 microstrains at 5 Hz in the diaphysis. The histomorphometric data in the diaphysis clearly showed enhanced bone formation. First, compared with nonloaded control the cross-sectional cortical area was increased by 11% at 5 Hz and 8% at 10 Hz (both P < 0.05). Second, the cortical thickness was elevated by 12% at 5 Hz (P < 0.01) and 8% at 10 Hz (P < 0.05). Third, mineralizing surface (MS/BS), mineral apposition rate (MAR), and bone formation rate (BFR/BS) were increased at 5 Hz (P < 0.01 for MS/BS; P < 0.001 for MAR and BFR/BS) and at 10 Hz (P < 0.05 for MS/BS; P < 0.01 for MAR and BFR/BS). Bone formation was enhanced more extensively in the medial side than the lateral or the posterior side. The results reveal that knee loading is an effective means to enhance bone formation in the tibial diaphysis in a loading-frequency dependent manner without inducing significant in situ strain at the site of bone formation.

Osteogenic potentials with joint-loading modality.

Osteogenic potentials with a novel joint-loading modality were examined, using mouse ulnae as a model system. Load-induced deformation of rigid bone is known to generate interstitial fluid flow and stimulate osteogenesis. However, in most of the previous studies, loads were applied to cortical bone. In the current study, we addressed the question of whether deformation of the epiphysis underneath the joint would enhance bone formation in the epiphysis and the diaphysis. In order to answer the question, we applied lateral loads to a mouse elbow and conducted a bone histomorphometric analysis, as well as measurements of strains and streaming potentials. Compared to the no-loading control, the histomorphometric results showed that 0.5-N loads, applied to the elbow at 2 Hz for 3 min/day for 3 consecutive days, increased the mineralizing surface (two- to threefold), the rate of mineral apposition (three- to fivefold), and the rate of bone formation (six- to eightfold) in the ulna. Strain measurements indicated that strains of around 30 microstrain, induced with the joint-loading modality, were under the minimum effective strain of around 1000 microstrain, which is considered necessary to achieve strain-driven bone formation. To evaluate the induction of fluid flow with the joint-loading modality, streaming potentials were measured in separate experiments, using mouse femurs ex vivo. We found that the streaming potentials correlated to the magnitude of the load applied to the epiphysis (r(2) = 0.92), as well as the flow speed in the medullary cavity (r(2) = 0.93). Taken together, the findings of the current study support the idea of joint-loading driven osteogenesis, through a mechanism that involves the induction of fluid flow in cortical bone.

Osteoblast responses one hour after load-induced fluid flow in a three-dimensional porous matrix.

When bone is loaded, substrate strain is generated by the external force and this strain induces fluid flow that creates fluid shear stress on bone cells. Our current understanding of load-driven gene regulation of osteoblasts is based primarily on in vitro studies on planer two-dimensional tissue culture substrates. However, differences between a flat layer of cells and cells in 3-dimensional (3D) ECM are being recognized for signal transduction. Proliferation and differentiation of osteoblasts are affected by substrate geometry. Here we developed a novel 3D culture system that would mimic physiologically relevant substrate strain as well as strain-induced fluid flow in a 3D porous collagen matrix. The system allowed us to evaluate the responses of osteoblasts in a 3D stress-strain environment similar to a mechanical field to which bone is exposed. Using MC3T3-E1 osteoblasts grown in the 3D collagen matrix with and without hydroxyapatite deposition, we tested the role of strain and the strain-induced fluid flow in the expression of the load-responsive genes such as c-fos, egr1, cox2, osteopontin, and mmp1B involved in transcriptional regulation, osteogenesis, and rearrangement of ECM. Strain-induced fluid flow was visualized with a microspheres approximately 3 microm in diameter in real time, and three viscoelastic parameters were determined. The results obtained by semi-quantitative PCR, immunoblot assay, enzymatic activity assays for collagenase and gelatinase, and mechanical characterization of collagen matrices supported the dominant role of strain-induced fluid flow in expression of the selected genes one hour after the mechanical treatment.

Bone formation induced by a novel form of mechanical loading on joint tissue.

Because of insufficient mechanical loading, exposure to weightlessness in space flight reduces bone mass. In order to maintain bone mass in a weightless condition, we investigated a novel form of mechanical loading--joint loading. Since some part of gravity-induced loading to our skeletal system is absorbed by viscoelastic deformation of joint tissues, we hypothesized that deformation of joint tissues would generate fluid flow in bone and stimulate bone formation in diaphyseal cortical bone. In order to test the hypothesis, we applied directly oscillatory loading to an elbow joint of mice and conducted bone histomorphometry on the diaphysis of ulnae. Using murine femurs ex vivo, streaming potentials were measured to evaluate a fluid flow induced by joint loading. Bone histomorphometry revealed that compared to no loading control, elbow loading increased mineralizing surface, mineral apposition rate, and bone formation rate 3.2-fold, 3.0-fold, and 7.9-fold, respectively. We demonstrated that joint loading generated a streaming potential in a medullar cavity of femurs. The results support a novel mechanism, in which joint loading stimulates effectively bone formation possibly by generating fluid flow, and suggest that a supportive attachment to joints, driven passively or actively, would be useful to maintain bone mass of astronauts during an exposure to weightlessness.

Effects of high-impact mechanical loading on synovial cell cultures.

Cartilage metabolism in response to mechanical loading is an important subject in sports science and medicine. In animal studies high-impact exercise is known to stimulate bone adaptation and increase bone mass. However, mechanical impacts potentially induce tissue swelling and occasionally degradation of connective tissues in synovium and articular cartilage. These detrimental outcomes should be properly evaluated clinically and biochemically. Using two synovial cell cultures derived from normal and rheumatic tissues, we examined the biochemical effects of impulsive mechanical loads on expression and activities of influential proteolytic enzymes in joints, matrix metalloproteinases (MMPs), and their natural inhibitors, tissue inhibitors of metalloproteinases (TIMPs). The molecular analysis demonstrates that an impact factor (Im ), the ratio of the maximum force to weight, served as a good indicator for assessment of the inflammatory responses. The results showed that high impact above Im = 40 to 80 elevated not only expression but also enzymatic activities of MMPs. Key PointsHigh-impact loading elevated expression and activities of MMPs in synovial cell cultures.The impact factor was used to define in vitro intensity of high-impact loading.

Stochastic resonance in osteogenic response to mechanical loading.

Stochastic resonance, in which noise enhances the response of a nonlinear system to a weak signal, has been observed in various biological sensory systems. We speculated that bone formation in response to mechanical loading could be enhanced by adding noise (vibration) to a standard exercise regimen. To test this hypothesis, three different loading regimens were applied to the ulnae of mice: (1) high amplitude, low frequency sinusoidal loading at 2 Hz with an amplitude of 3 N to simulate exercise; (2) low amplitude, broad frequency vibration with frequency components 0-50 Hz and 0.3 N of mean amplitude; (3) the sinusoidal wave combined with vibration (S+V) to invoke stochastic resonance. The simulated exercise regimen induced new bone formation on the periosteal surface of the ulna, however the addition of vibration noise with exercise enhanced the osteogenic response by almost 4-fold. Vibration by itself had no effect on bone formation. It was concluded that adding low magnitude vibration greatly enhanced bone formation in response to loading, suggesting a contribution of stochastic resonance in the osteogenic response.

Effects of broad frequency vibration on cultured osteoblasts.

Bone is subjected in vivo to both high amplitude, low frequency strain, incurred by locomotion, and to low amplitude, broad frequency strain. The biological effects of low amplitude, broad frequency strain are poorly understood. To evaluate the effects of low amplitude strains ranging in frequency from 0 to 50 Hz on osteoblastic function, we seeded MC3T3-E1 cells into collagen gels and applied the following loading protocols for 3 min per day for either 3 or 7 days: (1) sinusoidal strain at 3 Hz, with 0-3000 microstrain peak-to-peak followed by 0.33 s resting time, (2) "broad frequency vibration" of low amplitude strain (standard deviation of 300 microstrain) including frequency components from 0 to 50 Hz, and (3) sinusoidal strain combined with broad frequency vibration (S + V). The cells were harvested on day 4 or 8. We found that the S + V stimulation significantly repressed cell proliferation by day 8. Osteocalcin mRNA was up-regulated 2.6-fold after 7 days of S + V stimulation, and MMP-9 mRNA was elevated 1.3-fold after 3 days of vibration alone. Sinusoidal stimulation alone did not affect the cell responses. No differences due to loading were observed in alkaline phosphatase activity and in mRNA levels of type I collagen, osteopontin, connexin 43, MMPs-1A, -3, -13. These results suggest that osteoblasts are more sensitive to low amplitude, broad frequency strain, and this kind of strain could sensitize osteoblasts to high amplitude, low frequency strain. This suggestion implies a potential contribution of stochastic resonance to the mechanical sensitivity of osteoblasts.