Open Access
Add to BibSonomy
Abstract
Osteoporosis usually alters the chemical composition and physical microstructure of bone. Currently, most clinical techniques for bone assessment are focused on the either bone microstructure or bone mineral density (BMD). In this study, a novel multi-wavelength photoacoustic time-frequency spectral analysis (MWPA-TFSA) method was introduced based on the optical absorption spectra and photoacoustic effects of biological macromolecules, which evaluates changes in bone chemical composition and microstructure. The results demonstrated that osteoporotic bones had decreased BMD, more lipids, and wider trabecular separation filled with larger marrow clusters, which were consistent with multiple gold-standard results, suggesting that the MWPA-TFSA method has the potential to provide a thorough bone physico-chemical information evaluation noninvasively and nonradiatively.
© 2022 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Bone disease is a chronic condition that causes changes in the physical structure and chemical composition of bones. Osteoporosis is the most common bone disease globally, causing morbidity and mortality and placing significant strain on public health resources [1,2]. X-ray and ultrasound (US) are the most regularly used clinical tools for detecting bone quality [3]. Quantitative computed tomography and dual energy X-ray absorptiometry (DEXA) can be employed to assess bone mineral density (BMD), which is a crucial predictor of fracture risk [4]. However, these X-ray technologies require ionizing radiation and cannot provide bone microarchitecture information. Quantitative US (QUS) techniques are practical screening tools for detecting bone quality because they are faster than X-ray procedures and do not require ionizing radiation, although they are less accurate. Furthermore, they can provide information regarding the BMD, bone structure, and elastic characteristics [5,6]. However, none of the currently available clinical techniques can assess the changes in both the physical structure and chemical composition of bones in a non-invasive and non-radiative manner.
The photoacoustic (PA) technique is a novel biomedical detection approach that uses a pulsed or modulated optical wave to illuminate biological optical absorbers and generate US waves [7]. This technique can provide information regarding the chemical components of biological tissues, based on the optical absorption spectra of specific biomacromolecules [8,9]. Moreover, information pertaining to the histological microstructures of different chemical components can be characterized based on the acoustic frequency spectrum of the PA absorption sources [10]. Therefore, the PA technique shows potential for use in tracking biological lesions and has become a key research focus in biomedical detection technology.
The PA approach has been used for imaging and disease detection in soft biological tissues for nearly two decades. For instance, Wang et al. [11] achieved in vivo PA imaging of rat intracranial microvessels. Wang et al. [12] and Yang et al. [13] used the distinctive optical absorption of melanin to achieve melanoma imaging. Cheng et al. [14] explained the changes in the physico-chemical properties of numerous biological macromolecules based on multi-wavelength PA spectroscopy (MWPAS) combined with machine learning and realized the distinction between benign and malignant prostate cancers. These studies confirm the ability of MWPAS to determine the physical and chemical properties of biological tissues.
Cancellous bone is a porous material, with bone marrow clusters filling the pores of the trabecular meshwork [15]. Despite the porosity of bone tissue, which causes substantial scattering and results in the absorption of light and sound, several studies have demonstrated that PA bone detection is highly feasible. Feng et al. [16] demonstrated an optical and US penetration depth of 22 mm in bone tissue and performed three-dimensional reconstruction of bone microstructures. Mandelis et al. [17] proposed US and PA methods to detect the content changes of bone mineral and bone collagen. Steinberg et al. [18] determined the blood/fat ratio via marrow measurement in vivo, using a dual-wavelength PA system. Furthermore, Feng et al. [19,20] proposed an in vivo multi-wavelength PA (MWPA) method to obtain bone health information from calcaneus bone. However, previous PA measurement techniques have mainly focused on either assessing the bone composition or characterizing the bone microstructure of the trabecular bone. In our previous studies [21,22], we developed a PA time–frequency spectral analysis (TFSA) method to evaluate the trabecular thickness (Tb. Th) and BMD by converting the PA signals of trabecular bone into time and frequency dimensions based on wavelet transform, which provides the spatial distribution of trabecular bone. However, osteoporosis is accompanied by a decrease in BMD and a thinning in trabecular bone; occasionally, there is an increase in trabecular separation, producing larger marrow clusters [23]. The chemical composition also changes in osteoporosis, including decreases in bone mineral and collagen content and an increase in lipid content. Therefore, the microstructure and content of marrow clusters should also be evaluated for comprehensive bone assessment.
This paper proposes a new MWPA-TFSA method for detecting changes in the chemical composition and physical structure of bones simultaneously, based on our previous studies. Firstly, we investigated the changes in the chemical composition, such as those in the bone minerals and lipids, by analyzing the PA signals at different wavelengths corresponding to their optical absorption peaks. In addition, we abstracted the slope of the PA energy distribution over time for the BMD evaluation. Moreover, we employed the time–frequency spectrum at the typical optical absorption wavelength of bone marrow to determine the size of the bone marrow clusters, which is negatively correlated with the trabecular thickness.
2. Materials and methods
2.1 Bone specimen collection and processing
In this study, two groups of well-established rabbit models were used. After a week of acclimatization, 16 five-month-old female New Zealand white rabbits were randomly selected for ovariectomy to form the osteoporotic group (n = 8) and sham operation to form the control group (n = 8) [24]. All rabbits were kept in 50 × 42 × 40 cm3 individual cages, and they were fed standard commercial rabbit fodder. Five months after surgery, all the rabbits were euthanized, and the epiphysis of the long bone of the left leg was dissected and sawn into sheets with thicknesses of 1.5 mm for PA testing. The cortical bone and cartilage were removed to employ only the trabecular portion in the tests, as shown in Fig. 1(B). All experiments and procedures were approved by the ethics committee of Nanjing University of Science and Technology (No. 202100129) and were conducted in accordance with the ethical standards of the relevant institutional and/or national research committees.
Fig. 1. (A) Experimental setup for PA measurements. (B) Photograph of a bone sample. (C) Laser energy density at different wavelengths.
2.2 Gold-standard examination
2.2.1 DEXA imaging for area BMD analysis
Bone specimens from the left side of each rabbit in the two groups were scanned using DEXA (InAlyzer, Medikors, Inc., Gyeonggi-do, Korea) for BMD measurements. Figure 2(A) shows the regions of interest (ROIs) in the DEXA images of the bone samples, marked by yellow dotted boxes. Figure 2(B) presents the statistical analysis results for the area BMD of the two groups. There exists a strong correlation between the bone mineral content (BMC) and area BMD, where the BMC is equal to the area BMD times the area [25]. For bone tissues with approximately identical surface areas, the higher the BMC, the higher the BMD. The BMD is 8.8% lower in the osteoporotic group, indicating the reduced BMC in the osteoporotic bone samples.
Fig. 2. Gold-standard examination. (A) DEXA images of bones from the control and osteoporotic groups. The ROIs are marked by yellow dotted boxes. (B) Statistical analysis results for BMD (*p < 0.05). (C) MRI images of bones from the control and osteoporotic groups. (D) Statistical analysis results of lipid fraction (*p < 0.05). (E) Micro-CT images of bones from the control and osteoporotic groups. (F) Statistical analysis results for BV/BS (*p < 0.05).
2.2.2 MRI for lipid content analysis
Each bone specimen was scanned using a 9.4 T magnetic resonance imaging (MRI) system (Bruker Biospec 94/20 USR) to determine the bone marrow lipid contents of the control and osteoporosis groups. For the acquisition of the lipid and water separation images, measurements were performed with the MRI system using the three-point Dixon technique, with water–lipid shifts of (0, π, 2π) [26]. The measured data were processed via iterative water and lipid decomposition with echo asymmetry and a least-squares estimation algorithm. The resulting water and lipid images were used to develop the lipid fraction map. The primary parameters of the measurement were as follows: repetition time = 2500 ms, echo time = 24 ms, echo spacing = 8.0 ms, slice thickness = 2.0 mm, resolution = 0.313 ${\times} $ 0.313 mm, matrix = 128 ${\times} $ 128, field of view = 40 ${\times} $ 40 mm, and bandwidth = 100 kHz. The lipid fraction maps of a normal bone from the control group and an osteoporotic bone from the osteoporotic group are presented in Fig. 2(C). The quantified MRI results in Fig. 2(D) verify that the lipid fraction in the osteoporotic group is 10.8% higher than that in the control group, with significant differences between the two groups.
2.2.3 Micro-CT imaging for bone microstructure analysis
All the bone specimens were scanned via micro-CT (SCANCO, vivaCT 80), and 3D images of each bone specimen were reconstructed using a Hiscan Analyzer to characterize the microstructure of the trabecular bone and trabecular separation, which is strongly correlated with the size of bone marrow clusters. The reconstructed 3D images of a normal bone from the control group and an osteoporotic bone from the osteoporotic group are depicted in Fig. 2(E). The ratio of bone volume to bone surface (BV/BS) was calculated to characterize the trabecular separation. The statistical results in Fig. 2(F) show that the BV/BS in the osteoporotic group is 7.3% lower than that in the control group, with significant differences between the two groups. These differences correspond to a wider mean trabecular spacing of bone (pores filled with bone marrow clusters) in the osteoporotic group [27].
2.3 PA experimental measurement setup
The experimental setup is shown in Fig. 1(A). Laser pulses with wavelengths of 700, 760, and 930 nm were produced using a tunable optical parametric oscillator laser (Phocus Mobile, OPOTEK, Carlsbad, CA) with a width of approximately 2–5 ns. The laser power meter was used to record the energy fluctuation range and mean value of laser energy at different wavelengths ${L_\lambda }$ at the same position as the bone samples before PA measurement. The laser energy density ${D_\lambda }$ was calculated as ${D_\lambda } = \frac{{4{L_\lambda }}}{{\pi {d^2}}}$ to ensure that it is less than the ANSI laser safety standard, as shown in Fig. 1(C). Meanwhile, the blackbody was used to record and calibrate the laser energy during PA measurement process by dividing the laser beam into two parts to illuminate the bone samples and blackbody simultaneously since the laser energy varies with the wavelength and has a slight fluctuation over time. As shown in Fig. 1(B), the laser beam was irradiated uniformly onto the surface of each bone specimen; the diameter of the laser spot d was approximately 10 mm. A needle hydrophone with a bandwidth of 0–20 MHz (HNC1500, ONDA Corp., Sunnyvale, CA) was used to receive the PA signals generated from the bone specimens, whereas a 1 MHz focused transducer (Immersion Transducers, Olympus Corp., Tokyo, Japan) was used to receive the PA signals from the blackbodies. All the bone PA signals were amplified by 25 dB using an amplifier (5072PR, Olympus Corp., Tokyo, Japan) and recorded using a digital oscilloscope (HDO6000, Teledyne Lecroy, USA). The sampling rate was 250 MHz. Each PA signal was received an average of 50 times to improve the signal-to-noise ratio.
2.4 MWPA signal data processing
The MWPA signals were analyzed to obtain the chemical information. First, as shown in Fig. 3(A), as the average velocity of cancellous bone is approximately 2000 m/s [28], the valid PA signal length is approximately 5 µs with a laser spot diameter of 10 mm. All the PA signals generated by the bone specimens were calibrated using the peak-to-peak values of the PA signals from the blackbody to obtain the normalized amplitudes at different wavelengths. Second, as shown in Fig. 3(B), to obtain the PA time–frequency spectrum (TFS), all the PA signals were analyzed using the continuous wavelet transform (CWT) technique. The PA-TFS provides the acoustic frequency spectrum distribution at each time and the PA energy attenuation over time. As the pore size of bone is distributed, the PA signal is wide-band. Third, the energy of the PA signal at a wavelength of $\lambda $ nm (${E_\lambda }$) was calculated using the following equation:
Fig. 3. Data processing. (A) Typical normalized PA signal from a bone sample. The valid PA signal with a duration of 5 µs is marked with the red box. (B) Example of PA-TFS calculated using CWT of bone PA signal. (C) Typical power spectrum density of a bone PA signal. (D) Ratio of the PA signal energy at 760 nm to that at 930 nm (E760/E930), and ratio of the PA signal energy at 700 nm to that at 930 nm (E700/E930), calculated from PA-TFSs at three wavelengths for chemical component evaluation. (E) Example of normalized energy ($E(t )$) at different times. The slope was abstracted for BMD evaluation. (F) PWMF at different times in PA-TFS. The PWMF within 1 µs is denoted by the red dotted box for bone microstructure evaluation.
In the MWPA measurements conducted in this study, E700, E760, and E930 were selected to analyze the content changes in hydroxyapatite, Hb, and lipid. To determine the chemical component changes in the ratio of Hb to lipid, and the ratio of mineral and Hb to lipid, the ratio of the PA signal energy at 760 nm to that at 930 nm (E760/E930), and the ratio of the PA signal energy at 700 nm to that at 930 nm (E700/E930) were calculated, as shown in Fig. 3(D).
The attenuation of the transmitted US energy in bones depends on the BMD; the lower the BMD, the weaker the ultrasonic propagation attenuation [33]. To evaluate the BMD [34], the normalized PA energy at each moment ($E(t )$), reflecting the US propagation attenuation, was calculated using Eq. (2), as shown in Fig. 3(E):
Because the PA-TFS contains rich information regarding the microstructures of the PA absorption sources, it can be used to characterize the microstructure information of bone tissues. However, the PA-TFS is affected by not only the microstructure, but also the US propagation attenuation in bone tissues. To reduce the effects of US propagation attenuation in the characterization of bone microstructures, the PA-TFS was analyzed only during the first 1 µs, which contains rich microstructure information while the US propagation attenuation is less than 3 dB and can be neglected [35,36], as shown in the red dotted boxes in Fig. 3(B). To determine the dominant frequency at different times t, the power-weighted mean frequency (PWMF) at each instant in the PA-TFS was extracted, as shown in Fig. 3(F). The PWMF was calculated using the following equation:
In Eq. (3), the lower-limit frequency of the interval ${f_1}$ was 0.3 MHz to avoid strong low-frequency noise. The higher-limit frequency of the interval ${f_2}$ was 6 MHz, determined by subtracting 20 dB from the PA power spectrum, as shown in Fig. 3(C). In cancellous bone, the pores between bone trabeculae are filled with bone marrow clusters, where lipids are a major component [37,38]. The PWMF at a wavelength of 930 nm, corresponding to the optical peak of lipids, was further studied. As depicted in the red dotted box in Fig. 3(F), the Max-PWMF and Mean-PWMF within 1 µs at the wavelength of 930 nm, with acoustic attenuation that can be neglected, were quantified to characterize the microstructures of the bone marrow clusters.
An unpaired two-tailed independent sample t-test was conducted using GraphPad Prism 9.0 to assess whether statistically significant differences in the PA parameters were present between the osteoporotic and control groups.
3. Results
3.1 Bone mineral and lipid content characterization
Cancellous bone contains both organic and inorganic components. When osteoporosis develops, there are changes in the BMD and trabecular meshwork determined by the inorganic mineral and the composition of organic chemicals (i.e., lipids and Hb) in the bone marrow. MRI-based measurements have shown that osteoporosis is related to bone marrow lipids, as they significantly affect bone mineralization; additionally, a negative correlation exists between the bone marrow fat fraction and BMD [39,40]. Studies also show that an increase in marrow fat content caused a decrease in marrow perfusion in the osteoporotic bones, indicating less blood perfusion, that is, less Hb in bone marrow [37,41]. Thus, osteoporotic bones are typically higher in lipids and lower in bone mineral and Hb than normal bones.
In the MWPA measurements conducted in this study, different wavelengths were used to target different chemical components in the bone tissue. The results of PA-TFS at wavelengths of 700, 760, and 930 nm for the osteoporotic and control groups are presented in Figs. 5(A) and (B), respectively. The MWPA energy ratio of the normal and osteoporotic group is summarized in Table 1. As shown in Fig. 5(C), the energy ratio of E760/E930 is lower in the osteoporotic group than in the control group (p = 0.047), indicating a lower content ratio of Hb to lipid in the osteoporotic group. Moreover, the energy ratio of E700/E930 is lower in the osteoporotic group (p = 0.029), indicating a lower content ratio of bone mineral and Hb to lipid in the osteoporotic group than in the control group. The energy ratios indicate that more lipids, less bone mineral and Hb were present in the osteoporotic group than in the control group. These results were verified by the DEXA and MRI results.
Fig. 5. Ex vivo estimation of bone mineral and lipid content. (A) Time–frequency spectra of bone PA signals in the control group at wavelengths of 700, 760, and 930 nm. (B) Time–frequency spectra of bone PA signals in the osteoporotic group at wavelengths of 700, 760, and 930 nm. (C) Statistical analysis results for the ratio of the PA signal energy at 760 nm to that at 930 nm (E760/E930) and the ratio of the PA signal energy at 700 nm to that at 930 nm (E700/E930) (*p < 0.05).
Table 1. MWPA-TFSA Parameters of Normal and Osteoporotic Group
To evaluate the BMD information in bone tissue further, the slope of the PA energy distribution was determined and compared between the control and osteoporosis group. Figures 6(A), 6(B), and 6(C) present the PA-TFSs of a normal and an osteoporotic bone at wavelengths of 700, 760, and 930 nm, respectively. Moreover, Fig. 6(D) depicts the representative normalized energy curves of the two bone types. As shown in Fig. 6(E), the slope of the normalized energy was studied for BMD evaluation by analyzing the PA energy propagation attenuation. The MWPA energy distribution slopes of the normal and osteoporotic groups are summarized in Table 1. The results demonstrate that the normalized energy curve is relatively flat, and the quantified slope parameter is higher for the osteoporotic bone samples than for the normal bones (p = 0.001 at 700 nm, p = 0.008 at 760 nm, and p = 0.05 at 930 nm). This is because the osteoporotic group with low BMD has less bone mineral and larger bone marrow clusters. In general, the lower the BMD, the lower the propagating attenuation [33,42] and the higher the slope, and vice versa. These results also were verified by the DEXA results.
Fig. 6. Ex vivo estimation of BMC. (A) PA-TFSs of normal bone and osteoporotic bone at a wavelength of 700 nm. (B) PA-TFSs of normal bone and osteoporotic bone at a wavelength of 760 nm. (C) PA-TFSs of normal bone and osteoporotic bone at a wavelength of 930 nm. (D) Representative normalized energy of different bones from the control and osteoporotic groups. (E) Slope of normalized energy of the normal and osteoporotic bones (*p < 0.05, **p < 0.01).
3.2 Bone microstructure characterization
To characterize the microstructures of the bone marrow clusters, the Max-PWMF and Mean-PWMF within 1 µs obtained at a wavelength of 930 nm were quantified and compared between the two groups. The PA-TFSs of normal and osteoporotic bone samples are shown in Figs. 7(A) and (B), respectively, at a wavelength of 930 nm targeted at lipid clusters in the bone marrow [29,30]. The figures demonstrate that the acoustic frequency components in the normal bone are higher than those in the osteoporotic bone. This finding implies that the normal bone samples have smaller pores filled with bone marrow clusters, which is consistent with the results of previous studies [43]. Furthermore, the PWMFs of different bone samples were quantified using Eq. (3) to elucidate the changes in the acoustic frequency components. In particular, the quantitative parameters, Max-PWMF and Mean-PWMF, were obtained. The PWMF at 930 nm of the normal and osteoporotic groups is summarized in Table 1. Figures 7(C) and (D) demonstrate that at a wavelength of 930 nm, the osteoporotic group has a lower Max-PWMF (p = 0.013) and Mean-PWMF (p = 0.009) than the control group, with significant differences between the two groups.
Fig. 7. Ex vivo estimation of microstructure. PA-TFSs of (A) a normal bone sample and (B) an osteoporotic bone sample at a wavelength of 930 nm. The ROIs are denoted by white dotted boxes. (C) Representative PWMFs of normal and osteoporotic bone samples. The ROIs are marked with green dotted boxes. (D) Max-PWMFs and Mean-PWMFs of the two groups at a wavelength of 930 nm. The PWMF within 1 µs denoted by the green dotted box was studied (*p < 0.05, **p < 0.01).
Because the acoustic frequency spectrum of PA signals is dependent on the size of the PA absorption source (the smaller the size, the higher the frequency), the acoustic frequency spectrum of the lipid clusters in bone marrow can reflect the pore size of the bone [44]. The micro-CT results confirmed that in the osteoporotic group, the degeneration of the bone microstructure resulted in a wide trabecular separation filled with large bone marrow clusters; hence, the Max-PWMF and Mean-PWMF are lower than those in the control group.
4. Discussion
Herein, we proposed an MWPA-TFSA method that can simultaneously detect changes in the bone minerals, lipids, and bone marrow cluster size for comprehensive bone assessment. The MWPA-TFS contains information related to both the chemical and physical information of the bone tissues. Firstly, the chemical components of the bone tissue can be assessed by analyzing the multi-wavelength PA signals. For example, E760/E930 and E700/E930 can be used to detect differences in the content ratios of Hb to lipid and of bone mineral and Hb to lipid, respectively, between the two groups (control and osteoporotic groups). Secondly, the slope of the normalized energy can reflect the PA energy attenuation along the propagation distance in a wide ultrasonic frequency range. For example, the higher the BMD, the lower the slope of the PA energy over time due to the high-intensity US backscatter attenuation produced by the presence of more scatterers and the high attenuation caused by differences in the acoustic impedance [33,42]. Thirdly, as the acoustic frequency spectrum of PA-TFS within 1 µs neglects US propagation attenuation, the acoustic frequency spectrum is affected only by the PA absorption source. The Max-PWMF and Mean-PWMF within 1 µs at an optical wavelength of 930 nm reflected the main PA frequency components of bone marrow clusters, primarily filled with lipids, and correlated with the size of the bone marrow clusters. These three PA parameters were consistent with the gold-standard DEXA, MRI, and micro-CT results. These findings validate the potential applicability of MWPA-TFSA for ex vivo bone assessment.
The proposed MWPA-TFSA method has many advantages. Firstly, unlike in the previous study of bone composition assessment based on the spectral unmixing method [20], the MWPA-TFSA approach does not need the accurate absorption spectrum of each chemical component and thus will not introduce the decoupling error. Furthermore, this method only requires the measurement of several wavelengths of PA signals and thus has the advantages of low equipment requirements and being more economical. Secondly, for the bone microstructure analysis, the PA power spectral analysis method based on the Fourier transform spectral analysis (FTSA) of the PA signal is usually implemented, which targets a global acoustic frequency spectrum [20,45]. As the PA power spectrum is associated with not only the initial PA frequency spectrum of bone tissue, but also the frequency-related acoustic attenuation in the bone, the PA power spectrum is affected by both the bone microstructure and BMD. Therefore, the properties of the bone microstructure and BMD of the non-organic bone matrix are difficult to separate by using the method based on FTSA. The newly proposed PA-TFSA approach enables characterization of the bone microstructure and BMD by analyzing the acoustic frequency spectrum of trabecular bone and frequency-related acoustic propagation attenuation of bone trabecular meshwork, respectively [22]. Furthermore, based on the MWPA-TFSA method, not only the microstructure and content of the non-organic mineral (trabecular bone matrix), but also the organic tissue in the bone can be analyzed. For example, cancellous bone is a fluid–solid coupled medium, where osteoporosis is not only accompanied by thinning of trabecular bone, but also occasionally accompanied by an increase in trabecular separation, producing larger marrow clusters. The microstructure and content of marrow clusters can be evaluated by MWPA-TFSA at its characteristic absorption wavelength.
However, some technical issues need to be resolved before the MWPA-TFSA method can be translated to the clinic. Considering that the cancellous bone is covered by cortical bone and soft tissue, two issues must be resolved before in vivo clinical study: 1) how to extract the PA signal of cancellous bone from the whole signal, and 2) how to compensate for the influence of cortical bone and soft tissue on PA signal. For the first issue, our previous work demonstrated that the PA signal from cancellous bone tissue can be distinguished from that of cortical bone and soft tissue, and that the thickness of cortical bone and soft tissue can be determined, based on the analysis of PA signal amplitude and time duration [19,46,47]. For the second issue, because the optical effective attenuation and the frequency-related acoustic attenuation of cortical bone and soft tissue follow the law of exponential decay, the PA signal of cancellous bone can be compensated according to the information of several parameters, including optical attenuation coefficient, the acoustic attenuation coefficient, the thickness of cortical bone, and the thickness of soft tissue [32,48–50]. In addition, in order to further improve the accuracy in clinical study, the US imaging algorithm can be used to correct the thickness calculation errors caused by the acoustic impedance mismatching between different layers when the PA wave is incident at an oblique angle and the anisotropy of anisotropy of the ultrasonic wave-speed in cortical bone [51,52].
In addition to bone minerals and lipids, changes in collagen fibers significantly affect bone health. Previous studies have shown that collagen fibers affect calcium salt deposits and bone strength [53]. The PA signals of bone samples in the second near-infrared window (NIR-II) must be assessed further for bone hardness and biomechanics evaluation, as the specific wavelength of collagen is in the NIR-II window [54]. The cancellous bone is anisotropic, and PA signals generated by different positions differ [55,56]. The average value of the acoustic frequency spectrum of PA signals is typically used to evaluate bone microstructure; thus, the spatial distribution of an anisotropic bone microstructure has not been considered. Accurate bone assessment can be achieved through quantitative analysis of the cancellous bone at multiple locations. In contrast, the light absorption characteristics of trabeculae and bone marrow are different, as they have different specific absorption wavelengths. The size characteristics of trabeculae and bone marrow also vary, leading to different acoustic frequency spectrum distributions. As shown in Figs. 6(A) and (B), compared with the acoustic frequency spectrum at a wavelength of 700 nm, corresponding to the specific absorption peak of thinner trabecular bone, more low-frequency components are present at 760 nm and 930 nm, corresponding to the specific absorption peaks of large bone marrow clusters filled with Hb and lipids. The lower-frequency overlap in the acoustic frequency spectrum of the trabecular bone and bone marrow clusters can be eliminated based on the contents of the two chemical components calculated by employing multi-wavelength PA energies. Subsequently, by applying time inversion to high and low frequencies in PA-TFSs at multiple wavelengths, the relative distributions of thinner bone trabeculae and larger bone marrow can be obtained. In the future, through data mining on MWPA-TFS and machine learning on MWPA-TF spectral parameters in NIR-I and NIR-II, information regarding BMD, chemical component contents, and bone microstructure can be abstracted for more comprehensive bone quality evaluation.
Compared with the DEXA, MRI, and QUS methods, MWPA-TFSA can provide comprehensive bone assessment results. The gold-standard DEXA imaging enables the BMD assessment at the spine and femur, which are the most relevant sites. However, it cannot provide information regarding the bone microstructure, which is also a crucial factor affecting bone strength [57]. MRI represents an ideal method for assessing the properties of bone marrow fat; however, it cannot provide information about BMD, which is currently considered the best predictor of osteoporotic fractures [58]. QUS measurement can provide information about physical bone properties, such as BMD, bone microstructure, and elastic properties [59,60]. Nevertheless, it cannot provide chemical composition information. Considering that the PA method can provide information about not only the tissue microstructure, but also the biological macromolecules in biological tissues, it has great potential in bone disease diagnosis and treatment monitoring.
5. Conclusion
In conclusion, the MWPA-TFSA method proposed in this study can be employed to quantify relevant bone quality parameters with a reduced number of measurements via PA characterization. It can provide comprehensive information about the contents of both organic and inorganic chemical components, including lipids and bone minerals, through analysis of the PA energies at different wavelengths and the slope of the PA energy along the propagation distance, based on differences in the optical absorption spectra of different biomacromolecules and acoustic power spectra of biomolecular clusters in bone tissues. This method can also provide information regarding the bone microstructure (i.e., bone marrow cluster size) by analyzing the acoustic attenuation neglected PWMF of bone marrow clusters. Thus, MWPA-TFSA has considerable potential for enabling comprehensive clinical evaluation of bone quality in a noninvasive and nonradiative manner. It can be further applied to the monitoring of bone disease progression related to bone chemical composition and microstructure changes.
Funding
National Natural Science Foundation of China (11827808, 12034015); Program of Shanghai Academic Research Leader (21XD1403600); China Postdoctoral Science Foundation (2019M65156).
Acknowledgments
We would like to thank Editage (www.editage.cn) for English language editing.
Disclosures
The authors declare that there are no conflicts of interest related to this article.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. J. B. Richards, F. Rivadeneira, M. Inouye, T. M. Pastinen, N. Soranzo, S. G. Wilson, T. Andrew, M. Falchi, R. Gwilliam, K. R. Ahmadi, A. M. Valdes, P. Arp, P. Whittaker, D. J. Verlaan, M. Jhamai, V. Kumanduri, M. Moorhouse, J. B. van Meurs, A. Hofman, H. A. P. Pols, D. Hart, G. Zhai, B. S. Kato, B. H. Mullin, F. Zhang, P. Deloukas, A. G. Uitterlinden, and T. D. Spector, “Bone mineral density, osteoporosis, and osteoporotic fractures: a genome-wide association study,” Lancet 371(9623), 1505–1512 (2008). [CrossRef]
2. S. R. Cummings and L. J. Melton, “Epidemiology and outcomes of osteoporotic fractures,” Lancet 359(9319), 1761–1767 (2002). [CrossRef]
3. P. Pisani, M. D. Renna, F. Conversano, E. Casciaro, M. Muratore, E. Quarta, M. D. Paola, and S. Casciaro, “Screening and early diagnosis of osteoporosis through X-ray and ultrasound based techniques,” WJR 5(11), 398–410 (2013). [CrossRef]
4. H. K. Genant, K. Engelke, T. Fuerst, C. C. Glüer, S. Grampp, S. T. Harris, M. Jergas, T. Lang, Y. Lu, S. Majumdar, A. Mathur, and M. Takada, “Noninvasive assessment of bone mineral and structure: State of the art,” J. Bone Miner. Res. 11(6), 707–730 (2009). [CrossRef]
5. C. C. Glüer C, Y. Wu, M. Jergas, S. A. Goldstein, and H. K. Genant, “Three quantitative ultrasound parameters reflect bone structure,” Calcif. Tissue Int. 55(1), 46–52 (1994). [CrossRef]
6. K. Y. Chin and S. Ima-Nirwana, “Calcaneal quantitative ultrasound as a determinant of bone health status: What properties of bone does it reflect?” Int. J. Med. Sci. 10(12), 1778–1783 (2013). [CrossRef]
7. C. Li and L. V. Wang, “Photoacoustic tomography and sensing in biomedicine,” Phys. Med. Biol. 54(19), R59–R97 (2009). [CrossRef]
8. L. V. Wang and J. Yao, “A practical guide to photoacoustic tomography in the life sciences,” Nat. Methods 13(8), 627–638 (2016). [CrossRef]
9. H. W. Wang, N. Chai, P. Wang, S. Hu, W. Dou, D. Umulis, L. V. Wang, M. Sturek, R. Lucht, and J. X. Cheng, “Label-free bond-selective imaging by listening to vibrationally excited molecules,” Phys. Rev. Lett. 106(23), 238106 (2011). [CrossRef]
10. G. Xu, J. B. Fowlkes, C. Tao, X. Liu, and X. Wang, “Photoacoustic spectrum analysis for microstructure characterization in biological tissue: analytical model,” Ultrasound Med. Biol. 41(5), 1473–1480 (2015). [CrossRef]
11. X. Wang, Y. Pang, G. Ku, X. Xie, G. Stoica, and L. V. Wang, “Noninvasive laser-induced photoacoustic tomography for structural and functional in vivo imaging of the brain,” Nat. Biotechnol. 21(7), 803–806 (2003). [CrossRef]
12. J.-T. Oh, M.-L. Li, H. F. Zhang, K. Maslov, G. Stoica, and L. V. Wang, “Three-dimensional imaging of skin melanoma in vivo by dual-wavelength photoacoustic microscopy,” J. Biomed. Opt. 11(3), 034032 (2006). [CrossRef]
13. J. Staley, P. Grogan, A. K. Samadi, H. Cui, M. S. Cohen, and X. Yang, “Growth of melanoma brain tumors monitored by photoacoustic microscopy,” J. Biomed. Opt. 15(4), 040510 (2010). [CrossRef]
14. Y. Chen, C. Xu, Z. Zhang, A. Zhu, X. Xu, J. Pan, Y. Liu, D. Wu, S. Huang, and Q. Cheng, “Prostate cancer identification via photoacoustic spectroscopy and machine learning,” Photoacoustics 23, 100280 (2021). [CrossRef]
15. Z. E. A. Fellah, J. Y. Chapelon, S. Berger, W. Lauriks, and C. Depollier, “Ultrasonic wave propagation in human cancellous bone: Application of Biot theory,” J. Acoust. Soc. Am. 116(1), 61–73 (2004). [CrossRef]
16. T. Feng, K. Kozloff, M. Cao, Q. Cheng, J. Yuan, and X. Wang, “Study of Photoacoustic Measurement of Bone Health Based on Clinically Relevant Models,”, in B. Choi, N. Kollias, H. Zeng, H. W. Kang, B. J. F. Wong, J. F. Ilgner, G. J. Tearney, K. W. Gregory, L. Marcu, M. C. Skala, P. J. Campagnola, A. Mandelis, and M. D. Morris, eds. (San Francisco, CA, 2016), p. 96894f.
17. B. Lashkari, L. Yang, and A. Mandelis, “The application of backscattered ultrasound and photoacoustic signals for assessment of bone collagen and mineral contents,” Quant. Imaging Med. Surg. 5(1), 46–56 (2015). [CrossRef]
18. I. Steinberg, L. Shiloh, I. Gannot, and A. Eyal, “First-in-human study of bone pathologies using low-cost and compact dual-wavelength photoacoustic system,” IEEE J. Select. Topics Quantum Electron. 25(1), 1–8 (2019). [CrossRef]
19. T. Feng, Y. Zhu, R. Morris, K. M. Kozloff, and X. Wang, “Functional photoacoustic and ultrasonic assessment of osteoporosis: a clinical feasibility study,” BME frontiers 2020, 1–15 (2020). [CrossRef]
20. T. Feng, Y. Xie, W. Xie, D. Ta, and Q. Cheng, “Bone chemical composition analysis using photoacoustic technique,” Front. Phys., 8 (2020).
21. W. Xie, T. Feng, D. Ta, L. Cheng, and Q. Cheng, “Bone microstructure evaluation by photoacoustic time-frequency spectral analysis,” in IEEE International Ultrasonics Symposium (IUS); 2020 (IEEE Publications, Las Vegas, NV, 2020), pp. 1–4.
22. W. Xie, T. Feng, M. Zhang, J. Li, D. Ta, L. Cheng, and Q. Cheng, “Wavelet transform-based photoacoustic time-frequency spectral analysis for bone assessment,” Photoacoustics 22, 100259 (2021). [CrossRef]
23. B. Borah, G. J. Gross, T. E. Dufresne, T. S. Smith, M. D. Cockman, P. A. Chmielewski, M. W. Lundy, J. R. Hartke, and E. W. Sod, “Three-dimensional microimaging (MRmicroI and microCT), finite element modeling, and rapid prototyping provide unique insights into bone architecture in osteoporosis,” Anat. Rec. 265(2), 101–110 (2001). [CrossRef]
24. F. Sevil and M. E. Kara, “The effects of ovariectomy on bone mineral density, geometrical, and biomechanical characteristics in the rabbit femur,” Vet. Comp. Orthop. Traumatol. 23(01), 31–36 (2010). [CrossRef]
25. D. R. Carter, M. L. Bouxsein, and R. Marcus, “New approaches for interpreting projected bone densitometry data,” J. Bone Miner. Res. 7(2), 137–145 (2009). [CrossRef]
26. R. Korinek, K. Bartusek, and Z. Starcuk, “Fast triple-spin-echo Dixon (FTSED) sequence for water and fat imaging,” Magn. Reson. Imaging 37, 164–170 (2017). [CrossRef]
27. F. Padilla, F. Jenson, V. Bousson, F. Peyrin, and P. Laugier, “Relationships of trabecular bone structure with quantitative ultrasound parameters: In vitro study on human proximal femur using transmission and backscatter measurements,” Bone 42(6), 1193–1202 (2008). [CrossRef]
28. M. L. McKelvie and S. B. Palmer, “The interaction of ultrasound with cancellous bone,” Phys. Med. Biol. 36(10), 1331–1340 (1991). [CrossRef]
29. A. Pifferi, A. Torricelli, P. Taroni, A. Bassi, E. Chikoidze, E. Giambattistelli, and R. Cubeddu, “Optical biopsy of bone tissue: A step toward the diagnosis of bone pathologies,” J. Biomed. Opt. 9(3), 474–480 (2004). [CrossRef]
30. R. L. P. van Veen, H. J. C. M. Sterenborg, A. Pifferi, A. Torricelli, E. Chikoidze, and R. Cubeddu, “Determination of visible near-IR absorption coefficients of mammalian fat using time- and spatially resolved diffuse reflectance and transmission spectroscopy,” J. Biomed. Opt. 10(5), 054004 (2005). [CrossRef]
31. M. Firbank, M. Hiraoka, M. Essenpreis, and D. T. Delpy, “Measurement of the optical properties of the skull in the wavelength range 650–950 nm,” Phys. Med. Biol. 38(4), 503–510 (1993). [CrossRef]
32. T. Feng, Y. Zhu, K. M. Kozloff, B. Khoury, Y. Xie, X. Wang, M. Cao, J. Yuan, D. Ta, and Q. Cheng, “Bone chemical composition assessment with multi-wavelength photoacoustic analysis,” Appl. Sci. 10(22), 8214 (2020). [CrossRef]
33. M. B. Tavakoli and J. A. Evans, “Dependence of the velocity and attenuation of ultrasound in bone on the mineral content,” Phys. Med. Biol. 36(11), 1529–1537 (1991). [CrossRef]
34. K. I. Lee, H.-S. Roh, and S. W. Yoon, “Acoustic wave propagation in bovine cancellous bone: application of the modified Biot–Attenborough model,” J. Acoust. Soc. Am. 114(4), 2284–2293 (2003). [CrossRef]
35. M. O. Culjat, D. Goldenberg, P. Tewari, and R. S. Singh, “A review of tissue substitutes for ultrasound imaging,” Ultrasound Med. Biol. 36(6), 861–873 (2010). [CrossRef]
36. L. Baofeng, Y. Zhi, C. Bei, M. Guolin, Y. Qingshui, and L. Jian, “Characterization of a rabbit osteoporosis model induced by ovariectomy and glucocorticoid,” Acta Orthop. 81(3), 396–401 (2010). [CrossRef]
37. J. F. Griffith, “Bone Marrow Changes in Osteoporosis,” in Osteoporosis and Bone Densitometry Measurements, G. Guglielmi, ed., Medical Radiology (Springer, 2013), pp. 69–85.
38. S. Hwang and D. M. Panicek, “Magnetic resonance imaging of bone marrow in oncology, Part 1,” Skeletal Radiol. 36(10), 913–920 (2007). [CrossRef]
39. C. Cordes, T. Baum, M. Dieckmeyer, S. Ruschke, M. N. Diefenbach, H. Hauner, J. S. Kirschke, and D. C. Karampinos, “MR-based assessment of bone marrow fat in osteoporosis, diabetes, and obesity,” Front. Endocrinol. 7, 74 (2016). [CrossRef]
40. M. A. Bredella, E. Lin, A. V. Gerweck, M. G. Landa, B. J. Thomas, M. Torriani, M. L. Bouxsein, and K. K. Miller, “Determinants of bone microarchitecture and mechanical properties in obese men,” J. Clin. Endocrinol. Metab. 97(11), 4115–4122 (2012). [CrossRef]
41. J. F. Griffith, D. K. Yeung, P. H. Tsang, K. C. Choi, T. C. Kwok, A. T. Ahuja, K. S. Leung, and P. C. Leung, “Compromised bone marrow perfusion in osteoporosis,” J Bone Miner Res 23(7), 1068–1075 (2008). [CrossRef]
42. B. K. Hoffmeister, S. A. Whitten, S. C. Kaste, and J. Y. Rho, “Effect of collagen and mineral content on the high-frequency ultrasonic properties of human cancellous bone,” Osteoporos. Int. 13(1), 26–32 (2002). [CrossRef]
43. C. M. Bagi, P. Ammann, R. Rizzoli, and S. C. Miller, “Effect of estrogen deficiency on cancellous and cortical bone structure and strength of the femoral neck in rats,” Calcif. Tissue Int. 61(4), 336–344 (1997). [CrossRef]
44. M. W. Sigrist and F. K. Kneubühl, “Laser-generated stress waves in liquids,” J. Acoust. Soc. Am. 64(6), 1652–1663 (1978). [CrossRef]
45. L. Yang, B. Lashkari, A. Mandelis, and J. W. Y. Tan, “Bone composition diagnostics: photoacoustics versus ultrasound,” Int. J. Thermophys. 36(5-6), 862–867 (2015). [CrossRef]
46. T. Feng, Y. Zhu, Y. Xie, D. Ta, J. Yuan, and Q. Cheng, “Feasibility study for bone health assessment based on photoacoustic imaging method,” Chin. Opt. Lett. 18(12), 121704 (2020). [CrossRef]
47. T. Feng, Y. Zhu, C. Liu, S. Du, D. Ta, Q. Cheng, and J. Yuan, “Ultrasound-guided detection and segmentation of photoacoustic signals from bone tissue in vivo,” Appl. Sci. 11(1), 19 (2021). [CrossRef]
48. M. Kim, G.-S. Jeng, M. O’Donnell, and I. Pelivanov, “Correction of wavelength-dependent laser fluence in swept-beam spectroscopic photoacoustic imaging with a hand-held probe,” Photoacoustics 19, 100192 (2020). [CrossRef]
49. S. L. Jacques, “Corrigendum: optical properties of biological tissues: a review,” Phys. Med. Biol. 58(14), 5007–5008 (2013). [CrossRef]
50. C. Liu, D. Ta, B. Hu, L. H. Le, and W. Wang, “The analysis and compensation of cortical thickness effect on ultrasonic backscatter signals in cancellous bone,” J. Appl. Phys. 116(12), 124903 (2014). [CrossRef]
51. G. Renaud, P. Kruizinga, D. Cassereau, and P. Laugier, “In vivo ultrasound imaging of the bone cortex,” Phys. Med. Biol. 63(12), 125010 (2018). [CrossRef]
52. J. Shepherd, G. Renaud, P. Clouzet, and K. van Wijk, “Photoacoustic imaging through a cortical bone replica with anisotropic elasticity,” Appl. Phys. Lett. 116(24), 243704 (2020). [CrossRef]
53. B. Depalle, Z. Qin, S. J. Shefelbine, and M. J. Buehler, “Influence of cross-link structure, density and mechanical properties in the mesoscale deformation mechanisms of collagen fibrils,” J. Mech. Behav. Biomed. Mater. 52, 1–13 (2015). [CrossRef]
54. H. Lei, L. A. Johnson, S. Liu, D. S. Moons, T. Ma, Q. Zhou, M. D. Rice, J. Ni, X. Wang, P. D. R. Higgins, and G. Xu, “Characterizing intestinal inflammation and fibrosis in Crohn’s disease by photoacoustic imaging: feasibility study,” Biomed. Opt. Express 7(7), 2837–2848 (2016). [CrossRef]
55. J. L. Williams and J. L. Lewis, “Properties and an anisotropic model of cancellous bone from the proximal tibial epiphysis,” J. Biomech. Eng. 104(1), 50–56 (1982). [CrossRef]
56. W. J. Whitehouse, “The quantitative morphology of anisotropic trabecular bone,” J. Microsc. 101(2), 153–168 (1974). [CrossRef]
57. P. Ammann and R. Rizzoli, “Bone strength and its determinants,” Osteoporos. Int. 14(S3), 13–18 (2003). [CrossRef]
58. H. H. Hu and H. E. Kan, “Quantitative proton MR techniques for measuring fat,” NMR Biomed. 26(12), 1609–1629 (2013). [CrossRef]
59. J. J. Kaufman, G. Luo, and R. S. Siffert, “Ultrasound simulation in bone,” IEEE Trans. Ultrason., Ferroelect., Freq. Contr. 55(6), 1205–1218 (2008). [CrossRef]
60. J. J. Kaufman and T. A. Einhorn, “Ultrasound assessment of bone,” J. Bone Miner. Res. 8(5), 517–525 (2009). [CrossRef]
