Quantitative assessment of thrombosis-induced blood oxygenation change in deep tissues based on photoacoustic tomography: an ex vivo study

Abstract

The staging and classification of thrombosis hold significant clinical value for optimizing thrombus treatment strategies. In this study, we propose a quantitative method based on photoacoustic tomography for assessing thrombosis in deep tissues. By using inner chromophore signals as a correction factor, this approach minimizes the ‘spectral coloring’ effects caused by overlying heterogeneous tissues. Ex vivo experiments validate that the method acquires accurate spectra up to a depth of 30 mm across various tissue conditions. After calibration, the Pearson correlation coefficients calculated for the spectrum in deep tissue against the uncolored absorption spectrum is 15% higher, and the standard deviation of the Pearson correlation coefficients decreased by 58%. Sequential measurements capture time-dependent spectral changes of thrombus phantom during six days, providing a potential diagnostic reference for thrombus formation time and type. This method offers a non-invasive, practical tool for accurately quantifying thrombosis stages, which might be valuable for optimizing treatment strategies.

1. Introduction

Assessment of vascular health is crucial for clinical treatment and ongoing health monitoring. Thrombosis is a significant vascular health issue, which may lead to ischemic stroke, pulmonary embolism, tissue ischemia, and other serious complications [1,2]. Since the optimal clinical treatment depends largely on the type of thrombus, accurately determining its age and type is essential [3,4]. Current clinical methods for diagnosing vascular occlusion, such as X-ray and ultrasound, do not yet provide an accurate method of determining the stage and age of a thrombus. The formation of a thrombus typically commences with an intricate network of activated platelets and fibrin. A fresh clot contains more red blood cells (RBCs) [5]. After thrombus, the activity of agglutinated RBCs gradually declines, accompanied by changes in other thrombus components including recruited smooth muscle cells and fibroblasts [5,6]. Therefore, measuring alterations in the composition of a thrombus could be a key to diagnosing its age and type. Researchers have utilized ultrasound elastography to assess changes in vascular wall elasticity during thrombus evaluation [7], while MRI [8] and OCT [9] have been employed to analyze protein alterations during thrombus formation.

Photoacoustic tomography (PAT) [10,11] is an emerging functional imaging modality. Due to its biologically safe nature of being non-invasive as well as non-ionizing and its ability to combine optical contrast with ultrasound imaging depth, PAT has shown promising results in early tumor research [12], blood oxygen saturation measurement [13], brain imaging [14], breast cancer diagnosis [15], among others [11]. Based on the photoacoustic effect, where the molecule absorbs a short laser pulse and emits a thermoelastic acoustic wave, the intensity of the photoacoustic images is proportional to the light absorption of the chromophore. By using illumination with multiple wavelengths, PAT can capture the absorption spectra of chromophores [16]. Since each molecule in biological tissue has a unique light absorption spectrum, tissues with different molecular compositions can be differentiated according to their light absorption characteristics [16–18]. Given hemoglobin's optical absorption properties, photoacoustic tomography (PAT) enables the detection of progressive deoxygenation during thrombus formation, thereby facilitating the assessment of thrombus staging. Therefore, PAT could be a competitive candidate for the thrombosis classification. Tang et al. proposed that the changes in the oxygen saturation of hemoglobin (sO₂) of thrombus can be measured using photoacoustic imaging with intravascular light delivery [19]. However, further analysis is needed to quantitatively measure the sO₂ of deep - tissue thrombi through non - invasive methods and to characterize the changes in thrombus sO₂ over time.

However, laser propagation in complex tissue reduces the accuracy of measuring the light absorption spectra of chromophores deep within the tissue. Before reaching the thrombus, the laser beam passes through multiple tissue layers, which exhibit significant spectral differences in light energy at various laser wavelengths. This is due to wavelength-dependent attenuation of the light energy, a phenomenon known as ‘spectral coloring’ [20–22]. The optical absorption properties of layered tissue differ in terms of light attenuation and acoustic properties, influenced by variations in body composition, such as lipid thickness and skin pigmentation. To account for these differences in light scattering and absorption, ones need to know the tissue structure in advance to calculate the impact of overlying tissues. Numerous methods have been proposed to correct ‘spectral coloring’. For example, Monte Carlo simulations have been used to predict luminous flux [23], iterative methods have been suggested to optimize light absorption and scattering coefficients of the tissues [22,24], and some methods quantitatively estimate fluence attenuation using eigenspectra of tissues [17]. However, these methods often require high computational resources and prior knowledge of tissue structure, which complicates their clinical applications of classifying thrombi deep within the tissue. In a recent study [25], Thomas et al. proposed the use of known chromophores as fluence marker, enabling quantitative imaging of carotid plaques and providing a potential framework for evaluating spectral coloring and achieving quantitative measurements. However, the influence of deep tissue on quantitative thrombus imaging in deep as well as thrombus staging in deep tissues, require further investigation.

In this work, we propose a method based on PAT to quantitively assess the stage of thrombosis deep within tissue. Measurements at multiple wavelengths were conducted to obtain the optical absorption curves of both thrombus and adjacent blood vessels. Signals of the blood in the vascular near the thrombus were used to reduce the effects of ‘spectral coloring’ caused by the overlying tissue. A series of ex vivo experiments were conducted to validate the proposed method.

2. Method

With exposure to the laser pulse, the light-absorbing tissue absorbs the laser energy, thermally expands and contracts, and produces ultrasound. The pressure of the sound source p₀(r, λ) can be expressed as [26],

$$p_0(\mathbf{\text{r}},\lambda )=\mathrm{\Gamma }A(\mathbf{\text{r}},\lambda )I(\mathbf{\text{r}},\lambda ),$$ (1)

where A is the optical absorption coefficient, I is the light flux arriving at the absorbers, Γ is the Grüneisen parameter, r and λ represent the spatial coordinates and laser wavelength, respectively. A and Ι are spatially variant and wavelength-dependent. The induced ultrasound field p(r, t) can be governed by the following equation,

$$\frac{{\mathrm{\partial }}^{2}p(\mathbf{\text{r}},t)}{\mathrm{\partial }{t}^2}-c^2{\mathrm{\nabla }}^{2}p(\mathbf{\text{r}},t)=\mathrm{\Gamma }\frac{d\text{H}(t)}{dt},$$ (2)

where H(t) is the temporal waveform of laser pulse and c is the speed of sound. With the detected photoacoustic signals p(r, t), the sound source can be reconstructed as p₀'(r) = ∑p(r, τ) with τ = ||r – r _d ||/c using a delay-and–sum (DAS) method [27]. Here, r _d is the position of ultrasound detectors. Practically, considering the response of the signal detection system, the reconstructed p₀'(r) and the actual sound source p₀(r) have the relationship of p₀′(r, $\lambda$ ) = Kp₀(r, $\lambda$ ), where K is a constant relating to the system response. Therefore, the thrombus absorption spectrum A_t(λ) ≡ A(r, λ)|_r = rt, where r _t represents the coordinates of the thrombus, can estimated by

$$A_t(\lambda )=\frac{1}{\text{K}\mathrm{\Gamma }}\times p_0^{\mathrm{\prime }}(\mathbf{\text{r}},\lambda ){|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_t}\times I^{-1}(\mathbf{\text{r}},\lambda ){|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_t}.$$ (3)

However, the spectral estimation based on Eq. (3) is not practical, especially when the thrombus is located deep within the tissues. This is because the light flux I arriving at deeper tissue is generally unknown. The magnitude of I(r, t) depends not only on the laser source but also on the optical properties as the laser passes through the tissue, a phenomenon known as ‘spectral coloring’. A simple assumption of a uniform luminous flux must lead to measurement errors of light absorption coefficients [20], due to the uncertainty in the light flux. Therefore, obtaining an accurate spectrum of the thrombus deep within tissue is challenging because of the underdetermined light flux.

We proposed a method for accurately measuring the absorption spectrum of thrombosis in deep tissue by using endogenous chromophores as a calibrator. The approach relies on the fact that endogenous molecules, such as hemoglobin in the blood vessel near the thrombus, can be identified and their optical properties remain stable. Veins and arteries, along with their branches, often run in parallel within the extremities [28,29], allowing simultaneous measurement of the thrombus and the nearby blood. Moreover, hemoglobin in blood exhibits relatively stable optical properties, because the oxygen saturation in major vessels of the extremities remains normal [30,31]. Following thrombotic occlusion, biological systems activate collateral circulation to preserve relative stability in local oxygen supply [32,33]. Therefore, the nearby vascular network could provide a natural reference for correcting the spectral measurement of the thrombus. In more intricate scenarios, such as tissue hypoxia where the absolute sO₂ of the vessels is inaccessible, exogenous markers, like indocyanine green (ICG), can be utilized to calibrate the distortion induced by the overlying tissue. Therefore, we can make an assumption that the reference spectrum is either known or predictable.

Supposing the thrombus at r _t and the adjacent blood vessel at r _c can be measured simultaneously, the relationship between the reconstructed sound sources and their optical absorption be written as

$$p_0^{\mathrm{\prime }}(\mathbf{\text{r}},\lambda ){|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_t}=\text{K}\mathrm{\Gamma }{A}_t(\lambda )I(\mathbf{\text{r}},\lambda ){|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_t},$$ (4)

$$p_0^{\mathrm{\prime }}(\mathbf{\text{r}},\lambda ){|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_c}=\text{K}\mathrm{\Gamma }{A}_c(\lambda )I(\mathbf{\text{r}},\lambda ){|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_c}.$$ (5)

Considering the thrombus and the adjacent blood vessel are close to each other, i.e., r_t $\approx$ r_r, it has I(r, λ)|_r = rt $\approx$ I(r, λ)|_r = rc, Therefore, dividing Eq. (4) by Eq. (5), we have

$$A_t(\lambda )=A_c(\lambda )\times \frac{{p_0^{\mathrm{\prime }}(\mathbf{\text{r}},\lambda )|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_t}}{{p_0^{\mathrm{\prime }}(\mathbf{\text{r}},\lambda )|}_{\mathbf{\text{r}}={\mathbf{\text{r}}}_c}}.$$ (6)

In Eq. (6), both p₀'(r, λ)|_r = rt and p₀'(r, λ)|_r = rc is measurable, and the absorption spectrum of hemoglobin A_c(λ) has been well studied [34–36]. The spectral estimation based on Eq. (6) is practical. Moreover, the estimated spectral A_t(λ) does not depend on the measurement system or the propagation path of light. Therefore, A_t(λ) estimated by Eq. (6), only relying on the properties of the thrombus, can be used to assess the thrombus stages.

3. Ex vivo experiment

Figure 1 illustrates the experimental setup and the prepared samples. Finally, we apply this method to assess the development of a thrombus model from the acute phase to the subacute phase. Experiments were conducted to validate the proposed method. A Q-switched Nd: YAG laser (OPOTEK, LLC, USA) was used as a light source to illuminate the samples and produce photoacoustic signals. The laser pulse has a pulse duration of 4.5 ns and a repetition rate of 10 Hz, with a turntable wavelength of 690–950 nm. An ultrasonic linear array transducer, with a center frequency of 7.5 MHz and a –6 dB relative bandwidth of 80%, was used for signal acquisition. The transducer has 128 elements with an element pitch of 0.3 mm, providing a theoretical lateral resolution of 105 µm and a theoretical axial resolution of 284 µm. The signal acquisition system is based on a custom-designed ultrasound system, with a sampling frequency of 40 MHz.

Fig. 1.

Experimental setup and the samples. (a) Schematic illustration of the experimental system. Two vessel phantoms were placed parallel under different types of overlying tissue. (b) Side view picture of the system and phantom. (c) Top view picture of the phantom.

The imaging samples consist of soft transparent tubes filled with sheep blood and black ink, placed beneath various tissues. The tube used in this experiment was fabricated from polyvinyl chloride material, with an inner diameter of 4 mm and an outer diameter of 6 mm. The blood sample used in the experiment was defibrinated sheep blood. All experimental procedures were performed in conformity with protocols approved by the Animal Studies Committee of Nanjing University. Sheep blood shows a decline in erythrocyte activity and progressive deoxygenation. This process simulates the changes in blood following thrombus formation. After preparing the blood tube samples, we then proceeded with the experiment and carried out continuous measurements for six consecutive days. Black ink, with stable light absorption properties over time during the six-day-long measurement, was used to represent the adjacent blood, whose optical absorption properties are predictable according to the assumption. This helps mitigate potential experimental impacts caused by changes in the optical absorption coefficients of reference samples, allowing greater focus on the calibration and measurement in deep tissue. The two phantoms, made from sheep blood and black ink, represent the target and reference vessels, respectively. In our ex vivo experiment, the target vessel and the reference vessel were placed horizontally side by side and embedded under tissues, mimicking blood vessels located deep within tissue. We used different tissue types, including muscle, fat, and skin, to simulate complex conditions encountered in clinical deep vascular measurements.

Multispectral measurements were employed in this study, using wavelengths from 700 nm to 850 nm in an interval of 10 nm. This range encompasses the near-infrared band, where significant changes occur in blood optical absorption properties, including the isosbestic point [34,35]. We performed 30 repeated laser excitations at each wavelength to avoid the influence of fluctuations in laser energy and interference from electromagnetic noise. The data obtained at each excitation is used to reconstruct an image. The measured maximum radiant exposure on the tissues is about 15 mJ/cm² within the ANSI safety limit. To calculate the sound pressure of the absorber, we define the region of the sound source in the region of interest (ROI) by using image segmentation based on a simple thresholding method [37]. The average value within the region was used as the measured sound pressure signal. The average value of the 30 excitations is represented as marks in the box plot.

Additionally, the spectral curves were measured over 6 days with 24-hour intervals to observe temporal variations, thereby enabling the assessment of the progression through different stages of thrombosis.

4. Results

Figure 2 illustrates the process of quantitively assessing thrombosis in deep tissue. By illuminating the phantom with a laser at different wavelengths and repeating the imaging process, a series of images were obtained, as shown in Fig. 2(a). Figure 2(b) shows PA signals from the phantom. These signals were used to reconstruct an image of the phantom using the DAS algorithm. As shown in Fig. 2(c), both the target vessel and the reference vessel are visible in the reconstructed images. Figure 2(d) displays the image intensity of two phantoms as a function of the wavelength. The spectra of the two phantoms show significant differences. The optical absorption spectra of the reference shown in Fig. 2(e) was acquired by measuring the transmission spectroscopy of the sample using a fiber-coupled monochromator (Ocean Optics, USB2000+). Finally, the optical absorption spectra of the target vessel were calculated using Eq. (6), as shown in Fig. 2(f). In the following experiments, we used the above process to assess the vessel targes placed beneath various tissues.

Fig. 2.

Flow charts of the quantitative measurement of absorption spectrum deep within tissue. (a) PAT images obtained at multiple wavelengths. (b) PA signals in 128 channels. (c) Reconstructed PAT image derived from the signals. (d) Calculated optical absorption curves of the target and the reference. (e) Spectrum of the used ink sample. (f) Calibrated absorption spectrum of the target using the proposed method. The marks represent the mean value, while the whiskers indicate the standard deviation.

In the first experiment, we evaluated the performance of our method using tissues of varying thicknesses. The tissues used were chicken muscle and porcine muscle, with thicknesses of 0 mm, 10 mm, 20 mm, and 30 mm. The results are presented in Fig. 3. As shown in Figs. 3(a) and 3(d), the spectra directly extracted from the image intensity differ significantly, even for the same vessel samples. The differences are more pronounced at shorter wavelengths (700-800 nm). Additionally, the coloration effect caused by chicken muscle and porcine muscle is also distinct. On one hand, in the 700-800 nm wavelength range, the stronger light scattering and absorption by the upper tissue layers in this band may have led to the distortion of the spectral shape [35]. Compared to chicken muscle, pig muscle contains more myoglobin, which results in greater light attenuation. Consequently, the raw spectra measured beneath pig muscle exhibit more pronounced spectral distortion than those measured beneath chicken muscle. These results demonstrate that the spectral curve derived directly from the image intensity depends on both the tissue type and the thickness between the sample and the probe. The spectral variation can impact the measurement accuracy for targets deep within the tissue. In contrast, the spectral curves obtained using the proposed method (the right column) are closely aligned. This indicates that the measurement is independent of the tissue thickness through which the laser passes and the coloration effects are minimized.

Fig. 3.

Multispectral measurement through tissues of varying thicknesses. The left column (a) and (d) and the middle column (b) and (e) illustrate the image intensity of the target vessel and the reference vessel as a function of wavelength, respectively. The right column (c) and (f) represent the decolored spectra obtained using the proposed method. The tissue thicknesses through which the laser passes are 0 mm, 10 mm, 20 mm, and 30 mm, respectively.

In the second experiment, we measured the absorption spectra through four different types of tissues: chicken muscle, porcine muscle, porcine skin, and porcine fat. All tissues had the same thickness of 10 mm. The results are presented in Fig. 4. Figure 4(a) and 4(b) illustrate the spectral coloration of different tissues with the same thickness. As shown, different tissues have varying effects on the spectrum. For instance, porcine fat exhibits stronger attenuation of the spectra at shorter wavelengths, which may be attributed to the strong scattering properties of fat. Additionally, at 750-780 nm, the colored spectra display non-monotonic variations. Figure 4(c) presents the spectral curves obtained using the proposed method. It can be observed that the spectra obtained through different tissues agree well with each other, indicating that the coloration caused by different types of tissue is effectively reduced by the proposed method.

Fig. 4.

Multispectral measurement through different types of tissue: chicken muscle, porcine muscle, porcine skin, and porcine fat. (a) Image intensity vs. wavelength curves of the target vessel. (b) Image intensity vs. wavelength curves of the reference vessel. (c) Decolored spectrum of the target vessel.

In the third experiment, we examined the performance of our method using different multi-layered heterogeneous tissues. To simulate a more realistic scenario for spectral measurement of thrombus deep within tissue, porcine skin, porcine fat, and porcine muscle were stacked layer by layer as three distinct overlying tissues. In such cases, a model-based decoloring method often requires prior information about each layer and involves significant computational costs. Figure 5 presents the results. The multi-layered heterogeneous tissues have a more pronounced effect on spectral measurement. As shown in Figs. 5(a) and 5(b), after passing through these tissues, the spectra differ significantly from each other. This discrepancy may be due to additional spectral coloration introduced by the interfaces between different tissue layers. Figure 5(c) displays the spectra obtained using the proposed method. The consistent results demonstrate that the proposed method remains effective even when passing through multi-layered tissues with complex heterogeneity and interfaces.

Fig. 5.

Multispectral measurement through multi-layered heterogeneous tissues. Overlayer i is composed of porcine skin, overlayer ii is composed of stacked porcine skin and porcine fat, and overlayer iii is composed of stacked porcine skin, porcine fat, and porcine muscle, respectively. (a) Image intensity vs. wavelength curves of the target vessel. (b) Image intensity vs. wavelength curves of the reference vessel. (c) Decolored spectra of the target vessel.

We used the Pearson Correlation Coefficient (PCC) to evaluate the consistency of the measured spectra across different tissues [38]. The PCC, defined as the cross-correlation of two vectors, is a widely used statistical measure to quantify the correlation between two vectors. A PCC value closer to 1 indicates a stronger correlation. We calculated the PCC between the spectral curve measured through the tissue and the spectral curve measured directly in water. According to [35], the influence of water on the laser energy in this wavelength range is negligible for a short optical path. Thus, the spectral curve measured in water can be regarded as the one without coloration.

Figure 6 presents the results, where red boxes represent the results before decoloring and blue boxes correspond to the results after decoloring using the proposed method. It is evident that the PCCs of the spectra obtained using the proposed method are very close to 1, with a narrow distribution range. High PCCs can still be achieved even for multi-layered heterogeneous tissues with different thicknesses, types, and compositions. These indicate that the spectral modification caused by the spectral coloration of various tissues was effectively reduced. The spectra obtained by the proposed method are dependent solely on the optical absorption properties of the target vessel, but not on the tissue along the light propagation path. Correlation analysis on spectrum before and after calibration with the uncolored spectrum was then performed. Compared with direct measurements, the mean PCC values increased from 0.84 to 0.96 after calibration with an average improvement of 15% per group. Meanwhile, the mean value of standard deviation (SD) decreased from 0.038 to 0.014 with an average reduction of 58% per group. Therefore, the decolored spectra can be reliably used to quantify changes in the target deep within the tissue. Moreover, the proposed method does not require prior knowledge of the tissue properties along the light propagation path or complex computations, making it more practical for real clinical scenarios. Thus, the proposed method offers a robust quantitative measurement approach for assessing the stage of thrombus deep within tissue.

Fig. 6.

Evaluation of the consistency of the measured spectra using PCC. Red boxes represent the results before decoloring, while blue boxes correspond to the results after decoloring.

Finally, we applied the proposed method to examine ex vivo thrombus model and monitor the thrombus development process from the acute to the subacute phase, evaluating the feasibility of the method for thrombus staging. We simulated the blood flow obstruction during the acute phase of thrombus formation by blocking sheep blood in a transparent soft tube. Over time, the coagulated blood gradually turned dark red and eventually deep dark red, representing the progression to the subacute phase of thrombus formation. We conducted spectral measurements on the thrombi model at equal intervals over a continuous period to capture changes in the thrombus spectral curves. The spectra were measured continuously over six days, with a 24-hour interval between measurements. Polynomial fitting was used to analyze the spectral curves, and the first-order fitting coefficients were extracted to illustrate the time-dependent changes in the thrombi model.

Figure 7 gives the time-varying curves of blood. Figure 7(a) displays the images of the thrombi model on each day. Over time, the color of the thrombi model deepened, reflecting the changes following blood occlusion formation. The spectral curves obtained by the proposed method are shown in Fig. 7(b). The solid lines represent the mean value, and the shaded areas indicate the SD. Although these curves were measured through different multi-layered heterogeneous tissues, as in the previous experiments, the measurements are consistent with each other, and the SDs (shaded areas) remain within a small range. This suggests that these spectra can quantify the biochemical characteristics of the thrombus avoiding the interference from the complex tissues through which the light passes.

Fig. 7.

Assessment of the variation of an ex vivo thrombus model over 6 days. (a) Photographs of the model on each day. (b) Spectral curves of the model for each day. (c) Calculated slopes extracted from the spectra. (d) Calculated oxygen saturation of the target compared with the slope. The inset in (d) is the optical absorption coefficients of the oxyhemoglobin and deoxyhemoglobin [35].

As time progresses, the spectra of the thrombus model change significantly. To quantify these changes, we calculated the slope of the first-order fitting of the spectral curves. Figure 7(c) gives the spectral slope values on different days. The blue boxes represent the slope values from curves measured through tissues, while the red boxes represent the direct measurements without tissue.

The changes in the two groups of box plots show consistency over time. This indicates that our method can provide quantitative measurements of deep tissues without the influence of overlying tissues. Furthermore, as the thrombus progresses from the acute to subacute phase, the spectral parameters change monotonically. The mean value of the slope changed from 0.068 to −0.044. These results demonstrate the proposed PAT-based method has the potential to provide an effective non-invasive diagnostic approach for thrombus staging. By solving the system of linear equations, the sO₂ was derived from the calibrated curves as shown in Fig. 7(d). The inset in Fig. 7(d) is the optical absorption coefficients of the oxyhemoglobin and deoxyhemoglobin used [35]. As shown in Fig. 7(d), the sO₂ of the sample progressively decreased from high to low oxygen levels with a rapid change from the third day to the fifth day, depicting the deoxygenation process following thrombus formation. This provides valuable reference for diagnosing the thrombus stage. Furthermore, the results indicate a strong correlation between the curve slope and the variation in blood oxygen saturation which demonstrates that the curve slope is an effective parameter for assessing the thrombus stage.

5. Conclusion

In this study, we develop a PAT-based method for the quantitative assessment of thrombus stages deep within tissues. We performed measurements at multiple wavelengths to obtain the optical absorption curves of both the thrombus and adjacent blood vessels. The effects of ‘spectral coloring’ caused by overlying tissues are reduced by utilizing blood signals from nearby vascular as a calibrator. Therefore, this method does not rely on prior knowledge of tissue properties and high computational costs, making it simple and easy to use.

The method was validated through a series of ex vivo experiments, demonstrating its capability to accurately obtain spectra up to a depth of 30 mm, independent of overlying tissue. Four sets of experiments conducted under various overlying and multilayer tissue conditions confirmed the general applicability and practicality of the method. Compared to direct measurements that assume constant light flux, the spectra obtained using this method exhibited a higher correlation with the spectra of objects unaffected by tissue. Pearson correlation coefficients further validated the effectiveness of the proposed method. The mean value of PCCs improved by 15% while the SDs decreased by 58%, which suggests that the method effectively reduced the impact of overlying tissue, resulting in a more reliable absorption spectrum for deep thrombi. Additionally, time-dependent changes in ex vivo blood were quantitatively characterized using polynomial fitting, offering a potential diagnostic reference for evaluating thrombus formation time and type in clinical settings. There are still certain limitations in this study. A more precise thrombus model could be established to measure the changes in the spectrum during the aging process of the thrombus, and multimodal methods such as ultrasonic elastography could be used for comprehensive analysis.

In summary, the PAT-based method presented in this study provided a highly effective, non-invasive tool for quantifying the stage of thrombosis. By utilizing reference signals from endogenous chromophore nearby vascular structures, such as hemoglobin in blood vessels, the method achieves accurate and reliable spectral profiles with minimal spectral coloring effects. The successful tracking of thrombus progression from the acute to subacute phase, alongside significant spectral changes, demonstrates its potential for clinical applications in thrombus staging. This method could be valuable for optimizing thrombus treatment strategies.

Funding

National Natural Science Foundation of China10.13039/501100001809 (12027808, 12374436).

Disclosures

The authors declare no conflicts of interest.

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.