In vivo endoscopic optical coherence elastography based on a miniature probe

Optical coherence elastography (OCE) is a functional extension of optical coherence tomography (OCT). It offers high-resolution elasticity assessment with nanoscale tissue displacement sensitivity and high quantification accuracy, promising to enhance diagnostic precision. However, in vivo endoscopic OCE imaging has not been demonstrated yet, which needs to overcome key challenges related to probe miniaturization, high excitation efficiency and speed. This study presents a novel endoscopic OCE system, achieving the first endoscopic OCE imaging in vivo. The system features the smallest integrated OCE probe with an outer diameter of only 0.9 mm (with a 1.2-mm protective tube during imaging). Utilizing a single 38-MHz high-frequency ultrasound transducer, the system induced rapid deformation in tissues with enhanced excitation efficiency. In phantom studies, the OCE quantification results match well with compression testing results, showing the system's high accuracy. The in vivo imaging of the rat vagina demonstrated the system's capability to detect changes in tissue elasticity continually and distinguish between normal tissue, hematomas, and tissue with increased collagen fibers precisely. This research narrows the gap for the clinical implementation of the endoscopic OCE system, offering the potential for the early diagnosis of intraluminal diseases.


Introduction
Excessive dilation of the vagina during operative vaginal delivery can lead to vaginal injuries and the formation of hematomas.The onset of vaginal hematomas is covert, with no obvious symptoms or signs, making diagnosis difficult [1].Delayed diagnosis of hematomas can result in increased bleeding in patients, and it may cause maternal death in severe cases [2].Pathologic processes alter tissue elasticity, which often precedes visible morphological alterations [3][4][5].Therefore, elasticity assessment can be utilized for diagnosing vaginal issues.Traditional palpation methods, while commonly used for in vivo elasticity assessment, are limited by their reliance on the diagnostician's experience.Elastography, offering an accurate approach to evaluate tissue stiffness [6], and has the potential to enhance the diagnostic accuracy of diseases ranging from cancer to atherosclerosis [7].Hematoma is accompanied by alterations in tissue elasticity [8].Elasticity assessment of the vagina has the potential to improve the diagnosis of vaginal hematomas.
Elastography involves inducing stress in tissue using a mechanical stimulus, detecting resultant displacements and quantitatively reconstructing tissue strain to assess tissue stiffness.Various elastography modalities, such as magnetic resonance elastography (MRE), ultrasound elastography (USE), and optical coherence elastography (OCE), have been validated in clinical practice.MRE enables the non-invasive quantification of the biomechanical properties of soft tissues by directly visualizing propagating shear waves [9].Given its deep penetration depth, MRE proves beneficial for the general diagnosis and pre-operative assessment of large-scale tissues [10].Nevertheless, its resolution is limited to approximately 1 mm, making it unsuitable for evaluating tissue elasticity at a micrometer scale.USE offers a more flexible and intraoperative application, but it still faces limitations such as compromised spatial resolution, sensitivity and quantification accuracy [11,12], leading to fuzzy boundaries between elastic properties of the tissues and insufficient contrast to detect very small displacements.OCE serves as a functional extension of optical coherence tomography (OCT) [13], providing high-resolution (∼10 µm) imaging for elasticity assessment [14].Based on phase-sensitive OCT (PhS-OCT), it enables the accurate measurement of 1-nm axial displacement by accurately capturing phase changes in spectral interferograms [15], showcasing exceptional sensitivity and precise quantification [16,17].This positions OCE as a potent tool for early disease diagnosis by valuable insights into the intricate biomechanical properties of tissues at a microscopic level [18].OCE has been reported to be used for ex vivo elasticity assessment of the human vaginal wall [19], providing real-time and high-accuracy elastic images, thus revealing the potential clinical application of OCE in the vagina.Additionally, it has been reported that biomechanical properties of in vivo human cornea were measured by OCE [20], showing outstanding sensitivity and resolution of OCE technique.This underscores the potential clinical applications of OCE in diverse research areas.
However, current in vivo imaging capabilities of OCE are confined to external organs, such as the skin and eyes.The demonstration of endoscopic OCE imaging for deep-seated organs and tissues within the body is yet to be realized.This limitation arises due to the requirement for a specialized probe to conduct in vivo endoscopic measurements of biomechanical properties of internal tissues.Firstly, the probe should be as miniature as possible to navigate through lumens.Secondly, to meet the detection sensitivity requirements of OCE, it is necessary to apply a substantial excitation force from the probe to the tissues, inducing sufficient displacement within a short duration.Thirdly, the duration of the excitation should be significantly faster than the physiological motion cycles, including the cardiac and respiratory cycles, to mitigate the influence of potential motion artifacts.Various approaches have been undertaken to attain these objectives.Qu et al. introduced a 3.5-mm front-facing miniature probe with an 8.8-MHz ring transducer to generate elastograms of tissue with high mechanical sensitivity [21].They managed to perform imaging on an ex vivo human cadaver carotid artery sample and tissue abnormality was detected.This innovation has led to the creation of the first endoscopic OCE system capable of obtaining ex vivo OCE images and demonstrating the potential of endoscopic OCE in identifying plaque vulnerability through the detection of elasticity changes.Karpiouk et al. designed an OCE probe with the outer diameter of 2.5-mm, which consists of two 4.8-MHz ultrasound transducers for ultrasonic shear wave excitation and a commercial OCT catheter for detection [22].They used the ultrasound transducers to create shear waves in a rabbit heart and enabling the ex-vivo measurement of its elasticity properties.These two studies demonstrated the capability to perform elastography, but these probes are too oversize, hindering their ability to navigate within lumens for successful in vivo imaging.Latus et.al reported an OCE system with a 5.5-MHz hand-held ultrasound probe for shear wave excitation and a 1-mm dual-fiber probe for detection [23].Although they demonstrated the feasibility of endoscopic OCE to quantify elasticity in ex-vivo coronary arteries, they did not integrate the excitation source into the OCE probe.The separation of excitation and detection components in the OCE imaging systems pose challenges for conducting in vivo endoscopic imaging.Recently, Wang et.al reported an OCE system achieved a 1.1-mm motorized probe [24].This is the most promising research for achieving in vivo intravascular OCE imaging.It is the fastest endoscopic OCE system with an imaging speed reaching up to 160 frames per second.It induced a pressure change by varying fluid flow rate in the vascular lumen and achieved OCE imaging of tissue strain in human coronary artery samples.The lumens of blood vessels typically have symmetrical geometry, and this method is an optimal technique for intravascular OCE imaging.However, when subjected to fluid flow pressure, heterogeneous tissues such as the vagina and uterus have nonsymmetric stress distribution, which may lead to a decrease in quantification accuracy.Therefore, this method is not suitable for endoscopic OCE imaging of the vagina.Overall, to our knowledge, there is a lack of reports of in vivo endoscopic OCE imaging outcomes.Small animals are used for in vivo imaging in basic research.Present endoscopic OCE systems encounter challenges related to complex and bulky probe design, suboptimal excitation efficiency and long duration of excitation, which can hardly be used in small animals.In order for clinical translation, the ongoing basic research focus revolves around the development of probe miniaturization, and highly efficient excitation-detection methods to achieve rapid excitation and precise endoscopic elasticity detection.
In this study, we introduce a novel endoscopic OCE system, with a compact miniature imaging probe merely 0.9-mm in diameter.Enhanced excitation efficiency is achieved through the utilization of a single 38-MHz high-frequency ultrasound transducer with short duration of the excitation.Excitation-induced displacement refining method and surface extraction and filtering method were applied to suppress motion artifacts.The quantification results of agar phantoms with different stiffness levels show the high accuracy of our endoscopic OCE system in measuring elastic parameters.Based on these advantages of our system, we successfully achieved the first endoscopic OCE imaging in vivo and the imaging results of rat vagina illustrated that our system is capable of quantifying the elastic properties of normal tissue, hematoma tissue and tissue with collagen fibers increasement, which demonstrated the potential for clinical translation.

System setup and probe design
The system design, as illustrated in Fig. 1(a), employed a swept-source laser (Axsun Technologies, Massachusetts, USA) with a central wavelength of 1310 nm, a bandwidth of 105 nm, and a repetition rate of 50 kHz.The output energy onto the sample is 15 mW.The axial and lateral resolutions were 18.3 µm and 12.5 µm respectively (see section 1 of Supplement 1 for more details).OCT signals were acquired by using a balanced detector (PDB470C-AC, Thorlabs, USA) with a data acquisition (DAQ) card (ATS9350, Alazar Tech, Canada).A 38-MHz sinusoidal wave generated by a function generator (DG4102, Rigol Technologies, Beijing, China) was modulated by a rectangular wave with a pulse width of 0.8 ms.The modulated wave was amplified by a power amplifier (LZY-22+, Mini-Circuits, New York, USA) and transmitted to an ultrasound transducer (Insight Lifetech, Guangdong, China) with a center frequency of 38 MHz to generate the excitation acoustic waves.The pulse-echo response and the frequency spectrum of the transducer was shown in Fig. 1(b).The trigger signals from OCT laser were used to synchronize the excitation, data acquisition as well as the rotation and pull-back motion performed by stepper motors.The time synchronization was achieved by a programmed STM32 microcontroller.
The structural design of the home-made probe was shown in Fig. 1(d) (see section 2 of Supplement 1 for more details).A gradient-index (GRIN) lens (AMOS, Shaanxi, China) with both ends cleaved at 8 degrees, an optical reflection prism (Fuzhou OYeah Optronics Co., Ltd, Fujian, China), and a 0.5 × 0.6 mm 2 ultrasound transducer were sequentially arranged and encased in a stainless-steel housing.A single-mode fiber (SMF28) was used to transmit optical beams.The distal end of the fiber was cleaved at 8 degrees and the proximal end was a FC/APC fiber connector which connected the probe to the system's optical path.A double-layer coil (Tu's Cheng Fa, Guangdong, China) drove the whole probe to perform rotational and pull-backing movements.The photograph of the probe was shown in Fig. 1(c) and the zoomed-in photograph was shown in Fig. 1(e).

Data acquisition and processing
The system was synchronized to perform scanning in M-B-scan mode.As Fig. 2(a) shown, 500 A-lines were acquired to form one M-scan (motion-scan) for phase analysis.The excitation acoustic wave was emitted 0.4 ms after the start of each M-scan acquisition.The first 20 A-lines (corresponding to an acquisition duration of 0.4 ms) of each M-scan were averaged to form an OCT B-scan image.Triggers for the stepper motors were delayed 10 ms after the start of M-scan.A schematic diagram of the relationship between OCT B-scan image and M-scan images was shown in Fig. 2(b).Each M-scan contained information about the phase difference change over time at that location, from which the elasticity was calculated.The elasticity was characterized by Young's modulus: where σ is the axial stress on the sample, ε is the strain of the sample in the axial direction, ∆l is the axial deformation of the sample, and l is the thickness of the sample.The displacement in OCE images indicated the deformation of the sample.Assume that the samples had the same refractive index, each pixel unit in the data possessed the same spatial size, i.e. l is the same.Given the same stress, Young's modulus was proportional to the inverse of displacement.The displacement information was acquired by processing data from the PS-OCT.The overall data processing flow was shown in Fig. 2(d).Firstly, the phase difference was calculated separately on each M-scan using Kasai autocorrelation algorithm [25]: where z, x are the depth and lateral position, respectively, t denotes the t th A-line, F is the frequency-domain OCT complex signal.Then, the averaged OCT B-scan image was binarized to create a target mask to separate tissues from background.Morphological opening and closing operations were sequentially applied to the target mask to select continuous and non-porous regions of the tissues for further calculations.A low-pass filter was used to remove high-frequency phase noise from the phase difference data.The velocity of axial motion of each unit is related to the Doppler phase shift and was calculated by the following equation [14]: where v z,x is the axial velocity of the sample unit, λ is the center wavelength of the light source, n is the refractive index of the sample, τ is the time interval and θ is the Doppler angle.In this study, λ=1.31 µm, n = 1.38 [26], τ = 20 µs and θ = 26°.The cumulative displacement was obtained by integrating the velocity over time.The displacement of each unit was the maximum value of the cumulative displacement.The curve of velocity and cumulative displacement versus time in one excitation-detection cycle was shown in Fig. 2(c).The displacement was the result of cumulative calculation of phase difference.The system's phase noise was 357 mrad measured by a reflector placed at the focal point, corresponded to 27.0-nm displacement.After cumulative calculation, the phase noise was effectively smoothed out, approximating an average over about 40 A-line data points in the process (within excitation duration of 0.8 ms and A-line time interval of 20 µs).Finally, the phase noise was 10.8 mrad, corresponded to 0.82-nm displacement.

Motion artifact correction algorithm
The motion artifacts mean the relative motion between the probe and the sample surface, including physiological motion of the animal and shake of the probe.Motion artifact correction algorithm included displacement correction and image distortion correction.The idea of displacement correction is to deduct the displacement due to motion, leaving only the displacement due to the excitation.The schematic of the displacement correction was shown in Fig. 3(a) and (b).Physiological motion induced additional displacement during the 10 ms excitation-detection cycle, as shown by the red arrow in Fig. 3(a), leading to deviation in the displacement calculation.
The motion velocity due to respiration was assumed to be uniform over the period of 10 ms [27].
A time window from the fifth millisecond to ninth millisecond was selected, as shown by the green dashed box in Fig. 3(a), and the average result within this window was calculated as the motion velocity.The original velocity curve was calibrated by subtracted the motion velocity and the cumulative displacement was recalculated.The cumulative sum (CUSUM) control chart method was used to detect the velocity mutation due to excitation, as shown by the blue arrow in Fig. 3(b).The accumulated displacement corresponding to the velocity mutation time point was the baseline of the displacement.Finally, the displacement for each unit was determined as the maximum value derived from subtracting the baseline from the cumulative displacement.The schematic of the image distortion correction was shown in Fig. 3(c), (d) and (e).The idea of extracting and flattening tissue surface curve (or plane) to correct the distortion of OCT images has been reported previously [28,29].Here, a surface extraction and filtering method was used.First, the upper surface curvature of the tissue was extracted: the points with the minimum depth on each A-line of the target mask with true values were selected, and these points formed a curve called target curve.Target curve fluctuated due to the influence of physiological movements, as shown by the green curve in Fig. 3(c) and the dark blue curve in Fig. 3(d).Next, according to the frequency of respiration [27], low-pass filtering was performed on the target curve to obtain a smoothed curve, as shown by the red curve in Fig. 3(d).The difference between the smoothed curve and the target curve was used to calculate the translation, then all columns of the image were translated to obtain the motion artifact corrected images.Both OCT and OCE images applied the translation.The motion artifact correction algorithm was framed by green dashed line in the data processing flow, shown in Fig. 2(d).

Experimental setup of phantom studies
To simulate the acoustic attenuation properties of soft tissues, a uniform gelatin phantom with a concentration of 20% (g/100 mL) were prepared, with the addition of 0.4% (v/v) intralipid to enhance optical backscattering.The phantom was refrigerated at 5°C for 24 hours.The detailed recipe was identical to that described in the Ref. [30].Subsequently, the acoustic attenuation coefficient of the gelatin phantom was measured.Two identical ultrasound transducers were aligned facing each other, one emitting pulse waves and the other receiving.The phantom was placed between the two transducers, and the acoustic waves received before and after the insertion of the phantom were acquired.FFT analysis was performed, and the reduction in amplitude at the frequency of 38 MHz was compared to calculate the acoustic attenuation coefficient at this frequency.Finally, the phantom underwent pull-back scanning imaging, which was performed by pulling the probe back and scanning along the longitudinal direction to acquire longitudinal section images.The excitation peak to peak voltage (V PP ) input to the ultrasound transducer was 27 V.The excitation duration was set to 0.8 ms for all experiments in this study.
To validate the displacement accuracy of the OCE system, 6 uniform agar phantoms of different stiffness levels were produced, with concentrations of 0.4% (w/v), 0.6%, 0.8%, 1.0%, 1.2%, and 1.5%, each with 1.6% (v/v) intralipid added to enhance optical backscattering.The preparation process of these phantoms was similar to that in the Ref. [31].Granular agar was dissolved in distilled water at 25°C.Then, the solution was stirred, and heated until boiling.When heating was stopped, stirring continued, and the solution was naturally cooled to 70°C.Intralipid was then added, stirred for 5 minutes, and the solution was poured into cylindrical molds with a diameter of 38 mm and a height of 20 mm.The solution was then placed in a refrigerator at 4°C for solidification.Subsequently, each phantom underwent pull-back scanning imaging with an excitation voltage of 16.3 V PP .Imaging was repeated 10 times for each phantom, and the average displacement values were calculated in a consistent region of interest (ROI) size.Finally, a mechanical tester (QJ210, Qingji Instrumentation Technology, Shanghai, China) was used to conduct a compression test on the phantoms.The rate of compression was 0.1 N/s, keep compressing the phantoms until deterioration was seen.The Young's modulus of the phantoms was calculated from the measured pressure and strain.

Animal experimental setup and histopathology preparation
Three Sprague Dawley (SD) rats (female: 10-week-old, weighing about 250 g, Guangdong Vital River Laboratory Animal Technology Co. Ltd.) were bred and housed under standard conditions.All experiments were performed under isoflurane anesthesia.The experiments conducted on animals have adhered to the protocols that had been approved by the animal study committee of the Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences.To induce elasticity changes in the rat vagina, we inflicted injury to the rat vagina.The method of inducing vaginal injury was referenced to the method used for mechanical injury to the endometrium [32].This is because the vagina belongs to the external reproductive tract, and the morphological structure of vaginal mucosa is similar to that of the uterus.Vaginal mucosal surface injury was inflicted using a self-made blunt scraper, aiming to cause abrasions until a palpable fluctuation was felt beneath the surface without extensive mucosal bleeding.To observe the elasticity changes of the vagina during the hematoma and self-repair process after injury, the vaginas of a normal rat, a rat 24 hours after injury, and a rat 72 hours after injury were imaged separately.During imaging, a transparent protective tube made of PA12 material was used, with an outer diameter of 1.2 mm.Before conducting animal imaging, the protective tube was filled with distilled water, the probe was placed in, and air bubbles were removed.Pull-back scanning and rotational scanning was performed in each rat vagina in turn.Rotational scanning was performed by rotating the probe and scanning along the radial direction to obtain the cross-section image.An excitation voltage of 16.3 V PP was consistently applied to the transducer for each imaging session.After being imaged, the rats were euthanized, and their vaginas were collected, then Masson staining was performed and images were taken with a slide scanner (Olympus, Japan).Masson staining is a histological staining technique used to visualize connective tissues in tissue sections, particularly collagen.The collagen fibers of the vaginal tissue were stained blue after Masson staining.Tissue stiffening is associated with the increasement of collagen fibers [33].

Phantom experiments
To verify the effective imaging depth, we conducted a phantom experiment.The measured acoustic attenuation coefficient of the gelatin phantom was as follows: at a frequency of 38 MHz, the acoustic attenuation coefficient of the phantom was 20.81 ± 0.94 dB/cm.This result was close to the average value for soft tissues at the same frequency, which was 20.52 dB/cm [34].Therefore, the phantom was able to mimic soft tissue in terms of sound attenuation properties.The results of OCT and OCE imaging were shown in Fig. 4(a) and (c), respectively.To analyze these results, the ROI of the phantom was selected in Fig. 4(a).The curve of OCT signal intensity after horizontal averaging versus penetration depth was plotted, as shown in Fig. 4(b).The intensity curve was smoothed using a Gaussian filter and 0 dB was set to be the maximum value of the smoothed curve.The depth corresponding to a 3 dB decrease in the maximum intensity was considered to be the maximum penetration depth of OCT.The maximum penetration depth of OCT in the phantom was 0.94 mm.Similarly, the ROI was selected in Fig. 4(c), the curve of displacement after horizontal averaging versus penetration depth was plotted, as depicted in Fig. 4(d).The displacement curve was smoothed using a Gaussian filter and 100% of the displacement was set to be the maximum value of the smoothed curve.Displacement changes within 10% were considered constant.Within a depth of 0.88 mm, the displacement in the OCE image remained essentially unchanged with increasing depth.The results showed essentially identical displacement in OCE image within the penetration depth of the OCT imaging, indicated that the depth range of uniformly applied excitation force matched with the depth of OCT imaging.
The results of the experiment validating the accuracy of OCE imaging were presented in Fig. 5(a).The Young's modulus values of the agar phantoms measured through compression testing were plotted on the x-axis, while the inverse of the displacement values obtained from OCE images were plotted on the y-axis.The scatter points represented the inverse of the average displacement values for each ROI in the OCE imaging results.A linear fitting line was plotted through these points, with the fitting result R 2 = 0.9986.The results demonstrated a highly linear correlation between the displacement quantified by OCE imaging and the Young's modulus of the phantoms.Figure 5(b)-(g) showcased representative OCE images for each concentration of the agar phantoms.The result showed that the system possessed high quantitative accuracy in the range of 37.7-100.0kPa.

Motion artifact correction
In vivo OCT and OCE imaging were performed in the vagina of a normal rat, as shown in Fig. 6.It was observed that due to physiological movements such as breathing of the animal, the surface of the vagina exhibited spike-like distortions, causing abrupt changes in displacement values in the corresponding positions of the OCE image.After applying the motion artifact correction algorithm, the OCT and OCE results, as depicted in Fig. 6(b) and (d) respectively, showed a suppression of the vaginal surface distortions, smoothing of displacement value mutations.The result showed that the motion artifacts in the images were mitigated.

In vivo imaging of rat vagina
In vivo imaging was performed in rat vagina lumens of the three rats.The OCE and OCT images shown in Fig. 7(a-d) were the vaginal tissue of the rat without injury.Figure 7(f-i) showed OCE and OCT images of the rat vagina 24 hours after injury and Fig. 7(k-n) showed OCE and OCT images of the rat vagina 72 hours after injury.Figure 7(a), (f), and (k) were longitudinal OCE images, and the white dashed lines indicated the locations where the cross-sectional images were taken.Figure 7(b), (g), and (l) were cross-sectional OCT images and Fig. 7(c), (h), and (m) were cross-sectional OCE images.Figure 7(d), (i), and (n) were the zoomed-in OCE images at the white dashed boxes of Fig. 7(c), (h), and (m), respectively.Figure 7(e), (j), and (o) showed the corresponding Masson staining of the rat vaginal tissue without injury, 24 hours after injury, and 72 hours after injury, respectively.The blue arrows in Fig. 7(d) and (e) indicated the correspondence of tissue morphology between the OCE image and histological image.The green arrows in Fig. 7(n) and (o) indicated the correspondence of tissue morphology between the OCE image and histological image.The displacement at positions indicated by white asterisks in Fig. 7(h) was higher than the overall displacement in Fig. 7(c) (∼4.5 µm compared to ∼3.5 µm).These areas of higher displacement indicated softer tissue, with Young's modulus decreased by ∼22%.The displacement in Fig. 7(m) was lower than those in Fig. 7(c) (∼2.0 µm compared to ∼3.5 µm).These areas of lower displacement indicated stiffer tissue with Young's modulus increased by ∼75%.The blue color in Masson staining indicated collagen fibers.In normal tissue and tissue 24 hours after injury, the collagen fibers in the Masson staining image (Fig. 7(e) and  (j)) did not show obvious changes.In tissue 72 hours after injury, a slight increase in collagen fibers was shown in the Masson staining image (Fig. 7(o)).As indicated by the red arrows in Fig. 7(j), the epithelium was seen to be rough at 24 hours after injury, and blood cell clots were seen next to the rough epithelium, which were adjacent to the injured epithelium.In Fig. 7(i), the red arrows indicate the regions of high displacement, which correspond to the locations where blood clots appear in Fig. 7(j).

Discussion
OCE imaging technology is capable of detecting elastic changes caused by pathological alterations in tissues sensitively, holding potential to improve the diagnosis of vaginal hematomas and contribute to the maintenance of female reproductive health.In this study, we demonstrated a novel endoscopic OCE system capable of conducting the first elasticity imaging in vivo.We designed a compact OCE probe with the outer diameter of only 0.9-mm, which is the smallest integrated OCE probe to our knowledge.The probe consists of an optical fiber and a GRIN lens for phase-sensitive OCT detection and a miniature high-frequency ultrasound transducer to induce deformation in tissues effectively.Based on our displacement calculation method, we achieved elasticity images with high quantification accuracy and minimal motion artifacts by integrating excitation-induced displacement refining method and surface extraction and filtering method.In in vivo imaging experiments in the rat vagina, the system demonstrated the capability to offer supplementary insights into tissue mechanical properties beyond the morphological information provided by OCT alone.Thus, it exhibited the capacity to detect and differentiate lesions of varying severity.
A study utilizing compression OCE with high imaging speed to improve early detection of vaginal wall injuries [19], demonstrating the contribution of OCE technology to enhancing diagnostic accuracy.However, this method was limited by the use of static compression loading, making it difficult to achieve in vivo elasticity assessment.OCE methods based on acoustic radiation force hold promise for in vivo imaging but are limited by large probe size and suboptimal excitation efficiency.In our endoscopic OCE system, we proposed a solution using a highfrequency ultrasound transducer to achieve efficient excitation and reduce probe size, enabling in vivo endoscopic imaging.An ultrasound transducer with a frequency as high as 38 MHz is key for OCE, providing high excitation efficiency and short duration of excitation, which enables it possible in OCE imaging scenarios in vivo.Ultrasound transducer with higher frequency tends to provide greater acoustic radiation forces, resulting in higher axial stress applied to target tissues (see section 3 of Supplement 1 for details).Consequently, it only requires a lower excitation voltage to generate the same acoustic radiation force for elastography, contributing to enhanced efficiency.In addition, higher stress for elastography can be achieved by reducing the excitation area of target tissues.We opted to decrease the cross-sectional area of the acoustic beam by reducing the area of the transducer surface to 0.5 × 0.6 mm 2 .These adjustments provide sufficient axial stress on tissue, resulting in a more rapid deformation of the tissue, enabling us to shorten the excitation duration and minimize the influence of motion artifacts during in vivo imaging.In a previous work [21], a displacement of 0.75 µm was induced in a phantom of 23.1 kPa, using excitation voltage of 70 V PP and excitation duration of 1 ms.In our results in Fig. 5, a displacement of 2.97 µm was induced in a similar phantom of 37.7 kPa, using excitation voltage of 16.3 V PP and excitation duration of 0.8 ms.It means that we used a smaller excitation voltage and shorter excitation duration to induce a larger displacement in a stiffer phantom, indicating the high excitation efficiency of our OCE system.However, a challenge associated with the use of high-frequency transducers is that as the frequency increases, so does the attenuation of sound in tissue.This makes it difficult for high-frequency transducers to induce sufficient displacement at greater depths [35].Current elastography studies predominantly use low frequency transducers (1-10 MHz) for excitation [21,22,36,37], due to their ability to provide sufficient acoustic radiation force for tissue displacement at greater depths (1-10 mm).Unlike these elastography studies, our system was designed to match the penetration of OCT detection with that of the ultrasound excitation.OCT imaging penetrates only up to ∼1 mm within tissue, where the attenuation of high-frequency ultrasound is relatively low.With our endoscopic OCE system, we assessed the changes in OCT intensity and OCE displacement along the axial direction of a uniform phantom with an acoustic attenuation coefficient similar to that of soft tissue.The results, as shown in Fig. 4, indicated a relatively constant displacement across OCT penetration depths, confirming the viability of utilizing the 38-MHz high-frequency transducer within the penetration depth of OCT imaging.Furthermore, for applications in highly attenuating tissues such as vessel tissues, the use of a focused ultrasound beam is an effective means of expanding the depth of excitation and providing a uniform depth range of excitation [38].In future work, we will design and apply miniature ultrasound focusing elements [39] to make the sound field generated by the high-frequency ultrasound transducer more uniform in the depth direction.
Another advantage of using high-frequency transducer is the miniaturization of probe size.The center frequency of the transducer is inversely related to the thickness of the piezoelectric material [40].Compared to low-frequency transducers, high-frequency transducers are significantly thinner (the 38-MHz transducer has a thickness of only ∼0.3 mm) and have a smaller volume, facilitating the reduction in probe size.The compact size of the elastography probe for specific endoscopic imaging scenarios is crucial.For example, in order to access the small branches of coronary arteries, the imaging probe needs to satisfy an outer diameter of < 1 mm [41].However, none of the OCE probes in previous studies had an outer diameter less than 1 mm [21][22][23][24].Therefore our 0.9-mm probe is meaningful.Additionally, the use of a miniature probe offers the advantage of minimizing the pressure applied to the luminal wall, ensuring that the imaging captures the tissue in its natural state.In the section 4 of Supplement 1, we presented the in vivo imaging results obtained with a probe having an outer diameter of 1.2-mm in the rat vagina.This demonstration revealed that larger probes exert additional pressure on the luminal walls, resulting in inaccurate elasticity measurements.Such inaccuracies are detrimental to the effectiveness of in vivo imaging.This underscores the significance of employing a smaller probe size, such as our 0.9-mm probe, to ensure more accurate and reliable elastography outcomes, especially in scenarios requiring access to delicate and confined anatomical structures.It is crucial to reduce the size of the probe in endoscopic imaging.However, it often affects the performance of probe.Therefore, the key is to maintain excellent performance while reducing probe size.In the field of endoscopic elastography, efficient excitation is needed under the limited size of the probe.Therefore, our innovation lies in using high-frequency ultrasound as excitation, which not only addresses the efficiency issue but also helps reduce probe size.
Our endoscopic OCE system possesses excellent performance required for in vivo imaging.Results presented in Fig. 5 show a linear correlation between the inverse of displacement and the Young's modulus across various samples, affirming the high accuracy of OCE imaging.This accuracy can be attributed to the utilization of a high-frequency transducer, which enhances excitation efficiency.The efficient excitation achieved with the high-frequency transducer promotes rapid sample deformation, characterized by increased phase differences per unit time.In this scenario, the influence of phase noise becomes tolerable.Additionally, the system benefited from our displacement calculation method, where the accumulation process of phase differences further diminishes the influence of phase noise (improved from 357 mrad to 10.8 mrad).Physiological movements introduce challenges to in vivo OCE outcomes, such as morphologically distorted and elasticity-biased images.Such errors may pose the risk of potential misdiagnoses.To address this issue, we integrated various motion correction methods.First, a displacement refining method was implemented to correct elasticity bias by subtracting the motion-induced displacement from the excitation-induced displacement.Additionally, a morphological correction method was employed, involving the extraction of tissue surface.This extraction curve underwent processing with a low-pass filter, and subsequently, the OCE images were calibrated with the filtered curve.The outcomes presented in Fig. 6 not only showcased the robustness of our calibration methods but also highlighted their efficacy in mitigating the impact of motion artifacts.
In our experimental setup, we employed imaging techniques on rat vaginal tissues, deliberately inducing injury through mechanical abrasion to observe changes in elasticity.In Fig. 7, we tracked changes in tissue elasticity at two distinct time points, specifically after 24 hours and 72 hours.At 24 hours after injury, the injured vaginal tissue showed a decreased elasticity compared to normal vaginal tissue, as observed through our OCE system.This is because hematomas are formed by red blood cells and fibrin, causing a decrease in tissue stiffness [8].At 72 hours after injury, the self-repair process of the tissue had begun, marked by the increasement of collagen fibers, resulting in a higher elasticity as evidenced by OCE images [33].The concurrent examination using Masson staining reveals the presence of hematomas at the 24-hour mark and collagen fibers increasement at the 72-hour interval.The temporal correlation observed between tissue elasticity via OCE and histological changes identified by Masson staining provides a dynamic perspective to our comprehension of tissue response.This emphasizes the potential clinical significance of our approach in distinguishing distinct pathological states.Figure 7 also illustrates that the OCT intensity B-scan reveals minimal differences, whereas the OCE B-scan discerns different stiffnesses sensitively in the vaginal tissue after injury.These findings align with those obtained from Masson staining, providing consistent evidence of the induced tissue hematoma and collagen fibers increasement changes.OCE is positioned as a functional extension of OCT technology, providing additional information on tissue elasticity changes, aiding in the highly sensitive diagnosis of lesions.This reveals the potential of our endoscopic OCE as a tool for assessing vaginal injuries in clinical diagnosis.
A safety measurement (refer to section 5 of Supplement 1) was conducted to validate the viability of the efficient high-intensity output in in vivo elastography.The result indicated that under the present system settings and experimental conditions, the transducer's output intensity stayed within the diagnostic ultrasound limits regulated by the FDA.Consequently, our system has the capacity to exert excitation forces to the fullest extent while upholding biosafety standards, making it well-suited for in vivo elastography.
Our system faces limitation of imaging speed.The system's two-dimensional (cross-sectional, longitudinal) imaging rate is less than 1 frame per second, which lags behind the fastest reported endoscopic OCE technology (160 frames per second) [24].The imaging speed is limited by both excitation and detection.The current system features relatively long excitation duration and suboptimal phase stability.Shortening the excitation duration will lead to a more significant impact of phase noise on elasticity calculation resulting in a decrease in the accuracy of elasticity quantification.Shortening the excitation duration will also lead to a reduction in OCE image displacement resulting in a decrease in image contrast.To ensure adequate accuracy and contrast in OCE images, we chose longer excitation durations at the expense of imaging speed.In the future, we plan to make the following improvements: Firstly, enhancing the system's phase stability.The introduction of phase-stable methods, such as applying k-clock, λ-trigger and common-path configuration [42,43], could provide a solution.Secondly, increasing the excitation voltage and utilizing laser sources with higher A-line rates.Both increasing the excitation voltage and extending the excitation duration can increase displacement and enhance the contrast in OCE images [21].Increasing the excitation voltage accelerates tissue deformation, and employing a faster laser source can track the rapid deformation of tissue.With an appropriate excitation voltage, the excitation duration can be reduced to less than 0.1 ms, thereby increasing imaging speed by an order of magnitude.Future research directions include improving imaging speed to achieve high-speed endoscopic OCE imaging, enabling intravascular elasticity assessment; enhancing the sensitivity of the system for small displacement detection and expanding the dynamic range of elasticity quantification, thereby enabling accurate assessment of significant elasticity changes in vessels, tumors, and other tissues.

Conclusion
In summary, we presented a novel endoscopic OCE system, which features the most compact OCE probe currently available and successfully achieves the first OCE imaging in vivo.The utilization of a 38-MHz ultrasound transducer as excitation source enabled probe miniaturization and facilitated highly efficient OCE imaging of the rat vaginal wall.The motion artifact correction method we used effectively mitigated image distortions and minimized errors in elasticity calculations arising from physiological movements during in vivo imaging.The obtained imaging results demonstrated that our system has the capability of discerning alterations in tissue elasticity, which were manifested as hematoma and collagen fibers increasement in histological images.This validated the system's ability to assess alterations in tissue pathological status by detecting changes in tissue elasticity.This work lays the groundwork for clinical translation of endoscopic OCE system, holding promise for early diagnosis of intraluminal diseases.

Fig. 2 .
Fig. 2. (a) Scanning scheme.(b) Schematic diagram of OCT image and phase difference images obtained by M-B-scan mode protocol.(c) Diagram of phase difference versus cumulative displacement over time in one excitation-detection cycle.(d) Flowchart of data processing.

Fig. 3 .
Fig. 3. (a) Velocity and displacement curves affected by motion.The red arrow indicated the displacement caused by motion.The green dashed box indicated the time window for velocity calibration calculation.(b) Velocity and displacement curves after motion artifact correction.The blue arrow indicated a velocity mutation in the velocity curve.(c) OCT image of tissue affected by motion.(d) Diagram of target curve on the tissue surface and the smoothed curve.(e) OCT image after motion artifact correction.

Fig. 4 .
Fig. 4. Imaging results of the gelatin phantom simulating the acoustic attenuation properties of soft tissues.(a) OCT image.(b) The curve of average OCT signal amplitude versus penetration depth in the white dashed box in (a).(c) OCE image.(d) The curve of average displacement versus penetration depth in the white dashed box in (c).

Fig. 5 .
Fig. 5. Accuracy assessment of OCE image elasticity quantification.(a) Quantitative comparison between OCE imaging results and compression test results.(b-g) OCE imaging results for agar phantoms, the numbers below the images represent the concentration of agar phantom.

Fig. 6 .
Fig. 6.OCT and OCE images of rat vagina in vivo before and after processing by motion artifact correction algorithm.(a) OCT image before correction.(b) Corrected OCT image.(c) OCE image before correction.(d) Corrected OCE image.

Fig. 7 .
Fig. 7. Imaging results of the three rat vaginas.(a-e) In vivo normal rat vagina, (f-j) rat vagina 24 hours after injury, (k-o) rat vagina 72 hours after injury.(a,f,k) Longitudinal OCE image, (b,g,l) cross-sectional OCT image, (c,h,m) cross-sectional OCE image, (d,i,n) zoomed-in OCE image and (e,j,o) Masson staining image.(c), (h) and (m) were the cross sections taken relative to the white dashed lines in (a), (f) and (k), respectively.High displacement areas were indicated by the white asterisks in (h).Blood clots were indicated by the red arrows in (j).The red arrows indicate the regions of high displacement in (i).The blue arrows in (d) and (e), as well as the green arrows in (n) and (o), indicated surface concave features of tissue morphology.EP: epithelium and LP: lamina propria.