Open Access
Add to BibSonomy
Abstract
Current endoscopy techniques have difficulties to provide both high resolution and large imaging depth, which significantly hinders the early diagnosis of gastric cancer. Here, we developed a label-free, large-depth, three-dimensional (3D) chromatic reflectance confocal endomicroscopy. In order to solve the problem of insufficient imaging depth of traditional chromatic confocal microscopy, a customized miniature objective lens both with large chromatic focal shift and correction for spherical aberration was used to focus light of different wavelengths at different depths of the sample simultaneously, and a fiber bundle containing 50000 single-mode cores was used to collect the confocal reflectance signal. To acquire detailed information along the axial direction at a faster speed, a high-speed multi-pixel spectrometer was used to realize simultaneous detection of multi-depth signals. Specifically, we have built up a label-free fiber-optic 3D chromatic reflectance confocal endomicroscopy, with 2.3 µm lateral resolution, imaging depth of 570 µm in 3D phantom and 220 µm in tissue, and 1.5 Hz 3D volumetric frame rate. We have demonstrated that the fiber-optic 3D chromatic confocal endomicroscopy can be used to image human gastric tissues ex vivo, and provide important morphological information for diagnosis without labeling. These results show the great potential of the fiber-optic 3D chromatic confocal endomicroscopy for gastric cancer diagnosis.
© 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
Corrections
12 January 2022: Typographical corrections were made to the funding section.
1. Introduction
In order to accurately access the early gastric lesion, imaging depth is considered as important as resolution [1,2]. The tumor invasion depth is required to be no more than 1/3 of the superficial layer of the submucosa (SM1, <500 µm) for gastric cancer [3] to apply endoscopic resection. Once the tumor invasion exceeds these depths, radical resection is required to reduce the risk of lymph node metastasis [4]. Therefore, an endoscopy method with large depth (500 µm) is highly desirable for the treatment decision making [1,2]. As an “optical biopsy” tool, confocal endomicroscopy has been applied for early diagnostics in gastric cancer owing to its superior subcellular resolution [5,6]. Traditional confocal endomicroscopy technique requires mechanical Z-scan to achieve depth-resolved imaging and it requires fluorescent label for imaging [7,8]. For label-free imaging, confocal reflectance imaging has been applied in biology and provided high resolution in the lateral and axial direction [9,10]. Yoo et al. utilized a high-speed form of confocal reflectance confocal to accurately count intraepithelial eosinophils and identify other microscopic abnormalities associated with eosinophilic esophagitis on freshly excised biopsy samples [10]. However, conventional confocal reflectance microscopy is excited by single wavelength which limits the imaging depth to the range of 200–300 µm [11,12]. In recent years, chromatic confocal technology appears to have the possibility of fast multi-depth imaging with high resolution [13–19]. In chromatic confocal imaging technology, when white light passes through a lens, each monochromatic light component has different focus on the optical axis due to the different refractive index for different wavelengths of traditional lens. Thus, a dispersion element can be used to encode the optical axis with wavelength. Therefore, compared with traditional confocal [7], multi-wavelength imaging can be realized in chromatic confocal microscope without any mechanical axial scanning. In chromatic confocal technology, chromatic focal shift of lens is utilized for focusing multispectral continuous light source at different depths of sample [20]. Olsovsky et al. developed a chromatic confocal microscope using four aspheric lenses which contribute to the chromatic aberration and a water immersion lens with numerical aperture (NA) of 0.8 as objective, and achieved 150 µm imaging depth with a wavelength range of 590–775 nm and ∼3 µm axial resolution [13]. But they only performed volumetric imaging of buccal mucosa which was immersed in acetic acid for 30 seconds to enhance backscattering from cell nuclei. Liang et al. developed a digital mirror device (DMD)-based chromatic confocal microscope and achieved 45 µm imaging depth with wavelength of 505–650 nm and ∼12 µm axial resolution [18]. The works above have demonstrated the potential of using chromatic confocal system for medical application, but their systems were not miniaturized to an endoscopy setting. Lane et al. presented a fiber-based chromatic confocal microendoscopy by using a 1 mm-diameter gradient-index (GRIN) lens to achieve a range of chromatic focal shift of 40 µm over 200 nm wavelength range, and the optical sectioning was demonstrated by imaging a microprocessor chip [21]. Kang et al. designed a spectrally encoded confocal endoscopic probe, in which an objective lens with a titled arrangement was used to generate a focal line that was not parallel to the tissue surface to spanned over a large range of depths [22]. But its image quality of tissue was limited by insufficient signal-to-noise ratio. The imaging depths in these works are not sufficient for early cancer diagnosis which requires endoscopy techniques with both cellular resolution and large penetration depth [2]. Therefore, there is an unmet need for a chromatic confocal endomicroscope with both high resolution and large imaging depth with demonstration of performances on human stomach tissues without labeling.
Here we developed a label-free fiber-optic large-depth 3D chromatic reflectance confocal endomicroscopy by designing a customized miniature objective lens composed of five lenses. In chromatic confocal technology, the objective lens is the key to achieve large depth and high resolution. Especially in the chromatic confocal endomicroscopy, it is a technical challenge to make a tiny lens with both large axial chromatic aberration and small spherical aberration at the same time. In our work, we introduced a design of eliminating spherical aberration in developing a miniature objective lens composed of five single lenses to significantly eliminate spherical aberration while ensuring large axial chromatic aberration. With such design in chromatic confocal endomicroscopy, it can offer both high-resolution and large imaging depth. We used the customized miniature objective lens to focus light of different wavelengths at different depths of the sample simultaneously, and an imaging depth of 570 µm in 3D phantom and 220 µm in biological tissue with 2.3 µm lateral resolution were achieved. It is the first time that an imaging depth of more than 500 µm has been achieved in chromatic confocal endomicroscopy, to the best of our knowledge. It bridges the gap between high-resolution and large-depth of chromatic confocal endomicroscopy. We then used a high-speed multi-pixel spectrometer and a set of homemade programs to realize simultaneous acquisition of multi-depth confocal signals at 1.5 Hz 3D imaging volumetric frame rate. Compared with the method which using multiple photodetectors to detect signals of different wavelengths, our method is capable of reaching both high spectral resolution and large imaging depth at the same time. Given the 2.3 µm lateral resolution, large imaging depth, and fast 3D imaging capabilities without z-axis mechanical scanning, this confocal endomicroscopy has been used to image human gastric tissues and shown great potential for gastric cancer diagnosis.
2. Results
We developed a novel fiber-optic chromatic reflectance confocal endomicroscopy for label-free, large-depth and fast 3D imaging without z-axis mechanical scanning. As shown in Fig. 1, a customized miniature objective lens with large chromatic focal shift was used to focus light of different wavelengths at different depths. And correction for spherical aberration was introduced in design of the miniature objective lens to improve axial resolution. Then, light reflected from the tissue was collected by a single-mode multi-core fiber bundle and only the light in focus could be collected, thereby achieving high-resolution via confocal scheme. Finally, a high-speed spectrometer was utilized to achieve multi-depth data acquisition, and the image reconstruction was realized in real time. As a result, tissue information from multi-depth can be detected simultaneously. We have achieved a volumetric imaging speed of 1.5 frames per second with 2.3 µm lateral resolution and imaging depth of 570 µm in 3D phantom and 220 µm in human tissue. Regarding the axial resolution, it varied with wavelength at our working wavelength of 650–950 nm, and the averaged axial resolution was 12.2 µm and 26.6 µm with wavelength of 650–800 nm and 800–950 nm, respectively.
Fig. 1. Schematic of fiber-optic large-depth 3D chromatic confocal endomicroscopy containing probe design and depth-coded signal detection. L1-L4: Lens 1–4. M 1–2: Mirror 1–2. D1: Diameter of the probe. D2: Distance between fiber bundle and miniature objective lens. λ1…n: Different wavelengths focused at different depths. Gel: Fiber gel.
2.1 Ray-tracing simulation of the customized miniature objective lens
The miniature objective lens has a NA of 0.3 in object space and 0.21 in image space. As shown in Fig. 2(a)-(b), the chromatic focal shift of the miniature objective lens in the z-direction can reach 0–810 µm with the light wavelength of 650–950 nm, based on the trend of the dispersion curve traced. To achieve large imaging depth, the chromatic aberration of the miniature objective lens was utilized to simultaneously focus light with different wavelengths on different depths in tissue. A multi-core fiber bundle containing 50000 cores with NA of 0.4 was used for light delivery and signal collection. At the proximal end, dual-axis galvo mirrors were used for two-dimensional scanning. The signal at the focus of each core was reflected back to the fiber bundle and collected by photodetector for 3D chromatic confocal imaging.
Fig. 2. Ray-tracing simulations on the performance of the endomicroscopy. (a) Layout of miniature objective lens in optical simulation and enlargement of the dispersion range (wavelength from 650 nm to 950 nm). (b) Dispersive curve of miniature objective lens. (c) Simulation curve of maximum chromatic focal shift depending on the distance between fiber bundle and miniature objective lens. (d) Simulation curve of Airy radius and RMS radius depending on the distance between fiber bundle and miniature objective lens. (e) Simulation curve of object space NA and image space NA depending on the distance between fiber bundle and miniature objective lens. (f) Calculation data of magnification of miniature objective lens depending on the distance between fiber bundle and miniature objective lens.
With a micro lens at the distal end of fiber bundle, the resolution is limited by core-to-core space of the fiber bundle [23,24]. The theoretical lateral resolution of the endoscope system can be described as:
As shown in Fig. 2(c)-(d), when D2 was 19 mm (at the front focal plane of the miniature objective lens), the maximum focal shift was 810 µm with the wavelength of 650–950 nm and the maximum Airy radius was 1.546 µm at the wavelength of 800 nm. With the increase of D2, the maximum focal shift was increased. And when D2 increased, both the object space NA and the image space NA gradually decreased (Fig. 2(e)), while according to the formula [24]:
2.2 Performance characterization of fiber-optic chromatic confocal endomicroscopy
First, depth calibration was performed with 3D phantom of silicon wafer grooves and each groove of the silicon wafer was 50 µm wide. In order to achieve simultaneous measurement of multiple depths, a high-speed spectrometer with working wavelength of 650–950 nm was used. For large-depth imaging with the light wavelength range from 650 to 950 nm, D2 was adjusted to 19 mm (at the front focal plane of the miniature objective lens) to ensure sufficient chromatic focal shift and best resolution. A motorized stage was used to change the object distance between the phantom and endomicroscopic probe. Endomicroscopic reflectance confocal signals of silicon wafer grooves were collected by the high-speed spectrometer and 3D image of 256*256*2048 pixels was reconstructed. Besides, as shown in Fig. 4(a), two 3D images were obtained before and after moving the endomicroscopic probe 100 µm in the z direction and 3D images were merged together. And then XZ and YZ cross-sectional images (Fig. 4(b)-(c)) were used to measure the signal displacement of the same groove in the z direction to calibrate the depth vs axial pixels as shown in Fig. 4(b). It was clearly that the position change of the silicon groove in the z-axis direction was distinguished. The results clearly showed the capability of optical sectioning of our endomicroscopy.
Fig. 3. Measurements of the performance of the endomicroscopy. (a) Measurement of imaging depth in different wavelengths, specifically the depth ranges from 0 to 750 µm when using wavelength from 650 nm to 912 nm. (b) Corresponding spectra at different depths. (c) The measurement of the chromatic focal shift with different wavelengths when the distance D2 between the fiber bundle and the lens was different. (d) Endomicroscopic image of resolution target. (e) Measurements of the lateral resolution using resolution target shown in (d). (f) Measurement of axial resolution based on the spectral FWHM, FWHM: full width at half maximum.
Fig. 4. 3D imaging performance of the system. (a) 3D image in 3D chromatic confocal endomicroscopy of the silicon wafer groove with a groove width of 50 µm and two 3D images of the silicon wafer groove at two different positions in the Z direction were merged together. (b-c) The XZ and YZ cross-section images of the merged two 3D images in (a), specifically, the displacement of the two 3D images in Z direction was 100 µm. (d-f) Representative X-Y projection (MIP mode) of lens cleaning paper, onion and rhizome of celery, respectively. Different depths were represented by different colors. Scale bars, 50 µm.
Then, imaging depth measurement was conducted by imaging a 3D phantom which was made by mixing 10 µm-diameter Al beads in 2.5% agarose gel (see Visualization 1). According to the previous depth calibration, the imaging depth in the 3D phantom of 10 µm-diameter Al beads mixed in 2.5% agarose gel was 570 µm.
Furthermore, lateral resolution of the system was measured by imaging a resolution target and the Group 7 Element 5 can be distinguished (Fig. 3(d)). According to the measurement of the FWHM of the signal intensity distribution of Group 7 Element 5, a lateral resolution of 2.3 µm was achieved as shown in Fig. 3(e). And axial resolution of the system was measured based on the spectral FWHM. The spectral FWHM of different wavelengths corresponding to different depths was measured. Then, according to the relationship between wavelength and depth, the depth resolution was calculated. As shown in Fig. 3(f), the axial resolution varied with wavelength at our working wavelength of 650–950 nm, and the averaged axial resolution was 12.2 µm and 26.6 µm with wavelength of 650–800 nm and 800–950 nm, respectively.
In addition, in order to demonstrate the capability of 3D imaging of the system, we performed 3D imaging of the lens cleaning paper, onion and rhizomes of celery, respectively (Fig. 4(d)-(f)). Representative X-Y projection (maximum intensity projection (MIP) mode) of lens cleaning paper, onion and rhizome of celery were shown and information of different depths was represented by different colors. It can be seen from these results that our fiber-optic chromatic confocal endomicroscope has the capability of three-dimensional tomography.
2.3 Ex vivo tissue imaging using chromatic confocal endomicroscopy
To demonstrate high lateral resolution of our fiber-optic 3D chromatic confocal endomicroscopy, ex vivo imaging of fresh gastric tissue resected from rat was performed. The cell morphology and gland structure were clearly visualized (Fig. S1). Furthermore, human tissue imaging was conducted. The imaging results of the fixed human stomach tissue were compared with histological image with hematoxylin and eosin (H&E) staining. As shown in Fig. 5(a) and Fig. 5(c), representative images of a fixed normal human stomach tissue section and cancerous tissue section were obtained, respectively. And the corresponding H&E histological results of the adjacent tissue section were shown in Fig. 5(b) and Fig. 5(d). As shown in the magnified images (Fig. 5(e) and (f)) of (g)-(h) in Fig. 5(a) and (i)-(j) in Fig. 5(b), respectively, structural features of human stomach tissue were visible in the endomicroscopic images. On the images, the signal from the cell nucleus was dark and the signal from the cell membrane was bright, thus forming a natural contrast without labeling. The signal intensity of cancerous tissues was significantly stronger than that of normal tissues, and the difference in characteristic signals was greater than that of normal tissues. In addition, the cell morphology in normal tissues was relatively regular, but disordered in tumor tissues. It demonstrated the potential of in vivo pathology of the fiber-optic 3D chromatic confocal endomicroscopy.
Fig. 5. Comparison of the lateral plane image acquired by the fiber-optic 3D chromatic confocal endomicroscopy with the histology image of hematoxylin and eosin (H&E) staining. (a) Representative image of a fixed normal human stomach tissue section by the 3D chromatic confocal endomicroscopy. (b) H&E staining of the adjacent tissue section to the one shown in (a). (c) Representative image of a fixed human stomach cancerous tissue section by the 3D chromatic confocal endomicroscopy. (d) H&E staining of the adjacent tissue section to the one shown in (c). The red boxes point out the areas in confocal endomicroscopic images (a, c) which were similar to those shown in the histology images (b, d). (e-f) High-magnification confocal endomicroscopic images of (g-h) in (a) and (i-j) in (c), respectively, and yellow arrows mark the cell. Scale bars in (a-d), 100 µm. Scale bars in (e, f), 20 µm.
To demonstrate 3D imaging capability, normal and cancerous tissues of human stomach which were surgically resected were used for ex vivo imaging and 3D stack images were obtained (see Visualization 2 and Visualization 3). As shown in Fig. 6(a), the imaging depth was 220 µm and different depths in representative X-Y projection (MIP mode) of normal human stomach tissue were represented by different colors. And images of the human normal stomach tissue at different depths of 20–220 µm revealed multi-depth information (Fig. 6(b)). The tissue structures changed with different depths, and cells (marked by yellow arrows) were obviously visible. Furthermore, on X-Z and Y-Z section images (Fig. 6(a)) (axial view of tissue), structures similar to gland duct in X-Z and Y-Z section images were obviously visible. The yellow circles mark the lateral view of gland duct in Fig. 6(b) and the axial view of gland duct in X-Z and Y-Z section images in Fig. 6(a). Furthermore, as shown in Fig. S2, the middle of the glands (marked by yellow circles) resembling a circular black hole was a round dark crypt-opening (CO). The edge of the crypt-opening was the marginal crypt epithelium (MCE), which was mainly the cell plasma, and its signal intensity was high. The part between the two crypt-openings was the intervening part (IP), which signal intensity was lower than the MCE. The dark circles around the glands were the subepithelial capillary (SEC) with low signal intensity.
Fig. 6. Ex vivo normal tissue imaging by fiber-optic 3D chromatic confocal endomicroscopy. (a) Representative X-Y projection (MIP mode) of normal human stomach tissue with depth range of 0–220 µm. Different depths were represented by different colors. (b) XY cross-section images of human stomach tissue shown in (a) at different depths, cells (yellow arrows) and glandular duct (yellow circles) were visible. Scale bars, 50 µm.
As shown in Fig. 7(a), 3(D) imaging as conducted for human stomach cancerous tissue. The depth of 3D images of the cancerous tissue was 200 µm and different depths in its representative X-Y projection (MIP mode) were represented by different colors. Images of the human stomach cancerous tissue at different depths were shown (Fig. 7(b)). Compared with the normal tissue with relatively regular structure in Fig. 6, the tissue structure of cancer was irregular. As shown in Fig. S3, the size and shape of the gland ducts of cancerous stomach tissue were irregular. The yellow circle in Fig. S3 showed the duct at different depths. And the blue circles were suspected to point out a gland duct with a gap, because there was fusion of the gland duct in the tumor tissue. And from the reconstructed X-Z and Y-Z axial views of the tissue (Fig. 7(a)), there were no obvious duct structure. These results demonstrated the 3D imaging capability of the endomicroscopy to visualize the cell morphology and structures at multi-depth. And by comparing the 3D imaging results of the normal and cancerous stomach tissue (Fig. S4), we can see the differences of cell morphology and gland structure between normal and cancerous stomach tissue. It demonstrated that our endomicroscopy has the potential for optical diagnosis.
Fig. 7. Ex vivo cancerous tissue imaging by fiber-optic 3D chromatic confocal endomicroscopy. (a) Representative X-Y projection (MIP mode) of human stomach cancerous tissue with depth range of 0–200 µm. Different depths were represented by different colors. (b) XY cross-section images of human stomach tissue shown in (a) at different depths. Scale bars, 50 µm.
3. Discussion and conclusion
We have developed a novel fiber-optic chromatic confocal endomicroscope with capabilities of label-free, large depth, fast 3D imaging speed and high resolution. Its large imaging depth of 570 µm in 3D phantom and 220 µm in biological tissue provides the potential to determine tumor invasion depth. And owing to its fast 3D imaging speed, it has the potential to realize real-time optical imaging and diagnosis in vivo. Meanwhile, with the high lateral resolution, cell morphology and tissue structure of human tissue are visible to provide important morphological information for diagnosis without any labeling. These results demonstrated that the fiber-optic 3D chromatic confocal endomicroscopy has great potential for early diagnosis of gastric cancer. Taken together, this endomicroscopy has the advantages as discussed below.
First, the imaging depth of this fiber-optic 3D chromatic confocal endomicroscopy method enables its potential for diagnosis of gastric cancer invasion depth. In clinical applications of diagnosis of early gastric cancer, once the tumor has invaded deep into the submucosa, it will be at high-risk of Lymph node metastasis. Thus, it is necessary to obtain information from the submucosa of the tissue [3,25]. By using a miniature objective lens with large chromatic dispersion to focus the broad-spectrum excitation light of different wavelengths at different depths of the tissue, we have achieved an imaging depth of 570 µm in 3D phantom. We have also demonstrated ex vivo imaging of human gastric tissues with imaging depth of 220 µm. The major reason for the difference between the imaging depth in tissue and chromatic focal shift was the scattering and absorption of biological materials. And the low object space NA of the miniature objective lens further resulted in the low collection efficiency of reflected signals from the tissue. If we can increase object space NA of the miniature objective lens to improve the light collection efficiency, we may get some improvements. These results suggest that our method has the potential to provide sufficient imaging depth to assess tumor invasion depth of early gastric cancer if we make further improvements in lens design.
Second, the high 3D-imaging speed of our chromatic confocal endomicroscopy method has the potential for future application in optical diagnosis. A high-speed spectrometer was used to acquire confocal signals simultaneously from multi-depths. The 3D stack was then reconstructed by synchronizing the fast signal acquisition along the axial direction with 2D lateral scanning. We have eventually achieved a 3D volumetric imaging speed of 1.5 Hz, which shows a great improvement compared to conventional confocal endomicroscopy [8,26].
Third, high lateral resolution of this fiber-optic 3D chromatic confocal endomicroscopy meets the clinical needs for observation of cell morphology and structure. In our study, a lateral resolution of 2.3 µm has been achieved via confocal scheme. Specifically, a multi-core single-mode fiber bundle was used at the proximal end of a miniature objective lens for both delivery of excitation light and collection of reflectance signals from the sample. And a single-mode fiber was used for signals detection so that our endomicroscopy can provide high lateral resolution. With such resolution, cell structures within the thick tissue are visible, which shows a possibility to distinguish normal and cancer tissues.
Besides, for fiber scanning probe, piezoelectric tube (PZT) or micro electromechanical system (MEMS) can be used to drive the vibration of a single-mode fiber for two-dimensional scanning [27]. And these devices must be assembled in endoscopy probe. Alternatively, a fiber bundle containing numbers of single-mode fiber cores can be used to transmit light. And galvo mirrors or other scanning device are utilized at the proximal end of the fiber bundle to achieve two-dimensional scanning [28,29]. Thus, we chose the method using a fiber bundle containing numbers of single-mode fiber cores to minimize the probe size because there is no need to assemble a scanning mechanism in probe.
However, there still remain some limitations for translation. The diameter of the lens was approximately 11.5 mm and the size of probe housing containing lens and fiber bundle was 17 mm-diameter. We will solve the problem in future in vivo application to ensure that the probe can be combined with clinical commercial endoscopes.
In conclusion, we have developed a novel approach of fiber-optic 3D chromatic confocal endomicroscopy with capabilities of large depth, fast 3D imaging speed and high resolution. This study paves the way for its application in the diagnosis of gastric cancer.
4. Methods
4.1 Broadband light excitation
A broadband light source (SC-5, 470–2400 nm, YSL) was utilized to achieve broadband light excitation. A near-infrared filter (FF01-1010/SP-25, Semrock) was used to select light of 400–1000 nm. And the beam was expanded by telescope (L1, #49–360, f=100 mm, Edmund; L2, #49–370, f=200, Edmund) and split by a beamsplitter (BS, BSW26, 50/50, Thorlabs). Then the broadband light source was coupled into a fiber bundle with 50000 cores (FIGH-50-1100N, fiber diameter=1100 µm, Fujikura). In order to increase detection depth, a customized miniature objective lens (Golden Way Scientific) with large chromatic focal shift was selected as the objective lens of the endomicroscopy. This design enabled simultaneous multi-wavelength excitation by focusing light with different wavelengths into different depths of tissue.
4.2 Laser scanning
A pair of galvo mirrors (6200H 673 5 mm System, Cambridge) was utilized at the proximal end of the fiber bundle to achieve two-dimensional scanning. The light source was coupled into the fiber bundle using a microscope objective lens with NA of 0.5 and magnification of 20 (UMPLFLN20XW, 20X, 0.5, Olympus). At the proximal end, raster scanning was generated by applying a voltage to the x-axis and y-axis galvo mirrors. Both the x-axis and y-axis driven voltage signals were sawtooth waves. The dwell time for each pixel was 8 µs. The scanning area was 256 pixels (x-axis) * 256 pixels (y-axis) for each scan.
4.3 Probe assembly
The endomicroscopic probe consisted of three parts: fiber bundle, miniature objective lens and probe housing. The fiber was 1100 µm in diameter. Each core was approximately 2.5 µm in diameter with core-to-core spacing of 2.5 µm. The miniature objective lens was 11.5 mm in diameter and 37.5 mm long. The probe housing was made of polyetheretherketone (PEEK), a non-conductivity and rigid material. During the assembly, the fiber bundle with a fiber holder was fixed at the proximal end of the miniature objective lens through the probe housing by ultraviolet curing glue with a distance of 19 mm. Besides, the fiber bundle was fixed coaxial with the miniature objective lens, as shown in Fig. 1. And liquid matching was performed at both end of the fiber bundle to eliminate background and pixelation. The liquid was transparent and non-toxic fiber gel with a refractive index of 1.46 and effective wavelength >350 nm.
4.4 Multi-depth data acquisition
A high-speed spectrometer (CS800-800/300-250-OC2K, Wasatch Photonics) with 2048-pixel complementary metal-oxide-semiconductor (CMOS) monochrome line-scan sensor was used for simultaneous acquisition of multi-depth signals. Confocal reflectance signals reflected from different depths of the sample were coupled into the high-speed spectrometer through a single-mode fiber and dispersed to different pixels by a grating inside the spectrometer (Fig. 1). The single-mode fiber acted as a confocal pinhole and the coupling lens was an objective lens with NA of 0.75 and magnification of 20 (UPLSAPO20X, 0.75, Olympus).
Home-built LabVIEW based control software output all synchronous signals of system. All control signals were generated by the data acquisition (DAQ) card PXIe-6358 (1.25 MS/s, National Instrumentation, Texas). The pixel trigger digital output (DO), frame scan and line scan analog output (AO) were synchronized using the counter pulse output mechanism. Specifically, as shown in Fig. 8(a), the frame scan and line scan AO signals drove the galvo scanners within imaging, and the external pixel trigger DO signal synchronized the data acquisition of spectrometer. The exposure time for each 256*256 pixel was 8 µs with spectrometer data acquisition of 125k Hz. Through scanning the laser focus, the system probed the scattering signals of a 3D stack. A chromatic intensity array with 2048 pixels reflected the axial scattering signals of sample (Fig. 8(b)). Additionally, we could acquire multiple consequent chromatic stacks. As shown in Fig. 8(c), we also averaged the intensity of multiple stacks to increase the signal to noise (S/N) of images. Furthermore, the scanning and data acquisition mode of system were flexible with pixel size (256*256, 256*2048, 2048*2048). Through synchronous mechanism of Fig. 8(a), we acquired chromatic pixel-based intensity array of a stack image. By using the home-built image reconstruction program by MATLAB, we reshaped the intensity array by customized pixel size (256*256), and got 256*256*2048 pixel-based intensity matrix of a stack image. Through binning the pixels to 256*256 from 256*2048/2048*2048, we also could alternatively increase the pixel exposure time from 8 µs to 64 µs/512 µs, as shown in the Fig. 8(d). In this way, simultaneous multi-depth imaging was achieved.
Fig. 8. Diagram of multi-depth data acquisition. (a) Timing diagram of synchronous scanning and imaging of chromatic focal endoscopy. The frame scan and line scan analog output (AO) signals drove the galvo scanners within imaging. The external pixel trigger digital output (DO) signal synchronized the data acquisition of spectrometer. (b) Schematic of a chromatic 3D stack reconstruction. (c) Workflow of image reconstruction program and a typical image. (d) Workflow of image reconstruction with customized pixel size (Mode 1: 256*256, Mode 2: 256*2048, Mode 3: 2048*2048).
4.5 Data processing
In data processing, home-built MATLAB program was firstly used to reconstruct 3D images from the spectral data. Then, ImageJ was used to apply a gaussian blur (sigma=1) to solve the problem of fiber pixelation, and background subtraction was performed with a rolling bar with a radius of 50 pixels.
4.6 Materials and tissue preparation
For depth calibration, each groove of the silicon wafer was 50 µm wide. For depth measurement, the 3D phantom was made by mixing 10 µm-diameter Al beads in 2.5% agarose gel.
Regarding to 3D imaging of rat stomach tissue, the rats were anesthetized and dissected before imaging, and the stomach tissues were harvested, washed with phosphate buffered saline (PBS) and then the fresh tissues were fixed on the foam board with a pin with the mucosa side facing upwards. The stomach tissues were cut into approximately 2 cm*2 cm*3 mm.
Regarding to 3D imaging of human stomach tissue, the tissues were obtained from patients undergone gastrectomy. The tissues were then washed with PBS and then the fresh tissues were fixed on the foam board with a pin with the mucosa side facing upwards. The average size of cancer tissues was about 5 mm*5 mm*3 mm, and the average size of normal tissues was about 2 cm*2 cm*3 mm.
Regarding to 2D imaging of human stomach tissue section, the tissues were firstly fixed with formalin. After 24 hours of fixation, the tissues were immersed in a 30% sucrose solution for 24 hours for dehydration. After dehydration, the tissues were embedded with OCT embedding agent, then were made to frozen sections and hematoxylin and eosin (H&E) stained sections on a slide glass. The thickness of these tissue sections for imaging was 60 µm.
This study was approved by an institutional review board. Human stomach tissues were obtained from Department of Gastroenterology in Peking University Third Hospital. During imaging, the tissues were placed on the motorized stage.
4.7 Histopathology
The tissues were sectioned along the axial direction and stained with H&E. Then, two professional pathologists examined the H&E slides independently.
Funding
Beijing Municipal Natural Science Foundation (No. L172011); National Natural Science Foundation of China (No. 91959120No. 62027824; Basic Research Program for Beijing-Tianjin-Hebei Coordination (No. 19JCZDJC65500(Z)); Open Project Program of Wuhan National Laboratory for Optoelectronics (NO.2018WNLOKF026); Fundamental Research Funds for the Central Universities (No. YWF-21-BJ-J-549); China Postdoctoral Science Foundation (No.2020M670100).
Acknowledgments
We acknowledge Prof. Linhao Li from School of Biological Science and Medical Engineering at Beihang University for providing us the silicon wafer to perform imaging depth calibration.
Disclosures
The authors declare that they have no conflict 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.
Supplemental document
See Supplement 1 for supporting content.
References
1. J. Choi, S. G. Kim, J. P. Im, J. S. Kim, H. C. Jung, and I. S. Song, “Endoscopic prediction of tumor invasion depth in early gastric cancer,” Gastrointest. Endosc. 73(5), 917–927 (2011). [CrossRef]
2. H. Ono, K. Yao, M. Fujishiro, I. Oda, S. Nimura, N. Yahagi, H. Iishi, M. Oka, Y. Ajioka, M. Ichinose, and T. Matsui, “Guidelines for endoscopic submucosal dissection and endoscopic mucosal resection for early gastric cancer,” Dig. Endosc. 28(1), 3–15 (2016). [CrossRef]
3. J. G. C. Association, “Japanese gastric cancer treatment guidelines 2014 (ver. 4),” Gastric Cancer 20(1), 1–19 (2017). [CrossRef]
4. M. H. Ren, X. S. Qi, Y. N. Chu, Y. N. Yu, Y. Q. Chen, P. Zhang, T. Mao, and Z. B. Tian, “Risk of Lymph Node Metastasis and Feasibility of Endoscopic Treatment in Ulcerative Early Gastric Cancer,” Ann. Surg. Oncol. 28(4), 2407–2417 (2021). [CrossRef]
5. A. D. Wagner and C. Dejaco, “Development and assessment of morphologic criteria for diagnosing gastric cancer using confocal endomicroscopy: an ex vivo and in vivo study,” Endoscopy 38(9), 886–890 (2006). [CrossRef]
6. J. E. East, J. L. Vleugels, P. Roelandt, P. Bhandari, R. Bisschops, E. Dekker, C. Hassan, G. Horgan, R. Kiesslich, G. Longcroft-Wheaton, A. Wilson, and J. M. Dumonceau, “Advanced endoscopic imaging: European Society of Gastrointestinal Endoscopy (ESGE) Technology Review,” Endoscopy 48(11), 1029–1045 (2016). [CrossRef]
7. S. W. Paddock, “Principles and practices of laser scanning confocal microscopy,” Mol. Biotechnol. 16(2), 127–150 (2000). [CrossRef]
8. R. Kiesslich, M. Goetz, A. Hoffman, and P. R. Galle, “New imaging techniques and opportunities in endoscopy,” Nat. Rev. Gastroenterol. Hepatol. 8(10), 547–553 (2011). [CrossRef]
9. C. I. Unglert, E. Namati, W. C. Warger, L. Liu, H. Yoo, D. Kang, B. E. Bouma, and G. J. Tearney, “Evaluation of optical reflectance techniques for imaging of alveolar structure,” ” J. Biomed. Opt. 17(7), 071303 (2012). [CrossRef]
10. H. Yoo, D. Kang, A. J. Katz, G. Y. Lauwers, N. S. Nishioka, Y. Yagi, P. Tanpowpong, J. Namati, B. E. Bouma, and G. J. Tearney, “Reflectance confocal microscopy for the diagnosis of eosinophilic esophagitis: a pilot study conducted on biopsy specimens,” Gastrointest. Endosc. 74(5), 992–1000 (2011). [CrossRef]
11. C. Carrera, S. Puig, and J. Malvehy, “In vivo confocal reflectance microscopy in melanoma,” Dermatol. Ther. 25(5), 410–422 (2012). [CrossRef]
12. A. Levine and O. Markowitz, “Introduction to reflectance confocal microscopy and its use in clinical practice,” JAAD Case Rep 4(10), 1014–1023 (2018). [CrossRef]
13. C. Olsovsky, R. Shelton, O. Carrasco-Zevallos, B. E. Applegate, and K. C. Maitland, “Chromatic confocal microscopy for multi-depth imaging of epithelial tissue,” Biomed. Opt. Express 4(5), 732–740 (2013). [CrossRef]
14. M. A. Tadayon, S. Chaitanya, K. M. Martyniuk, J. C. McGowan, S. P. Roberts, C. A. Denny, and M. Lipson, “3D microphotonic probe for high resolution deep tissue imaging,” Opt. Express 27(16), 22352–22362 (2019). [CrossRef]
15. K. Shi, P. Li, S. Yin, and Z. Liu, “Chromatic confocal microscopy using supercontiuum light,” Opt. Express 12(10), 2096–2101 (2004). [CrossRef]
16. M. Rayer and D. Mansfield, “Chromatic confocal microscopy using staircase diffractive surface,” Appl. Opt. 53(23), 5123–5130 (2014). [CrossRef]
17. J. Garzón, T. Gharbi, and J. Meneses, “Real time determination of the optical thickness and topography of tissues by chromatic confocal microscopy,” J. Opt. A: Pure Appl. Opt. 10(10), 104028 (2008). [CrossRef]
18. S. Li and R. Liang, “DMD-based three-dimensional chromatic confocal microscopy,” Appl. Opt. 59(14), 4349–4356 (2020). [CrossRef]
19. Q. Cui and R. Liang, “Chromatic confocal microscopy using liquid crystal display panels,” Appl. Opt. 58(8), 2085–2090 (2019). [CrossRef]
20. M. Zint, K. Stock, R. Graser, T. Ertl, E. Brauer, J. Heyninck, J. Vanbiervliet, S. Dhondt, P. D. Ceuninck, and R. Hibst, “Development and verification of a novel device for dental intra-oral 3D scanning using chromatic confocal technology,” Proc. SPIE 9313, 93130U (2015). [CrossRef]
21. P. Lane, R. Elliott, and C. MacAulay, “Confocal microendoscopy with chromatic sectioning,” in SPIE BiOS (2003).
22. D. K. Kang, R. W. Carruth, M. Kim, S. C. Schlachter, M. Shishkov, K. Woods, N. Tabatabaei, T. Wu, and G. J. Tearney, “Endoscopic probe optics for spectrally encoded confocal microscopy,” Biomed. Opt. Express 4(10), 1925–1936 (2013). [CrossRef]
23. K. Carlson, M. Chidley, K. B. Sung, M. Descour, A. Gillenwater, M. Follen, and R. Richards-Kortum, “In vivo fiber-optic confocal reflectance microscope with an injection-molded plastic miniature objective lens,” Appl. Opt. 44(10), 1792–1797 (2005). [CrossRef]
24. J. Wang, H. Li, G. Tian, Y. Deng, Q. Liu, and L. Fu, “Near-infrared probe-based confocal microendoscope for deep-tissue imaging,” Biomed. Opt. Express 9(10), 5011–5025 (2018). [CrossRef]
25. P. V. Draganov, A. Y. Wang, M. O. Othman, and N. Fukami, “AGA Institute Clinical Practice Update: Endoscopic Submucosal Dissection in the United States,” Clin. Gastroenterol. Hepatol. 17(1), 16–25.e1 (2019). [CrossRef]
26. Z. Li, T. Yu, X. L. Zuo, X. M. Gu, C. J. Zhou, R. Ji, C. Q. Li, P. Wang, T. G. Zhang, K. Y. Ho, and Y. Q. Li, “Confocal laser endomicroscopy for in vivo diagnosis of gastric intraepithelial neoplasia: a feasibility study,” Gastrointest. Endosc. 72(6), 1146–1153 (2010). [CrossRef]
27. C. M. Lee, C. J. Engelbrecht, T. D. Soper, F. Helmchen, and E. J. Seibel, “Scanning fiber endoscopy with highly flexible, 1 mm catheterscopes for wide-field, full-color imaging,” J Biophotonics 3(5-6), 385–407 (2010). [CrossRef]
28. J. Wang, M. Yang, L. Yang, Y. Zhang, J. Yuan, Q. Liu, X. Hou, and L. Fu, “A confocal endoscope for cellular imaging,” Engineering 1(3), 351–360 (2015). [CrossRef]
29. A. Orth, M. Ploschner, E. R. Wilson, I. S. Maksymov, and B. C. Gibson, “Optical fiber bundles: Ultra-slim light field imaging probes,” Sci. Adv. 5(4), eaav1555 (2019). [CrossRef]
