Confocal laser endomicroscopy (CLE) is a sort of endoscopic imaging technique that allows real-time histological examination of tissue in the body [1]. The first commercial device in the world for this purpose was launched in 2006 [2]. This instrument adopted the CLE using a probe with a single optical fiber and piezoelectric translator (PZT) to scan in two dimensions (2D). This method could produce highly detailed images at megapixel resolutions [3]; However, it posed a risk of high voltage, which was inherent in products that enter the human body, owing to the use of PZT scanning in the probe [4]. In 2005, a new instrument that adopted a new method without the dangerous high-voltage system received approval from the U.S Food and Drug Administration [5]. This model used a fiber bundle imaging technique known as one of the CLE techniques [6], which uses a bundle of optical fibers instead of a single optical fiber with PZT to acquire image information, which is then processed by an external imaging device to obtain a 2D image of suspected lesions in the body. Since the probe contained only fiber bundles and lenses, there was no risk from the use of high voltage in the body. In addition, the bundle imaging type had the advantage of faster image acquisition owing to its use of a high numerical aperture (NA) [7].

An endoscope is a medical device that is inserted into a human or animal body to diagnose inflammation and, if necessary, remove or collect abnormal growths such as tumors that have formed inside the body. It is an optical system that allows wide-angle viewing of the internal organs and is used for noncontact imaging. The distal end of the endoscope is composed of instrument channels that can accommodate a nozzle, objective lens, illumination source, and other devices. Its diameter ranges from approximately 5.8 mm to 9.9 mm depending on the manufacturer and purpose.

Laser endomicroscopy is used to obtain precise images of lesions identified in wide-angle images from an endoscope by inserting an optical system into the instrument channel of the endoscope and contacting the probe with the biological tissue to acquire images in a narrow field of view (FOV). Some important examples of laser endomicroscopy include optical coherent tomography [8], multiphoton endomicroscopy [9], and CLE [10]. That is, CLE is a kind of laser endomicroscopy using a single fiber or fiber bundle to transfer fluorescence images radiated in lesions illuminated by laser light to a sensor.

Among all the types of CLE that rely on fluorescence imaging to determine the presence or absence of abnormalities in the lesion, the bundle imaging type CLE began to be used in earnest in Korea after its medical insurance pricing was finalized in 2019. This CLE uses a spatial filter to remove noise at a conjugate point where the laser light is irradiated into the biological tissue, and the fluorescence image is radiated from the subject at the suspected lesion. However, due to the volumetric nature of the fluorescence generated in biological tissues, the fluorescence emitted from different depths near the target point can become superimposed, resulting in a blurred image. This can be overcome by installing a spatial filter at the conjugate point, which removes the superimposed fluorescence and improves the axial resolution for more accurate measurements [11].

In general, when more detailed examination is needed after a conventional endoscopy, the affected tissues are collected and sent to the pathology department for examination under a microscope. The tissues are thinly sliced, in the depth direction, to facilitate the observation of dysplastic cells. The time required to determine the presence of cancer according to these procedures is a minimum of several days [12]. However, when CLE is used, it is possible to determine the presence of gastric cancer in real time by observing the tissue cells of the lesion during a gastric cancer screening procedure at a level comparable to that of traditional histopathological methods [13]. Additionally, early-stage cancer cells can be removed on the spot through endoscopic resection. Thus, it is possible to reduce time and costs by minimizing the various procedures required for the diagnosis of gastric cancer.

The objective lens used in CLE, as described above, has been the subject of active research recently. However, owing to the high design complexity, it is often challenging to use a large number of lenses or apply the lens system to products when the size of the lens system is large. Moreover, it is mostly challenging to produce a large quantity of products when using aspherical glass lenses or cemented achromatic lenses. For example, in order to reduce aberrations, nine glass lenses were used with a diameter of 3 mm and a long optical total track [14]. The optical total track is the length of the optical axis along which light originates in the tissue and reaches the fiber, as shown in Fig. 1. If it exceeds 10 mm, it is very difficult to apply clinically. However, seven glass lenses were used instead. When applying a combination of glass spherical lenses and cemented lenses [15], only five plastic lenses were used when applying aspherical plastic lenses. However, when using glass cemented lenses [16], five to six plastic lenses were used. When the optical total track exceeded 10 mm, as seen in cases [17, 18], it was considered necessary to develop a compact objective lens, while also increasing the productivity and satisfying the imaging performance.

Figure 1. Schematics of CLEs of the PZT scanning and bundle imaging types. CLE, confocal laser endomicroscopy; PZT, piezoelectric translator.

In this paper, we propose a design using one spherical glass lens and four aspherical plastic lenses to create an air-spaced achromatic-type lens. This design has both a compact size and satisfactory imaging performance, while ensuring productivity and ease of assembly. Furthermore, we aim to evaluate whether the objective lens developed and produced for CLE, along with the developed resolution measurement device, meets the required resolution performance for clinical use.

The CLE system uses 488-nm laser light, which is directed into the fiber through reflection from a dichroic mirror, as shown in the right optical configuration of Fig. 1. The light emitted from the fiber is then focused onto the skin lesion using an objective lens installed in the dotted red square, as shown in the left central device of Fig. 1. Fluorescence occurs over a wide area of the lesion that is being examined, which is then captured by the objective lens. The light re-enters and travels along the same optical path and enters the detector after transmission through the dichroic mirror. In the case of PZT scanning using the upper left device in Fig. 1, a PZT module installed inside the objective lens generates a 2D image. On the contrary, in the case of fiber bundle imaging adopting the lower left device in Fig. 1, a 2D scanner installed in the main equipment is used to acquire a 2D image. Therefore, both types of CLE can share the same specifications of the endomicroscopic objective lens.

Human tissue is composed of more than 70% water; Therefore, when the CLE contacts the lesion, the object space can be optically assumed to be water immersion. Furthermore, the object space NA is set to 0.55 to detect objects of approximately 1 μm in size using the objective lens. The image space is configured with a maximum image circle diameter of 600 μm to ensure compatibility with either the PZT scanning or the bundle imaging method in the CLE. The image space NA is set to 0.22. As a result, the magnification of the optical system is 2.5×, and the FOV on the object space is 240 μm. Furthermore, considering the laser fluence, the suitable skin penetration depth is generally 40 μm–70 μm; Therefore, 60 μm is used as the thickness of the object [10]. In addition, the optical total track should be less than 8.0 mm, and the lens outer diameter should be 2.0 mm to accommodate the instrument channel of the microscope. One spherical glass lens and four aspherical plastic injection lenses are used. In the CLE to be designed, the lateral resolution d_xy and axial resolution d_z of the focal radius can be calculated using Eqs. (1) and (2), respectively, by using the predetermined wavelength and NA_obj, which is the NA of the object. The calculated lateral resolution is 0.61 μm and the axial resolution is 4.85 μm, based on an NA of 0.55 at the target wavelength of 550 nm, which meets the desired specifications. Thus, it can be confirmed that the desired size of approximately 1 μm is decomposable.

Here, λ represents the central wavelength being used and η represents the refractive index of the immersion materials. The modulation transfer function (MTF) is designed with the goal of achieving the diffraction limit determined by the NA. The desired value of the MTF is 0.2 or higher at 500 cycles/mm.

The lens specifications determined through theory and experiments are summarized in Table 1, and the optical design was optimized using commercial optical design software (CODEV ver. 11.3; Synopsys, CA, USA) with the goal of achieving optimal results, as shown in Fig. 2. A highly refractive material, N-LASF31A (Schott AG, Mainz, Germany), was used because the incident angle of the marginal rays entering the first lens surface with NA of 0.55 was relatively large. Furthermore, it was determined that the first lens should be made of glass rather than plastic, since it needed to withstand direct contact and friction with the skin surface of the lesion while maintaining durability during cleaning. For the remaining lenses, plastic was used to enable the rapid production of a large quantity of lenses, since they did not require durability. In order to design an air-spaced achromatic lens, two types of plastic injection materials were used: OKP4HT-L (Doosung Chemis, Incheon, Korea), which possessed characteristics of the flint series, and Zeonex® E48R (Zeon Co., Tokyo, Japan), which possessed characteristics of the crown series. Furthermore, by symmetrically arranging the refractive powers of the lenses with respect to the aperture stop and by using aspheric surfaces, reductions were made for spherical aberration, chromatic aberration, distortion, and Petzval aberration. To reduce the differences in the sizes of objects in the depth direction, a telecentric was applied to the object space. Additionally, telecentric design was applied to the image space to minimize the difference in light intensity caused by the angle difference of the incident chief rays on the fiber bundles for each FOV. The design data for the objective lens of CLE, which was designed in this way, and the aspheric coefficients for the aspheric lenses are presented in Tables 2 and 3, respectively. In Table 3, K represents the conic constant, and A4, A6, A8, A10, and A12 are the aspheric coefficients corresponding to their respective orders.

Figure 2. 2D layout of the designed object lens.

TABLE 1. Optical design specification.

Specification	Design Values
Wavelength (nm)	488 to 550
Tissue NAa)	0.55 (Water Immersion)
Fiber NAa)	0.22
Magnification	2.5×
Field of View (μm)	240
Maximum Image Circle Diameter (μm)	600
MTFb)	≥0.2 @ 500 cycles/mm, All Field
Object Thickness in Water (μm)	60
Optical Total Track (mm)	≤ 8.0
Lens Diameter (mm)	≤ 2.0
Number of Lenses	≤ 5 ea

^a)NA, numerical aperture; ^b)MTF, modulation transfer function..

TABLE 2. Lens design data.

Surface	Radius	Thickness	Material	Diameter
Object	−12.5	0.06	Seawater	-
1	Infinity	0.90	N-LASF31A	2.0
2	−1.09	0.26	-	2.0
3	2.56	0.80	OKP4HT	2.0
4	0.75	0.10	-	2.0
5	0.88	1.10	Z-E48R	2.0
Stop	−1.12	0.66	-	2.0
7	0.91	1.00	Z-E48R	2.0
8	2.11	0.29	-	2.0
9	−0.75	1.70	OKP4HT	2.0
10	−1.10	0.80	-	2.0
Image	Infinity	-	-	0.6

TABLE 3. Modulus of aspherical lens surface.

Object	Ka)	A4b)	A6c)	A8d)	A10e)	A12f)
	−1.8e + 05	−5.23	1.81e + 03	−4.035e + 04	−1.041e + 06	2.66e + 07
3	0	−0.81	−1.57	6.51	−24.60	0
4	0	−1.20	−0.36	2.77	−5.07e + 03	0
5	0	−0.59	−0.13	1.06	−1.26	0
Stop	0	−0.19	0.69e + 03	−0.88	0.71	0
7	0	−0.53	0.31	−0.32	0.58	0
8	0	−2.07	4.20	−5.39	4.15	0
9	−5.73	−2.67	9.61	−19.13e + 03	19.4	0
10	−17.54	−1.14	4.82	−12.45	13.98	0

^a)K, conic constant; ^b)A4, 4^th aspheric coefficient; ^c)A6, 6^th aspheric coefficient; ^d)A8, 8^th aspheric coefficient; ^e)A10, 10^th aspheric coefficient; ^f)A12, 12^th aspheric coefficient..

The arrangement of the lenses with one spherical glass lens and four aspherical plastic lenses is depicted in Fig. 2. The designed magnification of this objective lens is 2.5×, with a FOV of 240 μm and calculated upper image circle diameter of 600 μm. Furthermore, it can be seen that all the specifications for the lens system presented in Table 1 were met, as the optical total track from the object to the image plane is 7.67 mm. Also, to allow sufficient light to pass through a lens with an outer diameter of 2.0 mm, the clear aperture of the largest lens was designed to be 1.7 mm.

Figure 3(a) shows an MTF chart of the designed objective lens, with values up to 520 cycles/mm. The wavelengths used in analysis are 488, 530, and 550 nm. The black alternating long and short dashed line represents the diffraction limits, and the red solid line represents the optical axis. The fields described next are based on half size from the optical axis. The green solid line represents the tangential field of 70 μm, while the green alternating long and short dashed line represents the radial field of 70 μm. The blue solid line represents the tangential field of 100 μm, while the blue alternating long and short dashed line represents the radial field of 100 μm. The brown solid line represents the tangential field of 120 μm, while the brown alternating long and short dashed line represents the radial field of 120 μm. Based on the convergence of the diffraction limit and the marked lines of all fields, it is concluded that aberration correction through optimization is successfully achieved. Numerically, it is confirmed that the modulation index of all fields exceeds 0.2, which corresponds to the desired resolution of 1 μm or 500 cycles/mm. Additionally, for the last field, a modulation of 0.24 was observed.

Figure 3. Diffraction modulation transfer function (MTF) chart: (a) Modulation vs. spatial frequency, (b) MTF through focus chart of 500 cycles/mm.

Furthermore, an analysis of the depth of focus was conducted using the MTF through focus chart, as shown in Fig. 3(b). The spatial frequency used for the analysis was 500 cycles/mm, and it was applied with the same wavelength as in the MTF chart. The field for each indicated color was consistent with the MTF chart, with all fields maintaining a modulation index of approximately 0.2 within the range of −2.5 μm to + 2.5 μm. In addition, it was predicted that the imaging performance of all fields would not differ significantly in the assembled state since the modulation index peaks of all fields were concentrated at the center.

The aforementioned results suggest that the astigmatic aberration and field curvature are also minimal. This was further confirmed in Fig. 4(a). In this graph, the horizontal axis represents the deviation from the focal point, while the vertical axis represents the field. The ray denoted by “S” represents the sagittal ray, while the ray denoted by “T” represents the tangential ray. It can be seen that the sagittal and tangential rays move in the same direction depending on the field, and the difference between these two rays, which indicates astigmatic aberration, is less than 10 μm. Furthermore, when analyzing the distortion as shown in Fig. 4(b), the highest distortion of approximately 0.74% is observed at the field position of 80 μm based on the half-value criterion. In the case of the last field, a distortion of approximately 0.51% is observed.

Figure 4. Aberration plots of the designed objective lens: (a) Astigmatic field curves, (b) distortion, and (c) spot diagram.

Furthermore, Fig. 4(c) represents a graph depicting a spot diagram. The four fields are represented based on the half-value axis, with respective values of 0, 70, 100, and 120 μm. The calculated spot colors for each fields are represented as red at 550 nm, green at 530 nm, and blue at 488 nm. The black circles indicated in the field are approximately 3 μm in size and represent Airy disks. However, it is possible to observe spot sizes that are much smaller than this, as indicated by the root mean square (RMS) spot size. It can be seen that the largest spot size at the last field is 1.18 μm by the RMS value.

After confirming that the performance of the objective lens meets the requirements, the designed objective lens data were used to create a 3D model of the assembled barrel using commercial 3D software (SolidWorks; Dassault Systèmes, Vélizy-Villacoublay, France), as shown in Fig. 5. To minimize the costs and tolerance management factors involved in the design, production, and assembly processes, lenses 1 and 2 were spaced apart using the barrel body itself, while a spacer was applied only between lenses 3 and 4 for spatial adjustment during assembly.

Figure 5. 3D model of the designed objective lens.

In particular, the instrument fabrication and assembly of lenses 2 and 3 and lenses 4 and 5 were designed similar to the design of mobile phone lenses [19], with the optical system and instruments touching the outer parts of the clear aperture so that they could be assembled without separate spacers.

Based on the optical system design of the objective lens for CLE in Fig. 2 and the 3D modelling design in Fig. 5, we designed the optical system for CLE as shown at the top of Fig. 6. According to Tables 2 and 3, one glass lens was machined, and four plastic lenses were injected as shown at the bottom of Fig. 6. The instrument comprised three parts: A spacer, a barrel, and barrel-impression part, and was fabricated using stainless steel for safety in the human body. For assembly, lens 1 was assembled from the left side of the barrel, as shown in Fig. 6, using an adhesive that was bio-certified to ISO 10993-1 [20]. If the seal is not proper, body fluids can enter the barrel. Next, the right side of the barrel was assembled in the order of lenses 2 to 5, and finally, the barrel impression part was used to secure the parts. In the case of the mounted impression part, the shape of the instrument could be changed so that a single optical fiber or a bundle of optical fibers could be mounted according to the CLE method.

Figure 6. Design of the 2D instrument and fabricated parts.

The objective lens for CLE, as assembled above, is depicted in Fig. 7. The total length of the assembly, including the barrel impression part, is 13 mm, and the outer diameter is 2.6 mm, making it compatible with the instrument channel of a commercial endoscope.

Figure 7. Assembled optics.

To verify the MTF of all FOVs at 500 cycles/mm, as shown in Fig. 3(a), a resolution measurement device for CLE, equipped with a sine grid chart, was constructed as depicted in the optical configuration (upper system) and actual arrangement (lower photo) of all devices in Fig. 8. All parts in Fig. 8 correspond to each other by the dotted red lines and dotted red square. The LED used (MWWHL4; Thorlabs, NJ, USA) is a model with a fluorescent coating on the blue light source. As shown in Fig. 9, the wavelength range was from 400 to 800 nm, with a characteristic blue LED center wavelength of 450 nm and a fluorescent center wavelength of 600 nm. For CLEs that used a 488 nm light source, the typical observed wavelength range was 500 nm to 700 nm. Therefore, it was determined that there would be no performance issues if the verified measurement results using an LED with a wider wavelength range met the specifications.

Figure 8. Objective lens resolution measurement system for confocal laser endomicroscopy (CLE).

Figure 9. Spectral distribution of LED used in the measuring instrument (LED, light-emitting diode).

The measurement device used a sinusoidal chart (#55-641; Edmund Optics^®, NJ, USA) with a minimum resolution interval of 256 cycles/mm. To achieve a resolution of 500 cycles/mm at the object position of the objective lens designed for CLE, a relay lens system with a magnification of 0.5× was constructed using a commercially available high-resolution 10× objective lens (Plan APO 10×; Mitutoyo Co., Tokyo, Japan) and a 5× objective lens (Plan APO 5×; Mitutoyo Co.), as shown in Fig. 7. When passing through the configured lens system, a sine grid with a frequency of 256 cycles/mm was reduced by a factor of 0.5×, forming lines on the CLE objective lens surface with a frequency of approximately 512 cycles/mm and width of approximately 1 μm. Furthermore, the sine grid was magnified to 2.5× after passing through the objective lens designed for CLE, as shown in Fig. 7, resulting in a single line of approximately 2.45 μm being formed. To evaluate the MTF of the objective lens, a relay lens system with a magnification of 10×, comprising a 50× objective lens (Plan APO 50×; Mitutoyo Co.), and two 5× objective lenses (Plan APO 5×; Mitutoyo Co.), was installed along with a camera (acA1300-30 μm; Basler AG, Ahrensburg, Germany). The sinusoidal image passing through the objective lens of the CLE is transmitted to the fiber bundle location by the relay lens system. Thus, a single line was ultimately resolved and observed by the camera at a resolution of 24.5 μm. The camera used was equipped with ICX445 CCD sensor (Sony Group Co., Tokyo, Japan). The effective sensor size was 4.8 mm × 3.6 mm, with a resolution of 1,280 pixels × 960 pixels. The size of each pixel was 3.75 μm × 3.75 μm. The magnification of one line captured by the camera, as determined by the relay lens system, was 24.5 μm. Therefore, it could be predicted that when the ideal objective lens was tested, approximately 98 pairs of lines would be formed on the sensor with a width of 4.8 mm. Figure 10 shows an image of a sinusoidal line drawn using the objective lens for CLE, measured using the MTF measurement device of Fig. 8.

Figure 10. Image measured at 512 cycles/mm.

The presence of 98 pairs of lines in the image shown in Fig. 9 validates the implementation of the FOV with the same specifications as the design. In addition, to assess the imaging performance of the objective lens, the MTF for a spatial frequency of 512 cycles/mm was calculated by measuring the pixel values for the on-axis field and four locations on the normalized 0.7 field in the captured image shown in Fig. 10. The pixel value was 8 bits, and it indicated the position of the image. Red 1 represents the top left corner, green 2 represents the top right corner, blue 3 represents the center, yellow 4 represents the bottom left corner, and purple 5 represents the bottom right corner. The higher the pixel value, the brighter the position in the image, and the lower the pixel value, the darker it is. By referring to Fig. 11 and quantitatively analyzing the image, it could be seen that the peak value (P-V) of the line pair consisting of 14 pixels at position 3, corresponding to the on-axis field, was higher than the P-V values at positions 1, 2, 4, and 5, corresponding to the 0.7 field. The value calculated using the Michelson contrast formula for the pixel measurements in Fig. 11 is presented in Table 4. The third value corresponding to the on-axis had the highest MTF value of 0.225. In addition, the upper part of the peripheral area with the 0.7 field had low MTF values, calculated to be 0.155 at position 1 and 0.154 at position 2. The lower part of the peripheral area with the 0.7 field had MTF values calculated to be 0.196 at position 4 and 0.172 at position 5, respectively.

Figure 11. Monochromatic pixel values of line pairs.

TABLE 4. Modulation transfer function (MTF) of the image at 512 cycles/mm obtained from the five image points marked in Fig. 10.

Position	Modulation
1	0.155
2	0.154
3	0.225
4	0.196
5	0.172

For the last field, some parts had MTFs less than 0.1. However, since the lines could be visually identified, the fabricated objective lens was considered to be sufficiently capable of resolving 1 μm for all fields.

Figure 12 shows an image of water-immersed 1-μm standard particles (Duke Standard^TM 4009A; Thermo Fisher Scientific Inc., MA, USA) captured using an objective lens for CLE by placing a preparation on the position of the sinusoidal chart in the measurement device shown in Fig. 8. Despite polystyrene particles having a refractive index of 1.59 at 589 nm with high transmittance, we confirmed 0.994 μm ± 0.015 μm could be resolved across the entire image area.

Figure 12. Image of water-immersed 1-μm standard particles captured at 0.5×. Scale bar: 20 μm.

An objective lens for CLE, which could be attached to the instrument channel of an endoscope, was designed, assembled, and evaluated for MTF using imaging techniques. The wavelengths used were 488 nm for the laser light source and 530 nm to 550 nm for the fluorescent wavelength. The imaging system was designed and fabricated using one spherical lens made of glass and four aspheric lenses made of plastic, considering productivity and ease of assembly. The system was designed and fabricated using an air-spaced achromatic-type design so that it would achieve a FOV of 240 μm and resolution of 1 μm when it came into contact with tissue inside the body. To reduce the toxicity to the human body, stainless steel material was used, and the device was fabricated by minimizing the components to one spacer, one barrel, and one barrel impression part. After assembling the lenses and instruments fabricated in this manner, it was confirmed that the assembly had a diameter of 2.6 mm and length of approximately 13 mm, making it suitable for use in the standard instrument channel of an endoscope. Additionally, a sinusoidal chart with a density of 256 cycles/mm was reduced to 0.5× to create a sine grid with a density of 512 cycles/mm for directly evaluating the MTF of the objective lens for CLE at the object position. When considering the size and resolution of the image sensor, which was tangential to the circular objective lens, the calculated MTF closely matched the theoretical value. In addition, the image’s on-axis field and four locations in the diagonal direction with the 0.7 field were selected. From each location, 14 pixel values corresponding to a pair of lines were extracted, and using the Michelson contrast formula, a minimum MTF of 0.154 and maximum MTF of 0.225 were calculated. For the last field, the MTF was below 0.1, but it was possible to visually identify objects as small as approximately 1 μm. Also, a 1 μm size standard particle preparation was created and placed at the sinusoidal chart position, which was then imaged at 0.5× reduction. The results confirmed that 0.994 μm ± 0.015 μm could be resolved across the entire image area. The asymmetrical modulation index of a 0.7 field in four locations, which was lower than the on-axis field, is believed to be due to the eccentric tolerance during the assembly of ultrasmall lenses. The reason for assembling the components with a stacking method while observing the physical center position using a tool microscope was because a commercial eccentric microscope was not available. In the future, we plan to implement a transmitting eccentric assembler or reflective eccentric assembler for ultrasmall lenses to research an air-spaced achromatic objective lens for confocal laser endomicroscopy with high and uniform MTF performance.

This work was supported by the Technology Innovation Program of Development of AI Base Multimodal Endomicroscope for in situ Diagnosis Cancer (Grant No. 20001533) funded by the Ministry of Trade, Industry & Energy (MOTIE, Korea).

The authors declare no conflicts of interest.

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.