Discriminative spatial localisation in confocal reflectance and carbon dots-based fluorescence imaging using mixed-mode endoscopic scanner

Abstract

Carbon dots (CDs)-based fluorescence imaging with confocal reflectance effectively localises labelled structures with clear sample visualization. Yet, integrating these modalities into single-fibre cantilever-based endoscopic scanners poses challenges. Thus, we present an endoscopic scanner with a mixed-mode fibre cantilever for dual modalities imaging. The cantilever features two-attached fibre construction for multiple illumination and detection channels without additional optical components. The measured lateral and axial resolutions are 5.81 and 6.52 µm, respectively with field-of-view of 408 × 488 µm. The fluorescence detection capability was first verified using fluorescent microspheres and CD-embedded polymer film, followed by confocal reflectance and CDs-based fluorescence imaging on labelled fibrous network and plant samples. The results demonstrate the complementary strengths of confocal reflectance, which extracts surface profiles, and CDs-based fluorescence, which captures fluorescence signals, leading to precise localization of fluorescence signals within the sample. Switching modality schemes at the scanner’s back-end enables implementation of different modalities.

Keywords:

• Lissajous scanning
• confocal reflectance
• fluorescence imaging
• endoscopic scanner
• carbon dots

Introduction

Optical imaging techniques offer non-invasive, high-resolution imaging of biological specimens and disease diagnosis, making them attractive in biomedical fields. These techniques include confocal reflectance, fluorescence, optical coherence tomography (OCT), and non-linear microscopy.^[1] Despite their ubiquity, these imaging techniques are specific and each only provides limited sample information.^[2–4] Confocal reflectance modality provides a high-resolution surface profile of the target sample but is less sensitive to molecular processes. Conventionally, the fluorescence modality has a high sensitivity for detecting molecular activities within the sample. The combination of confocal reflectance and fluorescence modalities has demonstrated spatial localisation of abnormalities with structural visualisation of samples.^[5–10] However, fluorescence imaging using traditional dyes has limited specificity.^[4,11] This limitation is due to each dye having different photostability, varying fluorescence intensity and may be toxic towards the biological sample.

Labelling methods based on fluorescent carbon dots (CDs) is a promising approach to overcome the limitations of traditional dyes. CDs have captivated many researchers since their discovery by chance in 2004.^[12] As zero-dimension carbon-based nanomaterials, CDs have diverse properties and favourable features such as excellent photostability, low toxicity, good biocompatibility, abundant functional groups, tuneable photoluminescence and low-cost processability.^[13–17] They are widely applied in various fields, including photovoltaics,^[18,19] dyes,^[20] fuel cells,^[21] drug delivery,^[22] food safety,^[23] optoelectronics,^[24] optical sensors,^[25,26] and biomedicals.^[27] Numerous investigations have been conducted in recent years to synthesise multiple kinds of CDs with varying functions and tunable ranges for fluorescence imaging to detect specific molecular activities within the sample.^[28–34]

Thus, there is a need to move towards CDs-based fluorescence imaging in conjunction with confocal reflectance to visualise the surface profile of target samples while also gaining molecular information. In this paper, we focus on integrating confocal reflectance with CDs-based fluorescence imaging in a fibre cantilever-based endoscopic scanning system. The combination of confocal reflectance and CDs-based fluorescence imaging has shown to be effective in tracking the localisation of labelled proteins while providing a clear visual of the target sample with a bench-top setup.^[33,35–38] Conducting multimodal imaging with an endoscopic scanner is a challenge in itself. The scanner requires multiple channels to accommodate different imaging modalities. To date, most endoscopic scanners employed double-clad fibre (DCF) as the fibre cantilever to accommodate two imaging modalities. DCF has a unique structure consisting of three layers of optical materials with different refractive indices (core, inner cladding, outer cladding) unlike the usual two layers in a single-mode fibre (SMF). The inner cladding offers an additional detection channel for other modalities. For example, the endoscopic scanner developed by Marques et al.^[39] used DCF as the fibre cantilever to accommodate fluorescence and OCT modalities. The fluorescence excitation and the NIR wavelengths were combined using a wavelength division multiplexer and transmitted to the endoscopic scanner probe. The inner cladding mode collected the fluorescence signal while the backscattered OCT signal was collected via the core. Both signals were then separated using a double-clad fibre coupler and routed to their respective detection methods. Using a similar method, many different combinations of imaging modalities can be used for multimodal imaging.^[40–44] However, many challenges are involved as this method involves multiple optical components working together to combine transmission light of different wavelengths and separate backscattered optical signals for detection. Consideration must also be given to the light transmission and detection scheme for different modalities. The involvement of multiple components also introduces attenuation to the backscattered optical signals.

In this paper, we report a mixed-mode, dual-core fibre cantilever-based endoscopic scanner for confocal reflectance and CDs-based fluorescence imaging. The proposed dual-core fibre cantilever was made from a pair of SMF and multimode fibre (MMF), and this was mainly done to provide mechanical asymmetry to produce the Lissajous scanning pattern.^[45,46] Interestingly, the inherent nature of this fibre cantilever also provides different channels from both fibres, allowing additional imaging modalities. The optical principle of multimodal imaging using the dual-core fibre cantilever was described. The CDs used in this paper were previously synthesised using hydrothermal methods, similar to another publication.^[47] The CD’s synthesis method, characterization and application of these CDs in sensors were presented in previous works.^[48,49] In other works of literature, the synthesised carbon dots were used in designing a sensor for chlorophyll detection. In this study, we focus on the application of the synthesised carbon dots to enhance the vascular features visualization for optical imaging using a mixed-mode fibre-based endoscopic scanner. The fluorescence imaging capability was first verified with standard fluorescent microspheres and polymer film containing CDs. The feasibility of this system was demonstrated through fibrous network samples, including those containing fluorescent microspheres and CD solution, as well as a plant sample fed with CDs. The surface profile of the target sample was distinctly revealed, with spatial localization of the labelled areas achieved through fluorescence.

Experimental detail

Optical principle of mixed-mode fibre cantilever

The dual-core fibre cantilever refers to the fibre cantilever that is made from a pair of optical fibres that are attached side-by-side, parallel with the fibre axis. In our previous work,^[45] this fibre cantilever was made with a pair of SMFs and it yielded a mechanical asymmetry necessary for the Lissajous endoscopic scanner.^[50] The scanner was only used for confocal reflectance imaging. Instead of collecting backscattered light via the core, it is possible to collect it via the cladding mode with a higher numerical aperture (NA). This dual-NA operation has been demonstrated in several reported works.^[51–53] The illumination is achieved by focusing the light emerging from the SMF core using a focusing lens. The backscattered light is then collected using the larger numerical aperture in the cladding and passes to the mounted photodiode.

When it comes to fluorescence imaging, the SMF core can be used for illuminating excitation light, but the cladding mode would unintentionally capture both the backscattered excitation light and fluorescence signal from the sample. Given the photodiode was mounted to the fibre cantilever using polymer cement, filtering out the excitation light proved challenging. An alternative approach involves utilising the core of the adjacent SMF to capture both backscattered excitation light and the fluorescence signal, allowing for filtering at the back end of the endoscopic scanner. However, subsequent testing revealed that the light captured through the core of the adjacent SMF was too weak in power for detection. To overcome this issue, a larger core is needed to allow more backscattered light to be coupled into its core. Therefore, a fibre cantilever made from an SMF and an MMF was proposed. This cantilever is labelled as the mixed-mode fibre cantilever throughout the paper.

Figure 1 shows a schematic of the multichannel detection using the mixed-mode fibre cantilever. The size of the components was exaggerated for easier visualisation. Figure 1(a) shows the optical principle for cladding detection mode, where the SMF core is used for illumination and the cladding with higher NA is used to capture the backscattered light. This configuration is primarily used for confocal reflectance imaging. Figure 1(b) shows another detection mode using the MMF core. Here, the SMF core is mainly used to transmit the excitation light to the target sample. The larger MMF core (core size: 62.5 µm) then captures both the backscattered excitation light and the fluorescence signal from the target sample. At the back end of the scanner, the excitation light is then filtered out from the fluorescence signal using a combination of dichroic mirror and emission filter. The intensity of the captured fluorescence signal was found to be sufficiently high for photodiode detection.

Figure 1. (a) Optical principle of multichannel detection using a mixed-mode fibre cantilever: (a) cladding detection mode, (b) multimode core detection mode.

Endoscopic scanner construction

Figure 2(a,b) show the top and front schematic view of the endoscopic scanner using the proposed mixed-mode fibre cantilever. The construction follows the same procedure mentioned in previous work.^[45] Briefly, the fibre cantilever was constructed using a custom 2-core drop fibre cable consisting of an SMF and an MMF. The jacket was removed and the exposed fibres were stripped, cleaned and cleaved together for a precise 90° cut. These bare fibres were then attached side-by-side using polymer cement. The constructed fibre cantilever was mounted on a 14 mm piezoelectric disc (238-000, RS component) at a 45° angle using a 3D printed triangle v-groove block. The cantilever’s length was fixed at 20 mm approximately. A surface mount PIN-photodiode (SFH2701, OSRAM) was positioned between the fibre cantilever and the v-groove block to capture the backscattered cladding modes. An apertured metal plate was added to ensure the scanning system operates in resonance.^[53,54] We enclosed the scanner within a 3D-printed housing. Figure 2(c,d) show the top and front photographs of the constructed endoscopic scanner respectively.

Figure 2. (a) Top and (b) front schematic view of the mixed-mode, dual-core fibre cantilever-based endoscopic scanner, along with its respective (c) top and (d) front photograph of the scanner. (e) Schematic of the overall scanning system.

Figure 2(e) shows the schematic of the endoscopic scanning system for confocal reflectance imaging and fluorescence imaging. For the confocal reflectance imaging, a fibre-coupled 635 nm laser diode (LP635-SF8, Thorlabs), operating at 4 mW was used for confocal illumination in the SMF core. The backscattered signal is detected by the mounted photodiode through the cladding detection mode. A 1-MHz data acquisition (DAQ) card (NI-USB-6251, National Instruments) was used to provide the scanner’s driving signals and capture the detected voltage signal from the mounted photodiode corresponding to the Lissajous scanning trajectory. An uncoated gradient-index (GRIN) lens (#64-520, Edmund Optics) was mounted before the endoscopic scanner using a holder on a manual x-y-z translation stage. The position of the GRIN lens was placed close to the scanner and adjusted to ensure the tip of the fibre cantilever was close to the centre of the GRIN lens. This placement formed a magnified image of the guided mode.

In the case of fluorescence imaging, a 405 nm fibre-coupled laser diode (WSLP-405-030m-4-B-PD, Wavespectrum), operating at 4 mW, was employed to provide illumination via the SMF core of the fibre cantilever. The backscattered light and fluorescence from the target sample were then collected via the MMF core. At the back end of the MMF, the light was then collimated using a focusing lens (C240TME-A, Thorlabs), passing through the dichroic mirror (67-064, Edmund Optics) and emission bandpass filter (67-031, Edmund Optics). Then the filtered fluorescence light was focused into an avalanche photodetector (APD) (APD120A2/M, Thorlabs) for detection using another focusing lens (C240TME-A, Thorlabs). The mounted photodiode and APD connections to the DAQ card were switched interchangeably between the confocal reflectance and fluorescence modalities. The fluorescence imaging was tailored to fluorescence labels that have an excitation wavelength, λ_ex close to 405 nm and emission wavelength, λ_em >500 nm.

The fluorescence imaging method was also extended for CDs-based fluorescence imaging. The 405 nm excitation wavelength was also used to illuminate the CDs through the SMF core of the fibre cantilever. Subsequently, the MMF core collects the backscattered light and fluorescence light emitted by the CDs within the sample. The fluorescence signal was then filtered out at the back-end of the fibre cantilever for detection.

Sample preparation

To validate the feasibility of fluorescence imaging using the scanning system with the proposed mixed-mode fibre cantilever, dry fluorescent microspheres with a diameter of 12 µm (Thermo Scientific™, Fluoro-Max Green Dry Fluorescent Particles, 35-3) were used. A 1% m/v (mass/volume) fluorescent microsphere solution with polyvinyl alcohol (PVA) was used to mount the microsphere on the microscope slide. The solution was prepared by first creating a 0.5% w/v stock PVA solution, which was simply dissolving 1 g of PVA (9002-89-5, Merck) into 200 ml of deionised water at 80 °C and stirred at 300 rpm for 30 minutes. Then 0.1 g of microspheres was added to 10 ml of the stock PVA solution.

Figure 3(a) shows the excitation and emission curve of the fluorescent microspheres obtained from the spectrometer. The intensity curves show that fluorescence can be produced when excited at 405 nm and the emission wavelength peaks at 513 nm, emitting a green fluorescence. To prepare the microsphere sample, a drop of the prepared fluorescent microsphere solution was added onto a microscope slide and spread out into a thin layer using a cover glass. Subsequently, the microscope slide was set aside to dry before multimodal imaging.

Figure 3. (a) Intensity curve of 1% w/v fluorescent microspheres solution under 405 nm excitation. The inset shows the photographs of the microsphere solution under daylight and UV (365 nm) irradiation respectively. (b) Photoluminescent excitation and emission spectra of the CDs used. The inset figure shows the CDs solution’s appearance under daylight and UV irradiation respectively. (c) Photoluminescent intensity of CDs and CD-PVA films kept for 30 minutes under constant UV (365 nm) irradiation. (d) Controlled PVA film and CD-PVA film under daylight and UV irradiation, respectively.

PVA film mixed with CDs (abbreviated as CD-PVA film) was used to further verify the feasibility of CDs-based fluorescence imaging. PVA polymer was chosen as a medium for embedding the CDs due to its hydrophilic nature, ease of preparation and high optical transparency, which minimizes interference with the excitation and emission of CDs.^[55,56] Additionally, PVA polymer has been reported to enhance the photostability of CD-PVA films.^[57–59] The CDs used for the CD-PVA film were synthesised in the Universiti Kebangsaan Malaysia (UKM) Photonic Technology Lab using two steps of hydrothermal methods based on a published work.^[47] Comprehensive details on the synthesis method, characterization and application of these CDs in sensors can be found in previous works.^[48,49] Briefly, 1.0507 g of citric acid was dissolved in 10 ml of deionised water, and 335 µl of ethylenediamine was added to the mixture. This mixture was heated at 200 °C for 12 hours using a Teflon-line autoclave and then was let to cool down to room temperature naturally. After that, 1 ml of polyethyleneimine solution was added, and the mixture underwent an additional 5 hours of heating at 80 °C. The solution was centrifuged at 4000 rpm for 30 minutes and filtered using a 0.22 µm membrane filter. Finally, the resulting supernatant solution was collected and dialyzed using a dialysis bag for a day, yielding the CDs. Figure 3(b) shows the CDs solution’s photoluminescent excitation and emission curve. Under ultraviolet (UV) irradiation, the solution produces bright blue fluorescence. The CDs have a peak excitation and emission wavelength at 512 and 551 nm, respectively. The excitation curve covers 375–550 nm, while the emission curve ranges from 490 to 650 nm. From this characterisation, it is possible to excite the CDs using a 405 nm excitation wavelength. The quantum yield of CDs was reported to be approximately 55% in a previously published paper utilising a similar synthesis method.^[60]

Synthesising the CD-PVA film involved several steps. Firstly, 0.1 ml of CDs was diluted in 10 ml of deionised water. Then, 1 g PVA was added to the solution, which was heated at 80 °C and stirred at 300 rpm for 2 hours. Once the PVA solids were dissolved, the mixture was transferred to an 8 mm glass petri dish and was left to dry in the oven at 50 °C for 2 hours. The dried CD-PVA film was then carefully peeled from the glass petri dish. A controlled PVA film without CDs was also prepared as a control sample using the same procedures mentioned earlier, excluding the addition of CDs. Small sections of PVA and CD-PVA film were cut using a razor blade and placed on a microscope slide for multimodal imaging.

The photostability of the CDs and CD-PVA film was studied by measuring the photoluminescent intensity under prolonged UV (365 nm) irradiation. Figure 3(c) shows the photoluminescent intensity of the CDs and CD-PVA film after 30 minutes of UV irradiation. For CDs, the intensity decreased by 52% after 15 minutes of exposure while the intensity for CD-PVA film only decreased by 0.05%. Figure 3(d) shows the photograph of the PVA and CD-PVA films under daylight and UV irradiation respectively. The high photostability of the CD-PVA film ensures that the emitted fluorescent light remains almost constant even after 30 minutes of UV irradiation. It is worth noting that the photostability of the CDs and CD-PVA film has minimal impact on the quality of the reconstructed image. The developed endoscopic scanner provides point-to-point illumination over the target sample throughout the scanning operation. Consequently, the scanning mechanism minimizes the prolonged exposure of CDs at every discrete point.

Expanding the array of samples for multimodal imaging, lens-cleaning tissues and plant samples were additionally incorporated. The lens-cleaning tissues represent complex details of a fibrous network composed of fibres with various widths that mimic a vascular network. The previous dry fluorescent microspheres were added to the tissue to act as foreign objects to be detected by fluorescence imaging. Furthermore, a portion of the lens-cleaning tissues were dipped into a 0.0001% v/v CD-PVA solution. The solution was prepared by mixing 10 µl CDs with 10 ml of the stock PVA solution. The use of the PVA solution was to ensure the small CDs were fixed within the fibrous network. Thus, parts of the fibrous network that come in contact with the CDs solution would produce fluorescence. Figure 4(a) shows the photograph of the prepared lens tissues with fluorescent microspheres and CDs under daylight and UV irradiation. A small section of the tissue sample was cut and placed on a microscope slide for multimodal imaging.

Figure 4. (a) Lens cleaning tissues with microspheres (left) and CDs (right) under daylight and UV irradiation respectively. (b) Controlled (left) and CDs-fed (right) vigna radiata plants under white light and UV light.

Another examined sample is the vigna radiata (mung bean) plant, where CDs were introduced into their transportation system for visualisation purposes. The utilisation of CDs in plants has been widely explored due to their low toxicity and potential to improve crop production.^[61] Using a similar methodology described in previous work,^[62] the mung bean seed was first sterilised and placed in two separate petri dishes (20 seeds per petri dish). One petri dish served as a control, receiving distilled water only, while the other petri dish received 10 ml of 0.002% v/v CDs solution. Both Petri dishes were covered and placed in an environment without light at 25 °C for growth. Regular checks were conducted to ensure sufficient water or CDs solution in each dish. After 8 days, when the first pair of leaves had grown to a reasonable size, a mung bean sprout was extracted from each petri dish and rinsed with deionised water thoroughly. Figure 4(b) shows the controlled and CDs-fed mung bean plant under daylight and UV irradiation, respectively. The CDs-fed mung bean plant emits the same blue fluorescence. The leaf of the mung bean was cut using a razor blade and placed on a microslide for multimodal imaging.

Image reconstruction

Each imaging technique was performed sequentially, from confocal reflectance to fluorescence imaging. The sample images were reconstructed based on the backscattered light using an algorithm written in MATLAB. The reconstructed images were coloured in grey depending on the reflected intensity. The brightness and contrast of the reconstructed images were adjusted to improve visibility for analysis. Psuedo-colouring was also done on the fluorescence reconstructed images to improve image clarity.

All reconstructed images of samples were compared to their respective images obtained using a standard microscope. The features exhibited on the samples from different imaging modalities were highlighted and compared. The similarities between the reconstructed images and those obtained from the standard microscope were discussed.

Results

Characterisation of endoscopic scanner

Figure 5(a,b) show the linear scan lines of the mixed-mode fibre cantilever for x- and y-axes respectively. The left and right fibres were coupled with 405 and 635 nm, respectively, to visualise the scan line. Both linear scan lines were successfully generated at both orthogonal axes without any mechanical cross-coupling. This results in a clear Lissajous pattern scan as shown in Figure 5(c). Due to the MMF having a larger core size, the scan line on the right is thicker in width and appears blurry in comparison to the scan line on the left (SMF’s core). Figure 5(d) shows the frequency response of the mixed-mode fibre cantilever for both orthogonal axes. The measured resonant frequencies are f_high = 596.5 Hz on the x-axis and f_low = 282.0 Hz on the y-axis. Both frequencies are sufficiently separated by 314.5 Hz. The frequency ratio is 2.1153, which is a non-rational number that signifies a non-repeating Lissajous scan.

Figure 5. Linear (a) x-axis and (b) y-axis scanning lines showing clear lines without mechanical cross-coupling from both fibre cores. The SMF and MMF core were coupled with 405 nm and 635 nm laser respectively. (c) Clear Lissajous scan from both fibre cores. (d) Measured frequency response curve of the mixed-mode fibre cantilever. f_high= 596.5 Hz (x-axis) and f_low = 282.0 Hz (y-axis).

After ensuring a successful Lissajous scan, the performance of the endoscopic scanner was characterised. Confocal imaging using the SMF as an illumination channel was performed on the United States Air Force (USAF) 1951 resolution test target shown in Figure 6(a) to estimate the resolution. Figure 6(b) shows the result of the confocal imaging on the resolution test target. Several line pairs in Group 6 can be visualized, and the estimated lateral resolution is 6.20 µm (80.6 lp/mm from Group 6, Element 3). The beam spot size at the focal plane was captured using a microscope, and its intensity profile was plotted shown in Figure 6(c). The measured lateral resolution is 5.81 µm. The result matches with the image obtained from the USAF 1951 resolution test target. In addition, microspheres were used to measure the axial resolution by capturing the reflected light intensity at the tip of the microspheres. The intensity profile is plotted in Figure 6(d) and the measured axial resolution is 6.52 µm.

Figure 6. USAF 1951 Resolution Test Target, Group 6 and 7: (a) Microscope reference and (b) reconstructed image from the mixed-mode fibre cantilever-based endoscopic scanner. (c) Beam profile at the focal plane with Gaussian fitting. The Inset figure shows the microscope image of the beam spot at the focal plane. (d) Intensity profile at the tip of microspheres with Gaussian fitting. The lateral and axial resolution based on the FWHM are 5.81 and 6.52 µm, respectively.

Verification of CDs-based fluorescence imaging

Figure 7(a) shows a microscope image of the fluorescent microspheres under daylight and UV irradiation. These microspheres measure approximately 10–12 µm in diameter and emit a green fluorescence as shown in Figure 7(b) when excited using UV light. Most of the microspheres are clustered into groups of 3–5. Figure 7(c) presents the fluorescence image result of the microspheres obtained using the proposed mixed-mode endoscopic scanner. The image was plotted in grayscale by default, but pseudo-colouring was applied by adding artificial colour green to enhance the visualisation of the image. The fluorescent signals from individual microspheres were successfully captured. Additionally, the fluorescent light also creates a highlight gradient on the microspheres, giving a 3D appearance. For instance, a bright spot can be located at the right of the microsphere indicated by the red arrow in Figure 7(c). As the light travels around the microspheres, the light gradient decreases, creating a 3D appearance.

Figure 7. Microscope image of fluorescent microspheres under (a) daylight and (b) UV irradiation. (c) Fluorescence image results of the microspheres. Microscope image of PVA and CD-PVA films under (d) daylight and (e) UV irradiation, with (f) corresponding fluorescence image result.

Subsequently, controlled PVA and CD-PVA film samples were used to demonstrate the endoscopic scanner’s capability to detect fluorescence signals from the CDs. Figure 7(d) shows a microscope view of the two films placed adjacent to each other under daylight. The PVA film has a clear, transparent appearance while the CD-PVA film has a transparent appearance with a light-yellow hue due to the presence of CDs. Under UV irradiation, as shown in Figure 7(e), the CD-PVA film emits a distinct blue fluorescence while the PVA film remains dark. Figure 7(f) shows the fluorescence image results obtained from the films. The fluorescence emitted from the CD-PVA film was successfully captured. Similarly, the image was plotted in greyscale by default, but pseudo-colouring was applied to the image by adding the artificial colour blue.

The findings on the fluorescent microspheres and CD-PVA films have validated the feasibility of CDs-based fluorescence detection using the mixed-mode fibre cantilever-based endoscopic scanner. In both cases, the presence of fluorescence emitted from the fluorescent source was successfully detected by the endoscopic scanner. The results also match with the reference images from the microscope.

Multimodal imaging results

To investigate further, multimodal imaging was performed on a fibrous network-like sample. Notably, each fibre has widths ranging from 15 to 26 µm. Figure 8(a) shows a microscopic view of the fibrous network under daylight. The fibre strands have a flexible rod shape with uneven surfaces, and they are intertwined together in the network. The microspheres are barely noticeable in the image because some are located in between or behind the fibre strands. Under UV light, the microspheres are more noticeable within the fibrous network as shown in Figure 8(b). There are clusters of microspheres located in between the fibre strands (marked with red dotted circles), as well as individual microspheres located in between the fibres at the top and bottom right (labelled with red arrows). The arrangements of the microspheres are at random positions, yielding a valid close-to-nature situation of a good target sample. Figure 8(c) reveals the microscopic cross-sectional view of the sample, showing that the sample has several layers, occupying a thickness of approximately 30 µm space. The microspheres can also be observed, located in between the fibre strands.

Figure 8. Multimodal imaging on lens-cleaning tissues with fluorescent microspheres. Microscopy images of the fibrous network with microspheres under (a) daylight and (b) UV light. The clusters of microspheres were indicated within the dotted circle and individual microspheres were indicated with arrows. (c) Microscopic cross-sectional view of the sample. Reconstructed images using (d) reflectance confocal imaging and (e) fluorescence imaging (pseudo-coloured), along with the (f) overlay image showing the location of the microspheres.

Figure 8(d) shows the confocal reflectance image of the fibrous network. The reconstructed image manages to show the individual strands of fibres, although some fibres appear out-of-focus due to the overlapping nature of the fibrous network. This finding is due to the scanner’s axial resolution, which is limited to 6.52 µm. The right side of the image is also brighter indicating that the fibres in that area are in-focus. Considering the thickness of the sample is about 30 µm, it is challenging to obtain a clear, full view of the sample. Additionally, the microspheres within the sample are also not noticeable due to the limited axial resolution. The lack of visibility is due to their varying height within the sample.

Subsequently, Figure 8(e) displays the fluorescent image of the sample, where the fluorescence emitted from the microspheres was successfully captured. Several clusters of the microspheres can be observed at the centre of the image where their fluorescent light also illuminated the nearby fibre strands. The individual microspheres at the top left and bottom right can also be observed. When comparing the brightness of the fluorescent light, the microspheres at the bottom right are the brightest, implying that those microspheres are mostly in focus. Meanwhile, the microsphere at the top left appears the least bright, suggesting that it is furthest away from the focus point of the scanner.

Figure 8(f ) shows the overlay image of confocal reflectance and fluorescence imaging. With the complementary surface detail of the fibrous network, fluorescence imaging pinpoints the locations of the fluorescent microspheres within the sample. The locations of the microsphere in the overlay image have a similar match with the microscopic reference image.

Moving on to the lens-cleaning tissue sample that’s partly soaked with CDs solution, Figure 9(a) shows the microscopic view of the fibrous network sample. Similarly, individual fibre strands of different sizes can be observed intertwined with each other within the network. Notably, there is no apparent indication of the CDs within the sample. This is because the CDs are small in size, which falls within the nanometre range. The minuscule CDs can only be visualised under UV light as shown in Figure 9(b). Within the image, the CDs are mostly visible in the top-right region indicated by the red dotted box. Remarkably, the CDs can penetrate the fibre strands, unlike the fluorescent microspheres. As a result, the fibre strands appeared to be producing the fluorescence. This presents an opportunity for the endoscopic scanner to detect the CDs within the fibre strands.

Figure 9. Multimodal imaging on lens-cleaning tissues partially soaked with CDs. Microscope images of the fibrous network sample under (a) daylight and (b) UV light. The corresponding (c) confocal reflectance and (d) fluorescence image of the sample with (e) their overlaying image.

Confocal reflectance imaging was conducted on the sample and its reconstructed image is presented in Figure 9(c). Likewise, individual fibre strands within the network can be observed in the reconstructed image. The light gradient on individual fibre strands was observed, giving a 3D appearance to the fibres. Switching to fluorescence modality, Figure 9(d) presents the fluorescent image of the sample, where the fluorescence signals are captured mostly at the top right of the image. The fluorescence signals manifest in strand-like patterns with several spots exhibiting bright fluorescent light. These bright spots indicate that these areas are in focus of the scanner.

Following that, the confocal reflectance and fluorescence images were combined to obtain a more comprehensive view of the fibrous network sample. Figure 9(e) illustrates the overlay image, showing that the fluorescence light does originate from the fibres on the top right of the image. These results confirm that the CDs solution can be found at the top right of the sample. It is also observed that the fluorescent light does not uniformly cover the entire width of the fibre strand, possibly due to the scanner’s limited axial resolution.

Further application of multimodal imaging was performed on CDs-fed mung bean plant samples. CDs solution was used because it has low toxicity, which is harmless to the plant sample. In addition, the minuscule CDs are small enough to be absorbed into the transport system of the plant via the veins. In contrast, the fluorescent microspheres were not chosen due to their relatively large size compared to the plant samples. The microspheres are made from polystyrene, which is a synthetic polymer that plants do not absorb.

Figure 10(a) presents a microscope image of a leaf taken from the plant sample, showing the leaf base and the branching veins. Within the image, the branching vein has a smaller vein measuring approximately 48 µm in width and a section of a larger vein that has a width of 169 µm approximately. The surface details of the leaf base and the vein can be observed but the CDs are not visible. Thus, placing the sample under a UV light allows the microscope to capture the fluorescence from the CDs solution within the leaf sample as shown in Figure 10(b). The microscopic image reveals that the CDs solution has travelled into the plant’s veins while transporting the water. Most of the CDs are concentrated within the larger branching veins, and some CDs travel into the leaf base where faint fluorescence can be observed. Notably, there are two unknown dark spots on the veins (indicated by the arrow) that do not produce any fluorescence.

Figure 10. Microscope image of a mung bean leaf under (a) daylight and (b) UV light. Reconstructed images of the respective leaf using (c) reflectance confocal imaging and (d) fluorescence imaging. (e) Overlay image from reflectance confocal image and fluorescence.

Using confocal reflectance imaging, the reconstructed image of the sample in Figure 10(c) was obtained, showing the surface profile of the leaf base and the branching vein. The larger vein appeared brighter because the top part of the vein was in focus. Subsequently, fluorescence imaging was performed on the leaf sample, producing Figure 10(d). The image shows the fluorescence signals in a large vein-like pattern, most likely from the CDs within the veins of the sample. Faint indications of the unknown dark spots are also noticeable within the fluorescence image. By overlapping the confocal reflectance and fluorescence images, the overlap figure in Figure 10(d) has confirmed that the fluorescence did originate from the CDs in the veins of the leaf sample. The fluorescence in the leaf base was not captured by the fluorescence detection, possibly due to a very weak fluorescence signal.

Discussion

The characterisation results demonstrated that the proposed mixed-mode fibre cantilever-based endoscopic scanner was capable of performing a Lissajous scan for imaging. The feasibility of fluorescence detection using the mixed-mode fibre cantilever was then verified with fluorescent microspheres and CD-PVA film. Results have shown successful detection of fluorescence signals.

Further exploration of multimodal imaging was conducted on more intricate samples such as the lens-cleaning tissue. It has a complex overlapping fibrous network, mimicking a superficial vascular network. Fluorescent microspheres and CDs solution were randomly deposited, creating a natural representation of foreign objects within the sample. The size of the microspheres is small relative to the fibre strands, which makes it difficult to locate them within the network. Whereas the size of the CDs is tens of nanometres. It is simply too small for the endoscopic scanner to pick up any weak individual back-scattered signal from a singular CD. The minuscule CDs can also penetrate into the fibre strands. Based on the findings, the structures of the fibrous network were successfully captured by the confocal reflectance imaging, while the fluorescence imaging successfully detected the fluorescence signals. When overlapping the images and combining the information from both imaging modalities, the origin of the fluorescence signal can be pinpointed to the microspheres and CDs.

Furthermore, the findings clearly show that multimodal imaging effectively identifies CDs within real, living samples. Confocal reflectance imaging reveals surface details and fluorescence imaging detects the emitted fluorescence from the CDs. The complementary strength of these two imaging methods allows for the accurate localization of CDs within complex living samples.

Throughout the multimodal imaging experiments, the confocal reflectance imaging often encounters challenges with the clarity of the reconstructed image. The lack of clarity is mostly due to the limited axial resolution of the endoscopic scanner. While a thin, flat sample would be preferable, confocal imaging is still able to provide a valuable surface view of the target sample. Besides that, the performance of the detection system in fluorescence imaging is dependent on the intensity of the fluorescence signals. When fluorescence emissions are weak, the detection system may have difficulty capturing them, resulting in darker images. CDs exhibit excellent optical properties, generating bright fluorescence that proves advantageous in the fibre-based endoscopic scanning system.

Conclusion

In this paper, we successfully integrated CDs-based fluorescence imaging with confocal reflectance imaging in a mixed-mode fibre cantilever-based endoscopic scanner. By utilizing two attached SMF and MMF as the mixed-mode fibre cantilever, it inherently produces a clear 2D Lissajous scan and also offers a notable advantage over the commonly used DCF by providing multiple illumination and detection channels. The fluorescence detection capability was first verified using fluorescence microspheres and CD-PVA film. Then, CDs-based fluorescence imaging with confocal reflectance was performed on fibrous network-like samples deposited with microspheres and CDs, as well as CDs-fed plant samples. The findings show that the confocal reflectance modality captures the surface profile, revealing distinct features of the target sample. Concurrently, the fluorescence modality complements confocal reflectance by capturing emitted signals from fluorescent microspheres and CDs. The combined results effectively localise the labelled structure within the target sample. The complementary strength of confocal reflectance and CDs-based fluorescence proves to be a powerful technique for endoscopic scanners, enabling comprehensive sample studies. This capability is particularly valuable in identifying foreign or cancerous cells within the sample. All that it needs is functionalised CDs that can chemically attach to specific molecules or proteins within the cell structure. Ultimately, this endoscopic scanner exhibits great potential for diverse applications, ranging from imaging blood vessels to identifying labelled structures within the gastrointestinal tract, as well as detecting specific proteins within sample structures.

Nomenclature
CD	=	carbon dots
SMF	=	single-mode fibre
OCT	=	optical coherence tomography
DCF	=	double-clad fibre
MMF	=	multimode fibre
NA	=	numerical aperture
DAQ	=	data acquisition
APD	=	avalanche photodetector
PVA	=	polyvinyl alcohol
USAF	=	United States Air Force

Acknowledgment

The authors would like to thank the Photonics Technology Laboratory, Department of Electrical, Electronic and Systems Engineering, Faculty of Engineering and Built Environment, Universiti Kebangsaan Malaysia for all the amenities provided.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The author acknowledges the Fundamental Research Grant Scheme (FRGS) [grant number FRGS/1/2021/TK0/UKM/02/15] funded by the Ministry of Higher Education (MOHE), Malaysia, and part of this research is supported by the Geran Universiti Penyelidikan (GUP), [grant number GUP-2022-009] funded by Universiti Kebangsaan Malaysia (UKM).