The optoelectronic excitation module

In the previous system [27], the microfluidic channel was illuminated from above along its entire length by means of a laser line generated by using a cylindrical lens. The resulting fluorescence, guided within the shaped upper part of the chip, was collected by a single plastic optical fibre and subsequently filtered by means of a high-pass interference filter and coupled to a photodetector. The interrogation of the 13 microfluidic channels was performed via sequential mechanical scanning. This procedure was characterised by a significant temporal duration (16 min), which considerably prolonged the overall assay time. Moreover, it introduced additional complexity in the device implementation, due to the requirement for a high-precision motorised translation stage capable of moving both the illumination module (laser source and cylindrical lens) and the detection module (optical fibre and photodetector) in a coordinated manner, in order to interrogate each of the 13 microchannels sequentially.

Both limitations can be effectively addressed through the use of the DOE, which enables the generation of a structured diffraction pattern composed of 13 distinct lines. This configuration allows the simultaneous interrogation of all 13 microchannels of the chip. This parallel excitation approach eliminates the need for mechanical scanning, thereby significantly reducing the total acquisition time and improving temporal resolution.

The customised DOE was designed with the following specifications:

material: glass

DOE dimensions (diameter × thickness): 8 mm × 1.2 mm

geometry of the diffraction pattern: parallel lines

number of lines: 13

design wavelength: 635 nm

inter-line spacing (at 15 cm): 2 mm

line length: 10 mm

pattern total size (at 15 cm): 24 mm × 10 mm

The use of a laser diode as an optical source led to the optimization of the diffracted pattern facilitating the system integration. Due to the intrinsic beam divergence of the laser diode, the emitted beam exhibited an elliptical profile. Partial collimation was achieved using the aspherical lens. This configuration reduced, without completely eliminating, beam divergence, resulting in an elliptical spot on the DOE with approximate dimensions of 1 mm (major axis) and 0.3 mm (minor axis). The LD was aligned with the DOE such that the major axis of the elliptical spot was parallel to the orientation of the diffracted lines. This alignment condition maximised the overlap between the excitation pattern and the microfluidic channel geometry, ensuring a more uniform distribution of optical power across all channels. Figure 1 shows the diffracted pattern projected onto a screen (a), as well as the transverse intensity profile at the centre of the line pattern (b), the average intensity distribution for all 13 lines (c), and the intensity profile of the central line (d). Figures S1 and S2 in the Supplementary Materials show the diffracted pattern, the intensity profile of the central line, the average intensity distribution, and the transverse intensity profile, both for the case in which the major axis of the elliptical LD spot is oriented transversely to the lines and for the case of a perfectly circular spot. It is worth noting that the typical anisotropic divergence of the laser diode, in this case, enables an extremely uniform optical power distribution across the 13 lines and, consequently, across the 13 channels of the microfluidic chip. This uniformity is critical to minimise channel-to-channel variability in fluorescence excitation, directly contributing to improved quantitative accuracy in multichannel measurements. The longitudinal intensity fluctuations observed in Fig. 1d originate from laser speckle, resulting from coherent illumination combined with phase modulation introduced by the DOE. While individual lines exhibit local variations (relative standard deviation around 12%), these fluctuations are largely uncorrelated and therefore significantly reduced when considering the total illumination. The integrated intensity over all 13 lines shows a much lower relative standard deviation of 1.7%, indicating stable overall optical power delivery despite local inhomogeneities. The influence of the illumination conditions was further assessed by comparing different beam profiles at the DOE plane. A perfectly collimated laser beam producing a circular spot (Figure S2 in the Supplementary Materials) leads to a more irregular intensity distribution (relative standard deviation of 54%), with fluctuations propagating to the integrated signal (10% standard deviation), due to inefficient speckle averaging. To mitigate this effect, a slightly non-collimated diode laser beam with an elliptical spot profile was employed. This reduces the effective spatial coherence at the DOE and promotes partial speckle averaging, improving illumination uniformity and reducing integrated intensity fluctuations from 10 to 1.7%.

Fig. 1

Characterisation of the diffraction pattern generated by the diffractive optical element (DOE): a photograph of the pattern projected onto a screen, showing the 13 parallel excitation lines; b transverse intensity profile measured at the centre of the pattern; c average optical power distribution across the 13 lines; and d longitudinal intensity profile of the central line

The rendering of the fluorescence excitation setup is shown in Fig. 2, highlighting the laser source, the bandpass interference filter, the DOE, and the 45° mirror used for chip illumination. The compact and fully static optical configuration enhances system robustness and eliminates alignment drifts associated with moving components.

Fig. 2

Rendering of the fluorescence excitation module. The setup includes the laser diode, bandpass interference filter, diffractive optical element (DOE), and 45° mirror used to direct the structured beam onto the microfluidic chip

The resulting module was then integrated into a light-shielded section of the final prototype, described in the following sections.

The optoelectronic detection module

To maintain the parallelism of the optical interrogation system and allow the possibility of rapid optical measurements, the multiplexing feature of the excitation system must necessarily be replicated on the signal collection one. This design ensures that the benefits of parallel excitation are preserved along the entire optical interrogation chain, avoiding bottlenecks in signal acquisition.

To achieve this, the WAF array had to be appropriately interfaced on one side with the microfluidic optical chip and on the other with the large-area CCD linear sensor. On the chip side, the end faces of its top part, which acts as a waveguide for the fluorescence signal propagating within it through successive total internal reflections, were angled at approximately 60°. Thus, most of the guided fluorescence signal resulted perpendicular to the sloped surface, passing through it [34]. As a result, the WAF array was appropriately tilted to achieve optimal facing and, consequently, maximise the fluorescence signal collection (Fig. 3). This geometrical optimisation maximises the coupling of the guided fluorescence into the WAFs, thereby improving signal collection efficiency.

Fig. 3

Rendering of the fluorescence detection module. The waveguide absorption filter (WAF) array is aligned with the microfluidic chip output and coupled to a linear CCD detector. The system geometry, including the tilt of the WAF array with respect to the chip surface, is optimised to maximise fluorescence collection efficiency while minimising background contributions

In addition, it was necessary to prevent even the smallest scattered light component (i.e. diffused stray light) from the surrounding environment from reaching the WAFs transversely, thus influencing the measurement based on the fluorescence signal. For this purpose, a special holder in black polymer (Vero® black resin) was realised, under the design of the authors, by the company Topp SRL (Sandrigo (VI), Italy) with selective laser sintering (SLS) 3D printing technique. The shielding block featured 13 square-section through-holes, each about the size of a WAF, angled to maximise the collection of the fluorescence signal from the chip microchannels. On the front side, the block featured a specially shaped cross-section, with the 13 holes spaced equal to the angled faces of the chip cover. This ensured proper interfacing between the two parts (Fig. 4a). Additionally, a suitable housing for the CCD camera was created on the rear of the block (Fig. 4b). The large-scale photodetector was then secured to the block with screws. The WAFs were manually inserted into each hole, so that their front face was slightly inside the hole and, at the same time, their rear face was as close as possible to the photosensitive surface of the detector. The use of a dedicated light-shielding structure significantly reduced the contribution of stray light, leading to an improved signal-to-noise ratio in fluorescence measurements.

Fig. 4

Custom-designed light-shielding holder for the WAF array, realised by 3D printing: a front view showing the alignment of the 13 angled channels with the microfluidic chip output; b rear view showing the housing for the CCD camera

On the CCD camera side, the first tested version was the conventional one (S11156-2048-01), featuring a quartz window that protects the sensitive area. This configuration implies a minimum distance of 1.47 mm between the photosensitive surface and the end (i.e. the output face) of each WAF, causing the different fluorescence signal contributions to spread and potentially overlap near the detector surface (Fig. 5a). In the second custom-made version of the CCD camera (S11156-2048N-01), the absence of the quartz window allowed the WAFs to be brought as close as possible—in a quasi-contact configuration—to the sensitive surface of the photodetector, drastically reducing the possible optical crosstalk (Fig. 5b). Reducing the distance between the WAF output and the detector surface limits the spatial spread of the fluorescence signal, thereby mitigating inter-channel overlap and preserving the spatial resolution of the multichannel readout. This configuration is essential to ensure accurate signal discrimination between adjacent channels, which is a key requirement for reliable multiplexed detection.

Fig. 5

Comparison of fluorescence signal acquisition using two CCD configurations: a standard camera with protective quartz window (S11156-2048-01 CCD camera), resulting in increased optical spread and potential inter-channel overlap; b window-less camera (S11156-2048N-01 CCD camera) enabling quasi-contact coupling between the WAF output and the detector surface, thereby reducing crosstalk and improving spatial resolution of the multichannel readout

Although all necessary countermeasures were taken to address unwanted phenomena (e.g. stray light; crosstalk during signal collection), inter-channel optical crosstalk occurring between adjacent channels within the microfluidic chip was still present. This issue arose from the close spacing of the channels on the chip (approximately 2 mm apart, along of more than 10 mm) which allowed fluorescence travelling along a specific channel to couple into neighbouring channels through scattering and multiple reflections. The parallel interrogation system approach makes it necessary to implement alternative strategies aimed at reducing the possible presence of inter-channel crosstalk. This issue was addressed through post-processing of the microfluidic chip by filling the spaces between the channels with commercial black silicone. The effectiveness of this approach was assessed by filling all the channels with PBS and then sequentially replacing the PBS with a fluorescent solution (IgG labelled with Alexa Fluor 647 5 µg/mL), acquiring the optical signal after each step. Crosstalk was evaluated by comparing the signal collected from a given channel filled with PBS before and after the sequential filling of the other channels with the fluorescent solution. To specifically evaluate the effect of the black silicone, the spaces between channels 5, 6, 7, 8, 9, and 10 were filled with this material. The improvement achieved with the use of black silicone is evident, as reported in Table 1, where thicker black vertical lines indicate the presence of the black material in the inter-channel spaces. Each row of the table provides the signal variation observed in channels filled with PBS when the i-th channel is filled with the fluorescent solution. For instance, filling channel 2 produces a 17.34% signal variation in channel 3, a 5.13% variation in channel 4, and so on. In contrast, filling channel 6—shielded from channel 5 by black silicone—results in only a 0.33% variation in channel 7 and negligible effects in the remaining channels. Minimising inter-channel crosstalk is essential to preserve the analytical specificity of each sensing channel and to prevent false signal contributions in multiplexed assays.

Table 1. Relative signal variation in adjacent channels induced by filling a specific channel with the fluorescent solution. Each row reports the signal variation measured in channels filled with PBS when the i-th channel is filled with the fluorescent solution. The thicker black vertical lines indicate the presence of black silicone in the interspace among adjacent channels from channel 5 to channel 10

The effectiveness of the approach was further assessed in a context closer to the heterogeneous assays by immobilising IgG labelled with Alexa Fluor 647 5 µg/mL and 2 µg/mL (IgG* in Fig. 6) directly only on channels 4 and 9, respectively. Figure 6 shows the histogram with the collected fluorescence signal with the channels filled with PBS before and after the immobilisation of the labelled antibodies on the surface of channel 4 and channel 9.

Fig. 6

Evaluation of inter-channel optical crosstalk under conditions mimicking a heterogeneous assay. The histogram shows the fluorescence signal measured in all channels filled with PBS before and after the selective immobilisation of labelled antibodies (IgG labelled with Alexa Fluor 647 5 µg/mL on channel 4 and 2 µg/mL on channel 9) on channels 4 and 9

The final integrated platform

The rendering of the optoelectronic and mechanical part of the final integrated platform is depicted in Fig. 7, highlighting the transition from individually optimised modules to a fully integrated POCT platform. In the lower part of Fig. 7, the excitation module previously described in the dedicated section is visible. The microfluidic chip, highlighted by the red circle, was mounted on a sliding stage that allows the chip to be inserted into the platform, aligning it with the diffracted laser lines and with the holder containing the WAF arrays and the camera. This configuration ensures consistent spatial alignment between excitation and detection, thereby reducing position-dependent variability in the measured fluorescence signal.

Fig. 7

Rendering of the integrated optoelectronic and mechanical platform. The excitation module is shown in the lower section, while the microfluidic chip (highlighted by the red circle) is mounted on a sliding stage for precise alignment with the excitation pattern and detection module. The inset provides a detailed view of the chip-detector interface, including the WAF array, CCD sensor, and microfluidic distributor

A stepper motor was used to precisely position and connect a custom-made stainless steel microfluidic distributor to the chip, ensuring reliable and reproducible fluidic interfacing through Teflon tubing and minimising operator-dependent variability during chip connection. The inset in the upper left provides a detailed view of the chip-detector interface, including the WAF array, CCD sensor, and microfluidic distributor. This integrated configuration improves mechanical stability and preserves optical coupling efficiency, thereby reducing signal fluctuations associated with misalignment.

This assembly constituted a stand-alone unit, which was then inserted into the enclosure of the final prototype. Figure S3 in the Supplementary Materials shows a photograph of the enclosure, where the pumps and microfluidic valves, the PC, the USB hub for data connections, and the power supply unit are visible.

A complete view of the platform, including the fluidic handling system, is reported in Figures S3 and S4 in the Supplementary Materials. The fluidic architecture comprised:

two syringe pump modules (Hamilton PSD/4) with different valve heads (one six-port and one four-port configuration, each including a syringe port);

three valve positioners (Hamilton MVP/4), including one four-port and two eight-port distribution valves located in proximity to the chip;

a fluidic mixer implemented as spiral-shaped tubing.

This modular configuration enabled flexible routing and mixing of reagents required for multi-step immunoassays. In particular, the proximity of the valve system to the chip allowed precise control of reagent exchange within the microchannels, improving assay repeatability and reducing the risk of cross-contamination.

The fluidic components were arranged to minimise the total internal volume and reduce dead volumes, thereby limiting the priming volume of the fluorescence module and improving assay response time, which is particularly advantageous in point-of-care settings where rapid time-to-result and low reagent consumption are essential.

A small form factor PC with MS Windows operating system was located inside the final prototype box (Figure S3 in the Supplementary Materials) and externally connected through HDMI and USB ports. A dedicated graphical user interface (GUI), developed in Java, enabled integrated control of the optical and fluidic subsystems, allowing both manual operation and fully automated execution of assay protocols. In Fig. 8, the Control Window of the Fluorescence Module organised into four sections is presented:

flow cell status (upper left): indicates and controls the fluidic connection of the flow cell

fluidic control (lower left): enables manual operation of pumps and valves for stepwise assay development

optical control and fluorescence acquisition (centre): allows the user to manage optical components and perform fluorescence measurements

procedure control (right panel): provides access to pre-programmed automated assay protocols.

Fig. 8

Graphical user interface (GUI) for system control. The software integrates fluidic handling and optical acquisition through four main sections: flow cell status, manual fluidic control, optical control and fluorescence acquisition, and automated protocol execution. This interface enables both stepwise operation and fully automated assay workflows

The integration of fluidic and optical control within a single software environment ensured proper synchronisation between reagent handling and fluorescence acquisition, thereby reducing timing-related errors and improving measurement reproducibility. Moreover, the availability of an integrated software interface enabling full control of both the fluidic and optical modules allowed the execution of fully automated measurement protocols. The availability of pre-programmed protocols supported the standardisation of assay procedures, which is a key requirement for reliable POCT deployment.

In this framework, the operational reliability was assessed by evaluating the stability and repeatability of the fluorescence signal upon repeated loading and unloading of the microfluidic chip with the sliding stage. For this purpose, the stability of the acquired optical signal as a function of repeated loading and unloading of the optical chip was evaluated. The chip microchannels, previously filled with anti-mouse IgG 5 µg/mL labelled with Alexa Fluor 647, were allowed to dry completely prior to each measurement in order to avoid any possible variation of the signal induced by leakage of the solution present in the channels during the loading-unloading procedure. The measurement of the emitted fluorescence was repeated ten times, and between each acquisition, the same chip was removed from the sliding stage and loaded again. The results are summarised in Fig. 9, where, for each channel, the acquired optical signal normalised to the average signal is reported with the indication of the CV value on the top of every channel. By inspecting the data reported in the histogram, a slight decrease in fluorescence signal can be observed for some channels when comparing the first insertions with the last ones. On average across the 13 channels, this corresponds to a small reduction in signal when comparing the first five measurements with the last five, with an overall variation of approximately 2%, while no systematic trend is present within each subset. This behaviour is attributed to a mechanical settling of the chip on the stage. For this reason, a short pre-conditioning (stabilisation) sequence is routinely performed when introducing a new chip, ensuring optimal measurement stability.

Fig. 9

Assessment of system repeatability through repeated loading and unloading cycles (n = 10). For each channel, the fluorescence signal is normalised to the average value, and the corresponding coefficient of variation (CV) is reported

Overall, the low variability observed upon repeated loading/unloading cycles highlights the robustness of the platform in conditions compatible with routine POCT use. The temporal stability of the intensity distribution was further confirmed by these repeated measurements, which showed minimal variability despite sample removal and repositioning (see Fig. 9), making the platform suitable for rapid point-of-care testing applications where time-to-result is critical. On this basis, the platform was subsequently validated at the bioanalytical level by implementing a model fluorescence sandwich immunoassay targeting C-reactive protein (CRP) on the multichannel microfluidic chip.

The assay relied on the monoclonal anti-human CRP capture antibody (clone C5) and an Alexa Fluor 647-labelled detection antibody (clone C7). The assay was performed through the sequential flow protocol described in SI, with a total assay time of 30 min: in particular, steps 1–5 in Table S1 in the Supplementary Materials were used for the injection and 10-min incubation of CRP followed by washing and background acquisition. The procedure steps 6–8 consisted in the injection and 10-min incubation of the fluorescently labelled detection antibody, followed by the final signal acquisition. To evaluate assay applicability in plasma, a calibration was performed by spiking CRP into CRP-deficient plasma diluted 1:5 in PBS (Fig. 10). The baseline signal measured for CRP-deficient plasma (approximately 8 a.u.) was subtracted from all data; considering the intrinsic complexity of plasma as a biological matrix, this condition already provides a stringent assessment of non-specific effects, and the residual background remains significantly lower than the signal obtained at the lowest CRP concentration tested. The experimental data were fitted by using a logistic model, where the parameters A1 and A2 represent the lower and upper asymptotes, x₀ corresponds to the CRP concentration at 50% of the dynamic range, and p is related to the slope of the curve at x₀. From this calibration curve, an LOD of 0.32 µg/mL was calculated.

Fig. 10

Calibration curve for CRP detection in plasma, obtained by spiking CRP into CRP-deficient plasma diluted 1:5 in PBS. Experimental data were fitted using a logistic model

Intra‑chip variability was assessed by measuring multiple channels on the same chip and resulted in coefficient of variation (CV) values of 12% at a CRP concentration of 0.1 µg/mL (n = 8) and 10% at 10 µg/mL (n = 5). Inter-chip variability was evaluated across four different chips (11 microchannels in total) at a CRP concentration of 10 µg/mL with a CV of 15% confirming good reproducibility between chips. Based on the characterisation of the optical platform reported in the previous sections, the observed variability can be reasonably attributed predominantly to the immunoassay steps [35, 36], while contributions from the optical subsystem and potential inter-channel crosstalk are expected to be comparatively minor.

Two simulated plasma samples were produced by spiking CRP-deficient plasma with CRP 5 and 50 µg/mL and then diluting them 1:5 with PBS (final CRP concentrations 1 µg/mL and 10 µg/mL). Results are reported in Table 2.

Table 2 Results on simulated plasma samples with CRP concentrations of 1 μg/mL and 10 μg/mL after dilution 1:5 in PBS

The results presented for the CRP immunoassay demonstrate the capability of the proposed platform to perform quantitative fluorescence measurements in clinically relevant conditions, while maintaining good reproducibility and low inter-channel variability. Beyond the analytical performance obtained for this specific biomarker, the main strength of the proposed approach resides in the underlying optical and system architecture, which differs significantly from conventional fluorescence-based POCT platforms reported in the literature.

In contrast to widely adopted fluorescence-based POCT configurations relying on imaging-based detection or sequential interrogation of individual sensing regions [1, 5, 6, 14, 16, 24], the present system integrates parallel excitation and parallel signal collection within a fully static optical architecture. In many reported implementations, multichannel detection is achieved either through wide-field imaging or time-sequential readout schemes, which can introduce trade-offs in terms of acquisition time, optical complexity, or scalability. The approach proposed here enables truly parallel interrogation, where excitation and signal collection occur simultaneously across all microchannels without the need for scanning or imaging optics.

From this perspective, it is important to emphasise that the overall assay duration in the proposed platform is no longer limited by the optical readout, but rather by the biochemical kinetics of the specific assay. In the case of CRP detection, a total assay time of 30 min was required to ensure adequate binding and signal development. However, the decoupling between optical interrogation and assay chemistry represents a key advantage of the system, as it enables the modulation and optimisation of the total analysis time depending on the target analyte and assay format. Faster assays could be implemented for biomarkers with more rapid binding kinetics, while longer incubation protocols could be adopted when higher sensitivity is required, without any impact on the readout speed.

This flexibility, combined with the intrinsic fast optical response, is particularly relevant in the context of point-of-care testing, where time-to-result and adaptability to different clinical scenarios are critical. Moreover, the integration of the optical and fluidic subsystems within a single automated platform ensures reproducible assay execution and minimises user-dependent variability, further supporting the suitability of the platform for decentralised diagnostic applications.

Overall, the proposed approach demonstrates how the implementation of a parallel optical interrogation strategy, coupled with an optimised microfluidic handling system, can address key limitations of current fluorescence-based POCT devices, particularly those related to sequential readout and acquisition time scaling with the number of channels. The proposed architecture enables acquisition times on the order of seconds that are independent of the number of interrogated microchannels, while maintaining a compact and fully static optical configuration. In this framework, it is worth noting that several commercially available POCT CRP systems, such as QuikRead go, Afinion, and LumiraDx, rely on optical detection schemes and are optimised for specific clinical use scenarios; therefore, the analytical performance reported here, including a limit of detection of 0.32 µg/mL, should be interpreted as a proof of concept demonstration of the platform capability rather than a fully optimised assay, while already falling within a concentration range that is relevant for clinical decision-making [37]. While the total assay time of 30 min is longer than that of rapid lateral flow tests, it should be noted that the latter rely on simplified assay formats based on passive capillary flow and limited control over fluid dynamics [38], whereas the present approach is based on a fully controlled microfluidic immunoassay designed to ensure quantitative and reproducible measurements. In this context, the current results should be considered as a proof of concept of the performance of a fully integrated optoelectronic platform capable of rapid, stable, and parallel fluorescence readout within a controlled microfluidic framework by means of the measurement of a single analyte.

Despite these promising results, some aspects still require further optimisation. In particular, from the optical perspective, although the current configuration already ensures efficient parallel excitation and detection with low crosstalk, additional refinements aimed at further improving signal-to-noise ratio and channel isolation could be beneficial, particularly in the context of low-concentration biomarkers or highly multiplexed assays. Finally, while the current study focuses on single-analyte detection as a controlled validation of the platform, the extension towards multiplexed assays will require dedicated optimisation of bioassay design, surface functionalisation, and cross-reactivity control, in order to fully exploit the scalability potential of the multichannel architecture.