Over the past few years, there has been a notable reduction in cancer-related deaths, largely attributed to improvements in oncology treatments, especially through the progress in next-generation sequencing, immunotherapeutic strategies, and the use of targeted medications [1]. However, the situation in hepatocellular carcinoma (HCC) is also distinct. With high rates of both incidence and mortality, HCC is one of the most prevalent malignant tumors of the digestive system. Approximately 60% of HCC cases are diagnosed at an advanced stage where the overall 5-year survival rate is below 18% due to the lack of effective early detection methods [2]. Early diagnosis of HCC is crucial for timely treatment and improving survival rates [3]. Despite the fact that imaging methods like magnetic resonance imaging and ultrasound have significantly increased the accuracy of HCC diagnosis, their use is restricted because of their high cost, invasiveness, and insensitivity to tiny tumors [4]. The presence of small molecules, such as proteins and nucleic acids, in the blood and body fluids of patients is considered a promising approach for the early diagnosis of lung cancer. Since its discovery in 1964 in the serum of HCC patients, alpha-fetoprotein (AFP) has served as the main biomarker for HCC diagnosis [5]. Protein induced by vitamin K absence or antagonist-II (PIVKA-II), also known as des-γ-carboxy prothrombin (DCP), has been considered a suitable serum biomarker specific for HCC since its first detection in HCC by Libert et al. in 1984 [6,7]. DCP is an abnormal prothrombin precursor produced during the malignant transformation of hepatocytes, showing superior performance in HCC detection compared to AFP and improving performance in terms of HCC prognosis, treatment response, and recurrence monitoring [8]. Furthermore, the assessment of DCP has demonstrated significant progress in identifying HCC occurrence in patients with cirrhosis related to hepatitis C virus or hepatitis B virus [9,10].

Currently, the general methods for detecting DCP are electrochemical detection and fluorescence detection. However, these methods require complex sample processing and possess limited specificity and repeatability [11]. Optical fiber biosensors are considered to possess significant application potential due to their strong anti-interference capabilities, high specificity and repeatability, as well as high sensitivity and rapid response time [12–15]. Optical fiber biosensors currently utilized for biomarker detection typically operate within the spectral range of 500–1000 nm. Their full width at half maximum (FWHM) is relatively large, with many reaching several tens of nanometers [16,17]. Therefore, the limit of detection (LoD) of current optical fiber biosensors is limited. One of the challenges in optical fiber biosensing is achieving ultra-high sensitivity and ultra-low LoD, facilitating better service for biological detection. The fiber ring cavity lasers have attracted widespread research interest due to the output spectra with narrow FWHM, high signal-to-noise ratio (SNR), and high extinction ratio (ER). The structure of fiber interferometers can serve as sensing elements integrated into ring cavity laser systems [18,19]. Therefore, combining optical fiber biosensors with fiber ring cavity lasers holds great promise for improving the performance of sensing devices. Chen et al. demonstrated an experimental single-mode tapered fiber immunosensor with an FWHM of 0.15 nm, achieving ultra-high sensitivity detection of Listeria monocytogenes [20]. Compared to sensors utilizing the 500–1000 nm light source range, optical fiber biosensors operating at the 1550 nm wavelength are capable of achieving high sensitivity, which allows for the highly effective detection of minute variations. Hu et al. utilized an optical fiber ring cavity laser based on the Michelson interferometer and flow control to achieve detection of small-volume and low-concentration chemical substances. The LoD for vitamin C detection was 0.06 mg/mL [19]. The above results demonstrate that the 1550 nm laser has stronger anti-interference capabilities, better beam collimation, and higher source brightness, which allows for more precise object recognition. Moreover, it is necessary to enhance the transmitted optical signal, further improving the performance of biosensors based on fiber ring lasers.

Localized surface plasmon resonance (LSPR)-based optical fiber biosensors offer several advantages, including high sensitivity and real-time detection capabilities [17]. The LSPR signal in fiber sensors can be enhanced by employing nanomaterials and specific fiber structures. Gold nanoparticles (AuNPs), due to their unique optical properties and biocompatibility, are widely used in the development of LSPR-based biophotonic optical fiber sensors. Compared to various other metals, AuNPs exhibit the highest sensitivity to changes in the surrounding RI [21]. Virk et al. developed a D-shaped fiber probe with AuNPs immobilized in the sensing area for the detection of GaHIgG. By varying the antigen concentration, an LoD of 0.6 μg/mL was achieved [22]. Singh et al. used a W-shaped LSPR sensor and achieved precise detection of casein using AuNPs and MXene. The LoD was 6.96 μM (1 M = 1 mol/L) and the sensitivity was 0.0385 nm/μM [23]. However, the LSPR range excited by uniformly sized AuNPs is generally in the visible light region. There is a need for further exploration of materials that can excite the LSPR effect at near-infrared wavelengths. The free electrons on the surface of gold nanorods (GNRs) will collectively oscillate under light excitation, resulting in the LSPR phenomenon. In addition, GNRs with different aspect ratios can lead to changes in the distribution of electron clouds and oscillation modes on the surface, thus affecting the position of the LSPR formant. Varsha et al. used GNRs with an aspect ratio of 10 [LSPR peak at 1400 nm; the solution has a relatively wide absorption bandwidth (∼380nm at 0.5 absorbance)] to achieve a Q-switched pulsed laser at the desired wavelength (1567 nm). This work demonstrates the feasibility of GNRs in the mid-infrared laser band [24]. These characteristics make GNRs suitable for near-infrared LSPR sensors.

In this work, a multi-mode fiber (MMF)-multi-core fiber (MCF)-MMF structure is fabricated and employed as a sensor probe. Due to the presence of MCF, the fiber probe exhibits good sensitivity to RI changes in the surrounding environment and enhances the interference effect of light. A W-shaped fiber probe is used to enhance the evanescent waves (EWs) by tapering the fiber probe and applied in the Er-doped fiber ring laser. To enhance the light signal in biosensors based on an optical fiber laser, GNRs with a diameter of 10 nm and a length of 102 nm are immobilized in the sensing area to excite the LSPR effect near the 1550 nm wavelength. The combination of special fiber structure and LSPR can improve the interaction between the light and the target substance. The specificity of the optical fiber laser biosensor is achieved by an immobilized DCP antibody in the fiber probe. The results of LoD of 367.6 pg/mL are obtained in a linear range of 0–1000 ng/mL with laser spectrum FWHM of 0.11 nm, significantly enhancing the specific recognition ability of biosensors and sensing performance. Moreover, the test of repeatability, reproducibility, stability, and selectivity is evaluated, demonstrating great potential for DCP detection applications.

In optical fibers, light travels through total internal reflection, while the EWs are oriented perpendicular to the interface. These electric fields exhibit rapid attenuation in the direction normal to the interface, and the range of their propagation can be characterized by the penetration depth dp: dp=λ2π1nco2sin2θ−ncl2,(1)where λ is the excitation wavelength. θ represents the incident angle of the core and cladding interface. nco and ncl are the RI of the fiber core and cladding, respectively. The proposed sensor probe is based on the principle of LSPR. Under the influence of the electric field present in GNRs, the electron cloud can initiate a collective oscillation, thereby generating surface plasmon waves (SPWs). Due to the size constraints of GNRs, SPWs are confined to the nanometer scale. Once the GNRs are fixed onto the sensor surface, the LSPR is activated by EWs stimulation, causing pronounced absorption at certain wavelengths. Changes in the RI cause shifts in the position of the peak wavelength [25]: Δλ=mΔnm(1−e−2ddp).(2)In this context, m and d represent the sensitivity of the GNRs to the electromagnetic field and the thickness of the electromagnetic layer, respectively. The presence of DCP polyclonal antibodies on the sensor surface enables the detection of DCP solutions across a range of concentrations, which in turn induces varying RI changes. Consequently, variations in concentration will result in distinct wavelength shifts, and these spectral shifts can be recorded by a spectrophotometer [26].

The LoD is a key parameter for assessing the performance of sensor devices, which is determined by the detection sensitivity and resolution of sensor probes. The formula for calculating the LoD is LoD=RS=3σS=σampl−noise2+σtemp−induced2+σspect−res23S,(3)σampl−noise≈FWHM4.5×(SNR)0.25,(4)σspect−res=RW23,(5)σtemp−induced=10fm.(6)Among them, S represents the detection sensitivity, and R represents the resolution of the sensor probe. According to Ref. [27], the resolution of the sensor is mainly related to the detection σampl−noise, the change in RI σtemp−induced, as well as the variability in the spectral resolution of the detection device’s standard deviation. Rw is the wavelength scanning resolution of the optical spectrum analyzer (OSA 0.1 nm). Given that a single noise source is statistically independent, the conventional criterion for resolution is set at three times the standard deviation of the overall noise σ [20]. R can be expressed by Eq. (3). The standard deviation of the signal peak position due to detection noise can be approximated by Eq. (4), assuming that the thermally stable laser sensor and interferometer have a standard deviation σtemp−induced of about 10 fm, which can be negligible in the calculation process.

The developed fiber structure utilized a combination of MMF and MCF. Compared to single-mode fibers, MMF has a larger core diameter, allowing more light to pass through. The MCF employed in the study featured a seven-core configuration, comprising seven identical germanium-doped silica strands, arranged with a central strand encircled by six others in a hexagonal pattern. The incorporation of multiple cores enhances the flexibility of waveguide degrees of freedom, thus conferring MCF-based sensors with enhanced sensing capabilities [28]. The fusion of MMF with MCF can leak a portion of the light energy to enhance EWs. However, the energy in this portion may not be sufficient to significantly enhance the sensitivity of the sensor. Therefore, a tapering process on the MCF section was performed. The taper region can greatly enhance the intensity of the EWs. The W-shaped taper region allows for sufficient leakage of light while ensuring that a considerable amount of energy can return to the fiber. This structure can enhance the EW and thus enhance the sensitivity of the sensor.

An amplified spontaneous emission (ASE) light source and an Er-doped fiber laser were utilized for testing RI changes of fiber probes without GNRs first. A 980 nm central wavelength laser served as the pump source, while a 1550 nm wavelength division multiplexing (WDM) component was utilized to amalgamate and differentiate various light wavelengths within the ring resonator setup. An erbium-doped fiber (EDF) with a low dopant concentration was chosen to function as the amplification medium. To guarantee the unidirectional propagation of light, a 1550 nm fiber isolator was integrated into the ring cavity. A polarization controller (PC) was connected between the isolator and the sensor probe to control the polarization state. Subsequently, a 90:10 coupler at 1550 nm was utilized to bifurcate the cavity’s light into two beams. The port with 10% of the output light was connected to an OSA, while 90% of the light continued to propagate within the ring cavity. To reduce the effects of external environmental disturbances on the ring cavity laser’s output, the entire laser assembly was mounted on an optical table designed to dampen vibrations. The layout of the laser ring cavity is depicted in Fig. 1.

In order to evaluate the functionality of the ring laser, several tests were conducted. The fiber probe was first tested by an ASE light source, and deionized water was added to the sensor; the results are shown in Fig. 2 with the black line. After setting the pump source power to 200 mW, the constructed fiber ring cavity was connected to the sensor probe, and the output spectrum is shown in Fig. 2 with the red line. Compared to the interference spectrum (FWHM=10nm), the FWHM of the laser fiber ring cavity is narrower (0.11 nm), which is improved by nearly two orders of magnitude. Moreover, the ER and SNR have been enhanced to a large, certain extent.

To assess the refractive index (RI) sensing capabilities of the sensing system, a series of measurements was undertaken. We prepared sodium chloride solutions with different concentrations, with RIU values ranging from 1.3340 to 1.3480. The experimental setup is shown in Fig. 1, where the solution was added to the sensor from low to high concentration. To ensure the reliability of the experimental results, deionized (DI) water was used to clean the sensor before each change in concentration. The experimental results are shown in Figs. 3(a) and 3(b), where the laser emission wavelength appears to red shift as the solution RI is changed from 1.3340 to 1.3480. As can be seen from Fig. 3(c), both sensors exhibit good linearity of fit, and the experimental results are consistent with the theoretical analysis. The sensitivity of the sensors with GNRs was 148.6 nm/RIU, while that of the sensors without GNRs was 144.5 nm/RIU, indicating a slight improvement in the sensitivity of the GNR biosensors. Meanwhile, the SNR and stability test results of the two sensors are shown in Fig. 3(d). The SNR of biosensors without and with GNRs were 41 dB and 48 dB, respectively, demonstrating that the presence of GNRs enhances the SNR of the laser spectrum and significantly improves stability. GNRs can enable the fiber sensor to generate the local surface plasmon resonance (LSPR) effect near 1550 nm. The changes in the LSPR signal instantaneously respond to the dynamic processes such as adsorption, binding, or dissociation of biomolecules on the sensor surface. When biomolecules approach and bind to specific antibodies on the sensor surface, the original response signal can be enhanced by LSPR. A stronger detection signal helps to increase the signal intensity. Under the condition that the noise level is relatively stable, the increase in signal intensity directly leads to the improvement of the signal-to-noise ratio, which is consistent with the increase in the signal-to-noise ratio of the sensor decorated with GNRs that we have observed.

The DCP solutions with a concentration range of 0–1000 ng/mL were prepared for testing, and the shift of laser wavelength was recorded. To reduce experimental errors, the DCP concentration was gradually increased from low to high, and the probe was rinsed with PBS before each concentration change. Due to the presence of DCP antibodies on the probe, the RI around the fiber probe changes with the concentration of the DCP solution. The experimental results are shown in Fig. 4(a). It can be observed that with the increase in DCP concentration, there is a red shift in the peak wavelength. The results of data analysis are shown in the inset of Fig. 4(b). The peak wavelength tends to be stable when the detection concentration of DCP is 1000 ng/mL. However, the relationship between the wavelength shift and the logarithm of the concentration exhibits a good linear relationship (R2=0.9848), as shown in Fig. 4(b). By analyzing the fitted data, the sensitivity of the fiber probe is 0.265 nm/lg(ng/mL). The resolution R and LoD of the fiber probe can be calculated to be 0.09726 nm and 367.6 pg/mL, respectively. If an ASE light source with an FWHM of 10 nm and an SNR of approximately 12 dB is used for DCP detection, the calculated LoD is 4.25 ng/mL. It can be seen that the LoD of the biosensors based on ring lasers is approximately 10 times higher when using ASE sources. The compared results show that biosensors based on light sources with narrow FWHM possess lower LoD. Additionally, due to the combination of special fiber structure and GNRs signal enhancement technology, the fiber probe developed on the basis of a fiber laser source has obtained excellent sensing performance.

The practical application capabilities of the developed probe were evaluated through tests of repeatability, reproducibility, selectivity, and pH. Reproducibility experiments were conducted twice for 0 and 1000 ng/mL DCP solutions. The spectrum in Fig. 5(a) shows that the spectra remained essentially unchanged during the two test processes, indicating that the developed probe has good repeatability. The reproducibility test assesses if probes of identical construction exhibit consistent sensing behavior. For this purpose, three probes were fabricated and used to analyze a 1000 ng/mL DCP solution, with the outcomes presented in Fig. 5(b). All three probes showed congruent peak wavelengths at the 1000 ng/mL concentration, indicating the probes’ high reproducibility.

Due to the presence of DCP polyclonal antibodies on the probe surface, the fiber probe has the capability for specific selection. The solutions of carcinoembryonic antigen, bovine serum albumin, bovine serum globulin, human serum albumin, and myoglobin at concentrations of 0 ng/mL and 1000 ng/mL were prepared for selectivity testing. The results in Fig. 6(a) show that when the detection target is a solution of another substance, there is almost no change in wavelength shift. This suggests that the sensor probe is capable of fulfilling the requirements for targeted detections, free from interference by extraneous substances. The probe’s purpose is to quantify DCP in human serum, making it essential to ascertain that the PBS solution used provides the ideal testing conditions. To achieve this, either sodium hydroxide or acetic acid is added to deionized water to regulate the pH levels to 3, 6, 10, and 13. Subsequently, DCP solutions at concentrations of 0 ng/mL and 1000 ng/mL are evaluated under these varied pH conditions to establish the best-suited environment for the probe’s efficacy. The wavelength shifts detected by the sensor probe at 0 ng/mL and 1000 ng/mL are shown in Fig. 6(b). When the pH was 7.4, the maximum red shift in wavelength occurred, which means that the developed sensor can meet the requirements for detecting DCP in the human body.

The stability of fiber laser ring sensors with and without GNRs was tested. The stability analysis was performed by introducing DI water into the sensor. The experiments were conducted with the ambient temperature set at 25°C, and the laser’s spectral data were captured at 5-min intervals. In each measurement cycle, the spectrum’s peak wavelength and intensity were noted, and these scanning outcomes are depicted in Figs. 7(a) and 7(b). Upon analysis of the experimental results, as shown in Fig. 7(c), during the 20-min monitoring process, the laser spectrum with GNRs showed almost no change. The maximum shifts in the peak wavelength and intensity were 0.02 nm and 0.216 dBm, respectively. As illustrated in Fig. 7(d), the laser spectrum without the material exhibited slight changes, with the maximum shifts in the peak wavelength and intensity being 0.08 nm and 0.922 dBm, respectively. In conclusion, the fiber sensor with GNRs demonstrated better stability.

The performance of various sensors is shown in Table 1. Compared with biosensor data, the proposed fiber optic sensor based on a ring fiber laser has a lower LoD and wider detection range. The FWHM of the laser light source is only 0.11 nm. As can be seen from Eqs. (3) and (4), narrow FWHM can reduce the value of LoD. Similarly, the fiber biosensors with GNRs can excite LSPR, enhancing the SNR and improving the detection sensitivity, further reducing LoD. Our experiment improves the sensitivity of the sensor from many aspects and realizes the low LoD of the sensor. As described in the paper, the fiber biosensors can accurately detect low concentrations of DCP.Table 1.

Comparison of the Performance of This Sensor with Other Existing Sensors

In this study, a W-shaped DCP fiber probe based on MMF-MCF-MMF was developed. The GNRs were immobilized on the probe to excite the LSPR effect at near-infrared wavelengths, enhancing the sensitivity of the sensor. By integrating the fiber laser ring cavity system with the sensor probe, the FWHM was significantly reduced (0.11 nm), thereby reducing LoD of the developed fiber probe. Compared to the ASE light source, the LoD of the laser ring cavity is reduced by a factor of 10. The incorporation of DCP polyclonal antibodies into the probe design provides it with the capability for specific targeting, which mitigates interference from other biomolecules and facilitates the detection of DCP in the near-infrared spectral region. The developed sensor underwent DCP testing in the range of 0–1000 ng/mL, with a sensitivity and detection limit of 0.265 nm/lg(ng/mL) and 367.6 pg/mL, respectively. The experimental results indicate that the developed near-infrared ring cavity biosensor can meet the detection requirements for DCP over a broad range. Additionally, other characteristics of the sensor were evaluated, demonstrating the potential application of this sensor in the detection of liver cancer biomarkers. Furthermore, this experiment provides new insights for enhancing the capabilities of fiber optic biosensors.