Skip to main content
NCBI home page
Food Chemistry: X logoLink to Food Chemistry: X
. 2026 Feb 13;34:103665. doi: 10.1016/j.fochx.2026.103665

Non-destructive monitoring of cyenopyrafen residues on strawberry via 280 nm UV-excited fluorescence of co-formulated organic solvents

Zichen Huang a, Tetsuhito Suzuki b, Panintorn Prempree c,d, Ken Abamba Omwange e, Kazunori Ninomiya f, Yoshito Saito g, Takahiro Hayashi d, Ryohei Nakano h, Tetsuya Nakazaki h,i, Siyao Chen d,j,⁎, Naoshi Kondo d
PMCID: PMC12933749  PMID: 41756594

Abstract

Cyenopyrafen is widely used to control spider mites during strawberry cultivation, and its residues should be monitored non-destructively after harvest. This study explores a fluorescence imaging system for estimating cyenopyrafen residues from a commercial pesticide formulation (STARMITE®, Nissan Chemical co., Japan). Excitation-emission matrix (EEM) analysis guided the imaging setup using 280 nm UV light. While cyenopyrafen itself is non-fluorescent, persistent formulation components associated with the solvent system emitted strong blue fluorescence, serving as an indirect index of residue level. Fluorescence images of 71 samples were captured and correlated with HPLC measurements. A partial least squares regression (PLSR) model developed using 53 samples (<5 mg/kg) achieved R2 = 0.92, RMSE = 0.38, and a detection limit of 0.44 mg/kg. This method enables non-destructive, accurate, and cost-effective residue estimation and is suitable for high-throughput post-harvest screening. The approach also shows potential for broader application to other formulation-based pesticide detection tasks.

Keywords: Strawberry, Cyenopyrafen residue, EEM, Non-destructive, Fluorescence image

Highlights

  • •

    A non-destructive fluorescence imaging system was developed to estimate cyenopyrafen residues on strawberries.

  • •

    While cyenopyrafen itself is non-fluorescent, formulation-associated components in the commercial product exhibit strong blue fluorescence under 280 nm UV excitation.

  • •

    Formulation-associated fluorescence intensity showed a strong correlation with cyenopyrafen residue levels.

  • •

    A PLSR model based on blue channel intensity achieved high accuracy (R2 = 0.92, RMSE = 0.38 mg/kg, LOD = 0.44 mg/kg).

  • •

    The method enables real-time, low-cost, and high-throughput screening for pesticide residues in post-harvest handling.

Chemical compounds list

1. Cyenopyrafen PubChem CID: 18772482. Active pesticide ingredient studied in this work.

2. Acetonitrile (ACN) PubChem CID: 6342. Extraction solvent used for HPLC analysis.

3. Methanol PubChem CID: 887. Organic solvent used in mobile phase for chromatography.

4. p-Coumaric acid PubChem CID: 637542. Strawberry metabolite mentioned as a natural fluorescent compound.

5. Cyclohexanone (co-formulant candidate solvent) PubChem CID: 7968. Discussed as a typical solvent used in β-ketonitrile formulations.

6. N-Methyl-2-pyrrolidone (NMP) PubChem CID: 13387. Discussed as a typical formulation solvent with known UV absorption and fluorescence behavior.

1. Introduction

Strawberry fruits (Fragaria × ananassa Duch.) are popular worldwide for their rich antioxidants (Ayala-Zavala et al., 2004) and distinctive flavor (Hikawa-Endo, 2020). Pesticide residues on strawberry fruits persist due to their fuzzy surface and large surface area, which promote pesticide adhesion. As a result, residue levels often exceed regulatory limits, leading to failed pesticide residue tests (You et al., 2020). In Japan, consumers prefer to eat fresh, unwashed strawberry fruits. Therefore, ensuring that pesticide residues remain within maximum residue limits (MRLs) is crucial for food safety, and non-destructive sensing methods playing a particularly important role in this process.

Among the pesticides used during the cultivation of the strawberry fruits in Japan, cyenopyrafen (C24H31N3O2, structure shown in Fig. S1) (Yu et al., 2012) is commonly applied for spider mite control, which was developed and commercialized by Nissan Chemicals Co. in 2009, and is also widely used in many other countries (Liu et al., 2019). However, Cyenopyrafen has demonstrated toxic effects in animal studies, including liver, kidney, uterine, and eye toxicity, as well as reproductive issues such as reduced implantation and increased incidence of uterine tumors in rats. These health risks underscore the importance of strict residue control (Food Safety Commission of Japan, 2023). In Japan, the acceptable daily intake (ADI) for cyenopyrafen is set at 0.05 mg/kg body weight, corresponding to a maximum of 3 mg per day for an average adult (Food Safety Commission of Japan, 2011). Studies have shown that cyenopyrafen residues on strawberries are stable, with less than 5% degradation during long-term storage, and the residues are located primarily on the surface of the fruits (Li et al., 2021). For strawberry-exporting countries in East Asia, compliance with varying maximum residue limits (MRLs) is necessary. The MRLs for cyenopyrafen are 3 mg/kg in Japan (Search Engine for MRLs of Agricultural Chemicals in Foods, 2010), 2 mg/kg in Korea (Kim et al., 2017), and 0.1 mg/kg in Canada (Consolidated Federal Laws of Canada, Pest Control Products Act, 2020). Hence, pre-export sampling and residue testing are essential. Currently, cyenopyrafen residues are detected using high-performance liquid chromatography (HPLC) (Liu et al., 2019) or ultra-performance liquid chromatography/tandem mass spectrometry (UPLC-MS/MS) method (Tan et al., 2018). Although accurate, these methods are destructive time-consuming, and expensive. Given the perishable nature and high market value of fruits like strawberries, non-destructive measurement of pesticide residues is essential to preserve sample integrity and enable rapid, large-scale screening (Hu et al., 2023; Kanmani et al., 2020). This highlights the need for a non-destructive, real-time sensing method to estimate cyenopyrafen residues efficiently.

Several non-destructive sensing methods for estimating pesticide residues have been reported. Previous studies have demonstrated that specific pesticide residues in agricultural products can be estimated using visible and near-infrared spectroscopy with a measurement error of 1.48 mg/kg, as well as Raman techniques with a measurement error of 6.69 mg/kg (Jamshidi et al., 2016). Moreover, hyperspectral imaging has been applied to detect specific pesticide residues by exploiting spectral information across multiple wavelength bands (Dong et al., 2014). Other potential methods for pesticide residue detection include terahertz spectroscopy (Suzuki et al., 2011) and plasmon resonance based sensors (Rajan et al., 2007; Saylan et al., 2017).

As a competitive alternative, fluorescence imaging systems offer the advantages of cost-effectiveness and real-time measurement. At its simplest, it can be constructed with an ultraviolet (UV) LED and standard imaging devices such as a digital color camera or a smartphone. Fluorescence is the emission of light by a substance at a longer wavelength than the excitation light. This technique is widely used for detecting agricultural by-products (Huang, Takemoto, et al., 2023).

This study focuses on the pesticide STARMITE® (Starmite 30 SC by Nissan Chemical), which is widely used in Japan and has been adopted in multiple countries worldwide following regulatory registration. The pesticide was selected from a preliminary screening of approximately 94 commonly applied compounds, representing about 20% of those registered in Japan. STARMITE® is a suspension concentrate (SC) containing approximately 30% cyenopyrafen, with the remainder comprising water, surfactants, and undisclosed organic “functional solvents” (Lewis et al., 2016). While the exact composition of these solvents is confidential, preliminary tests revealed that the formulation exhibits fluorescence under UV excitation. This finding suggested the feasibility of using a fluorescence-based method to estimate cyenopyrafen residues in treated agricultural products.

The objective of this research is to explore feasibility to monitor and quantitatively estimate cyenopyrafen residues in strawberry fruits based on fluorescence characteristics of pesticides. In this research, the fluorescence characteristics of both strawberry fruit and pesticide were measured. Experiments were conducted with different pesticide concentrations, followed by fluorescence imaging. The exact cyenopyrafen residue levels in each sample were quantified using HPLC, and a partial least quares regression (PLSR) model was developed using color parameters from fluorescence images to estimate the residue concentrations.

2. Materials and methods

2.1. Materials

The strawberry fruits (cultivar. ‘Sagahonoka’) were cultivated in a table-top system. The pesticide STARMITE, produced by Nissan Chemical Co. (Tokyo, Japan) was sprayed one day before harvest when the strawberry fruits had reached market-ready maturity (> 80% red color) (Hayashi et al., 2014). In addition to the active ingredient cyenopyrafen, the pesticide formulation included Mairino (マイリノー), a spreading agent produced by Nichino, Japan, used at the recommended concentration of 0.5 mL/kg. Spreading agents, primarily composed of surfactants, reduce the surface tension of chemical solutions, thereby improving pesticide coverage and effectiveness (Myers, 2020). The spreading agent was confirmed not to exhibit fluorescence emission in the visible region. Six pesticide concentrations (0.25, 0.5, 2.5, 3.75, 5, and 7.5 mL/kg) were prepared, to generate different levels of residues on the strawberry fruit, with 0.5 mL/kg representing the recommended application rate. Sixteen strawberries were treated with the 0.5 mL/kg concentration, while ten strawberries were assigned to each of the other concentrations. The pesticide solution was applied using a hand sprayer at a distance of 5 to 15 cm, with two sprays per fruit at a 90° angle interval at a distance of 10 cm to create uniform residue distribution. Then strawberry fruits were harvested at 10:00 am of 0 day after harvest (DAH), which was about 24 h after pesticide application. This timing was designed to allow sufficient spreading and initial stabilization of pesticide residues on the fruit surface, while minimizing additional environmental variability prior to subsequent measurement. In addition to the 66 strawberry fruits with pesticide treatment, another 5 strawberry fruits without pesticide treatment were harvested one day after pesticide treatment as the control group. Two hours after harvest, all fruits were transported to an experimental room at Kyoto University and stored in a cool incubator (A1201, ASONE. Japan) at 10 °C and 80% humidity. Finally, the fluorescence characteristics of these 71 strawberry fruits were obtained using the proposed machine vision system for building the cyenopyrafen estimation model.

2.2. EEM and HPLC measurement

To characterize the fluorescence excitation-emission matrices (EEM) of the pesticide and strawberry fruits for building a suitable machine vision system, typical cyenopyrafen pesticide and strawberry fruit samples were measured by a spectrofluorometer (Jasco FP-8300, JASCO Inc., Japan). EEMs provide a three-dimensional dataset comprising excitation wavelength, emission wavelength, and fluorescence intensity (Al Riza et al., 2019). The excitation and emission ranges of the spectrofluorometer were set to 200–450 nm, and 210–550 nm, respectively. The photomultiplier (PMT) tube was set to a medium sensitivity. These settings were decided after examining the pre-experiment results. The resolution of the measurements was 5 nm, with a scan speed of 5000 nm/min and a response time of 50 ms. The EEM of the samples were measured using a front-face method, which is suitable for solid samples. The strawberry fruit surface sample was cut into an 18 mm diameter section to be placed into a sample holder (Fig. 1(a)). The sampling points were selected based on fluorescence images, focusing on areas exhibiting strong blue fluorescence under UV light at a wavelength of 280 nm. Pesticide and spreading agent used in this experiment were placed into a quartz cuvette for EEM measurement (Q-204, ASONE Inc., Japan).

Fig. 1.

Fig. 1

Open in a new tab

Experimental workflow and equipment used for strawberry fruit analysis: (a) EEM measurement, (b) HPLC measurement and (c) Image acquisition system.

The amount of cyenopyrafen residue in strawberry fruits was measured by HPLC. The HPLC procedure and settings followed that established by the India Ministry of Agriculture & Farmers Welfare (MAFE, 2020) for cyenopyrafen analysis. Prior to measurement, a calibration was performed using standard cyenopyrafen samples. SA-certified standard cyenopyrafen (99.9% purity) was acquired from Wako Pure Chemical Japan (Osaka, Japan). HPLC-grade acetonitrile (ACN) and methanol were purchased. The water was purified before use by a Milli-Q instrument (TANKMPK01, Merck Millipore, USA). In order to prepare a 20 mg/L standard stock solution, 2.00 mg of cyenopyrafen sample was dissolved in 100 mL of solvent. A 20-mg/L working solution was prepared from this stock solution, and a calibration curve was constructed by diluting the working solution to obtain the following concentration levels of 0.05, 0.1, 0.2, 0.5, 1.0, 2.0, and 5.0 mg/L, a practice used in previous research for measuring cyenopyrafen residue (Kabir et al., 2017).

The cyenopyrafen extraction method follows to previous researches on cyenopyrafen residue analysis (Fig. 1(b)) (Kabir et al., 2017; Li et al., 2021). First, a strawberry fruit was pulverized using a glass rod. Then, 5 g homogenized strawberry fruit sample and 30 mL of acetonitrile were added to a 50-mL Teflon tube. The mixture was shaken for 15 min using a vortex mixer (Vortex-2 Genie, Scientific Industries, USA). Next, a 15 mL aliquot was transferred to the 15-mL Teflon tube and centrifuged at 6000 rpm for 5 min. The supernatant was filtered through a 0.22 μm nylon syringe filter and transferred to a screw-cap vial for HPLC analysis. The vial was stored at −60 °C awaiting HPLC measurement.

A Shimadzu HPLC (SCL-10AVP; Shimadzu, Tokyo, Japan) system, equipped with an autosampler (SIL-20 AC), a column oven (CTO-20 AC), a dual pump (LC-20 CE), and an ultraviolet (UV)–vis detector (SPD-20 A), was used for liquid chromatography (LC) analysis. Data acquisition and processing were performed using Chromato-PRO software. For chromatographic separation, the mobile phase was made up of acetonitrile (55%), water (25%), and methanol (20%) after 15 min degas using an ultrasonic cleaning (US CLEANER, Asone, Japan). For the HPLC settings, the injection volume was 5 μL, and the flow rate of the mobile phase was 1 mL/min. The analyte was detected at a wavelength of 280 nm. Separation was performed using a C18 column (4.6 × 150 mm, 5 μm particle sizes, Shim-pack Scepter C18–120, Shimadzu, Japan) maintained at 40 °C.

2.3. Image acquisition and processing

The color and UV-induced fluorescence images of strawberry fruits were captured prior to HPLC measurement. The image acquisition system consisted of a white LED (LDL2-80X16SW2, CCS Inc., Japan) and a 280 nm ring LED (LDR2-100UV2–280-W, CCS Inc. Japan) (Fig. 1(c)). In this research, the 280 nm light served as a preventative measure to minimize the emission of blue fluorescence originating from p-coumaric acid, which tends to accumulate in overripe strawberries when excited within the 310 to 398 nm range (Huang et al., 2022). A USB camera (DFK 23 U445, Imaging Source Co. Ltd., Germany) with UV cut filter (50% transmittance at 390 nm, Sony Inc., Japan) was positioned 300 mm above the strawberry fruits. The white balance of the camera was calibrated using a whiteboard before image acquisition. To prevent external light interference, the experiment was conducted in a dark room, and the image acquisition system was enclosed with a UV-cut curtain for operator protection. For each strawberry fruit, four side views were captured at 90° intervals, recording both color and UV fluorescence images at each angle. For color images, the camera was set to a gain of 1 dB and an exposure time of 0.008 s. For UV images, the camera was set to a gain of 18.75 dB and an exposure time of 0.5 s.

After acquiring images of the strawberry fruit, image processing was performed using MATLAB (Matlab, 2018b, MathWorks. USA). The fruit was extracted from the background using Otsu algorithm applied to the R channel of the color image. The region of interest (ROI) was identified as the entire strawberry surface segmented from the background in the color images. Based on the extracted ROI, the corresponding region in the UV fluorescence images was determined, and the average R, G, and B pixel values and the corresponding intensity ratio (ratioR, ratioG, ratioB) were calculated (Eq.1). In addition, the average value of HSV color space and La*b* were extracted for building a non-destructive estimation model based on PLSR. The PLSR model utilized 12 principal component scores (R, G, B, H, S, V, L, a*, b*, ratioR, ratioG, ratioB). A 5-fold cross-validation was used in the construction of the PLSR model to prevent over-fitting. The performance of the prediction model was evaluated using the coefficient of determination (R2), and root mean square error (RMSE) (Forouzangohar et al., 2008). To assess the efficacy of the suggested models, an established metric known as the performance-to-deviation ratio (RPD) was integrated (Cortés et al., 2017; Huang, Omwange, et al., 2023; Williams et al., 2017). This metric functions as a criterion for evaluating the effectiveness of the PLSR model. Generally, a satisfactory model is anticipated to manifest an RPD value exceeding 2.5, while values surpassing 3.0 are regarded as indicative of excellent performance. An effective model is characterized by higher R2 and RPD values coupled with a lower RMSE.

ratioR=RaverageRaverage+Gaverage+Baverage
ratioG=GaverageRaverage+Gaverage+Baverage
ratioB=BaverageRaverage+Gaverage+Baverage (1)

where, Raverage, Gaverage and Baverage are the average R, G, and B pixel value.

3. Results and discussion

3.1. EEM and HPLC results

Typical EEM results are shown in Fig. 2, where the x-axis is emission wavelength (nm), the y-axis is excitation wavelength (nm), and the color bar is intensity in Raman Units. The EEM of the pesticide alone (Fig. 2(a)) (at a recommended concentration of 0.5 mL /L) shows the pesticide fluorescence emission between 400 nm to 500 nm when excited using wavelength ranging from 240 nm to 350 nm. In this experiment, the USB camera detects the fluorescence emission through its blue and green channels. To clarify the origin of the observed fluorescence, the EEM of 5 mg/L pure cyenopyrafen in acetonitrile was also measured; however, cyenopyrafen itself exhibited no fluorescence (Fig. 2(d)), confirming that the observed emission originates from formulation components. Due to the proprietary nature of the formulation, the exact fluorescent species in STARMITE® cannot be explicitly identified. As a β-ketonitrile derivative, cyenopyrafen is formulated as a suspension concentrate using organic solvents such as cyclohexanone and N-methyl-2-pyrrolidone (NMP), together with surfactants and stabilizing auxiliaries. While these solvents mainly act as dispersion media and are unlikely to contribute directly to blue-region fluorescence, certain persistent formulation-associated components may fluoresce under UV excitation and thus serve as indirect indicators of residue concentration. Fluorescence excited at approximately 280 nm with emission in the 400–500 nm range is commonly associated with π–π electronic transitions in aromatic or conjugated molecular systems (Lakowicz, 2006). In commercial suspension concentrate formulations, dispersants and stabilizing auxiliaries often contain substituted aromatic or heteroaromatic structures with electron-donating functional groups, which are known to contribute blue-region fluorescence under UV excitation (Ohkouchi & Tsuji, 2022).

Fig. 2.

Fig. 2

Open in a new tab

Typical EEM spectra of (a) pesticide, (b) strawberry fruit, (c) strawberry fruit with pesticide, and (d) cyenopyrafen solution.

The typical EEM of strawberry fruit EEM (Fig. 2(b)) reveals a single fluorescence peak within the measured range, with no detectable emissions in the visible spectrum (400–700 nm). The detected fluorescence peak is suspected to be associated with amino acids (Sikorska et al., 2020; Zhang et al., 2011). The strawberry fruit with pesticide exhibits two distinct peaks (Fig. 2(c)): one corresponding to pesticide and the other to the strawberry flesh. This unique pesticide fluorescence signature allows for residue detection in strawberry fruit, which could be integrated into a machine vision system with an excitation light source between 240 nm to 350 nm.

The retention time of cyenopyrafen in HPLC measurement was approximately 13.6 min, consistent with a previous report (MAFE, 2020). As shown in Fig. S2, chromatograms of residue-free strawberry fruit flesh and samples containing cyenopyrafen residues at 0.53 mg/kg, 1.57 mg/kg, and 3.10 mg/kg exhibit a clear single peak at this retention time. To quantify residue levels, a calibration method was applied following previous research (Kabir et al., 2017). A linear calibration curve was obtained with the equation y = 0.0004x - 0.0463 (R2 = 1.0000), where x is the peak area and y is the cyenopyrafen residue (mg/L). These results confirm the clarity and effectiveness of the HPLC method for residue determination in this study.

3.2. Pesticide residue estimation using fluorescence image

The color and UV images of strawberry fruits from four sides at 90° intervals are shown in Fig. 3. The first sample is the residue-free strawberry fruit. Without pesticide treatment, there is a weak blue halation area caused by reflection. This artifact could potentially be reduced in future studies by incorporating a polarizing filter into the machine vision system (Cheng et al., 2015). The cyenopyrafen residues levels of the other three samples were 1.065, 2.717, and 4.873 mg/kg. As cyenopyrafen residue increased, the blue color intensity in the UV images increased. Given that cyenopyrafen residues remains stable during storage (Li et al., 2021) and over 95% persist on the strawberry surface even after two weeks (Food and Agricultural Materials Inspection Center (FAMIC), 2005), images taken one day after harvest can reliably indicate the residue levels. Moreover, in Japan, fruit grading agencies typically assess strawberry quality one day post-harvest, making the proposed method well-suited for automated residue detection during the grading process.

Fig. 3.

Fig. 3

Open in a new tab

Representative color and UV images of strawberry fruit.

To further evaluate the temporal stability of the proposed method within a practical post application period, cyenopyrafen residue levels were further estimated at 1, 3, 5, 7, and 9 days after spraying using the established PLSR model based on UV image analysis (Fig. S3; n = 6 per time point). No significant temporal variation was observed over this period, and no statistically significant difference was found between day 1 and day 9 (p = 0.547), indicating that the fluorescence-based estimation remains stable after the initial residue stabilization.

For the coefficient of determination analysis, the average pixel value of the B channel within the strawberry fruit ROI was evaluated, based on EEM results indicating blue fluorescence emission from the pesticide under 280 nm excitation. A strong correlation was observed between cyenopyrafen residue levels and the B channel pixel value, with a coefficient of determination of 0.88 (Fig. 4(a)). This result demonstrates that blue-channel fluorescence intensity provides a reliable indicator for estimating cyenopyrafen residues. The observed relationship is consistent with the fluorescence response shown in the UV images (Fig. 3).

Fig. 4.

Fig. 4

Open in a new tab

Coefficient of determination (R2) between pesticide residue levels and B channel pixel value using (a) linear regression, (b) curve regression, and (c) short-scale (0–5 mg/kg) regression.

A slight change in the shape of the increasing trend can be observed between the residue from 0 to 5, and 5 to 10 mg/kg, which is attributed to partial saturation of the fluorescence intensity in the images. As the pesticide concentration on the strawberry surface increases, the captured images appear brighter. However, at higher residue levels, fluorescence intensity approaches saturation, resulting in a slower increase in the B channel pixel values. A curve regression using second-order polynomials had a higher coefficient of determination (Fig. 4(b)). Alternatively, when the regression is restricted to a narrower residue range of 0–5 mg/kg, a strong linear relationship is maintained, with a coefficient of determination of 0.87 (Fig. 4(c)). Accordingly, for the subsequent PLSR analysis, regression results for the 0–5 mg/kg and 0–10 mg/kg ranges are discussed separately.

3.3. PLSR model for estimating cyenopyrafen residue

Two PLSR models were developed to predict cyenopyrafen residue on the strawberry fruit using 12 PLS components. The best performance was achieved using all components, with ratio R, ratio G, and ratio B identified as the top three contributing variables. The first model, built with 71 samples covering residue levels from 0 to 10 mg/kg, achieved an RMSE of 0.84 and R2 0.90 (Fig. 5(a)). The second model uses 53 samples with residues within range from 0 to 5 mg/kg, exhibited an RMSE of 0.38 and R2 of 0.92 (Fig. 5(b)). The RPD values for the first and second models were 3.22 and 3.58, both exceeding the commonly accepted threshold of 3 for reliable prediction. Notably, the RMSE of less than 1 mg/kg meets the standard for practical application (Dhakal et al., 2014). The limits of detection (LOD) was calculated as 0.44 mg/kg using the formula LOD = 3σ/S, where σ is the standard deviation of the blank and S is the average slope of the model. This LOD suggest the proposed fluorescent imaging system has potential for classifying strawberry fruit samples in accordance with market standards of Japan (3 mg/kg) and Korea (2 mg/kg), though it may not meet the stricter requirements of markets such as Canada (0.1 mg/kg).

Fig. 5.

Fig. 5

Open in a new tab

Regression results for pesticide residue in the ranges of (a) 0–10 mg/kg and (b) 0–5 mg/kg. (c) Distribution of fluorescence peak positions for 94 commercially available pesticides.

Overall, both models demonstrated robust performance, and the second model confirmed that excluding high-residue samples can yield more consistent predictions by reducing potential image saturation. These results highlight the practical utility of the proposed fluorescence-based system for residue estimation.

A previous report by Nissan Chemical Co. investigated the presence of cyenopyrafen on the surface, peel and pulp of various fruits and vegetables, such as oranges, eggplants and strawberry fruits using a carbon-14 labelling method (Food and Agricultural Materials Inspection Center (FAMIC), 2005). The findings indicated that over 95% of the cyenopyrafen residues were located on the surface and could be absorbed by the surface cleaning solution during the experiment. These results suggest that the proposed machine vision system has the potential to be applied not only to strawberries but also to other produce such as oranges and eggplants.

4. Discussion

The developed fluorescence imaging system demonstrates potential for non-destructive, indirect quantification of pesticide residues. When a specific pesticide formulation is pre-calibrated, quantitative estimation becomes feasible. Given that STARMITE® has been widely used in Japan since 2010, the proposed approach offers practical value for farmers and producers in managing and grading pesticide-treated fruits. Moreover, the system could be adapted for other fruits or vegetables that exhibit blue fluorescence under 280 nm excitation. In our broader screening of 94 commonly used pesticides, the majority exhibited detectable intrinsic fluorescence across a broad range of excitation and emission wavelengths (Fig. 5(c)). A small subset showed weak responses at short excitation wavelengths around 230 nm, with intensities close to the background level. Overall, these results highlight the feasibility of designing multi-wavelength fluorescence imaging systems tailored to different pesticide types.

The proposed method requires no chemical pretreatment and enables non-destructive measurement of intact fruits, with a total analysis time of approximately 2 min per sample. In contrast, conventional HPLC analysis involves multiple preparation steps and chromatographic separation, resulting in a total analysis time of approximately 40 min per sample. Although HPLC provides direct and highly accurate chemical quantification and remains indispensable for confirmatory analysis, the imaging-based approach offers sufficient predictive accuracy for screening purposes, as demonstrated by its strong correlation with HPLC reference measurements. A simplified comparison of workflow, analysis time, and equipment cost between the two methods is presented in Table 1.

Table 1.

Performance comparison between UV fluorescence imaging and HPLC.

Parameter UV fluorescence imaging
(this study)		 Conventional HPLC
Sample preparation time 0 min ∼25 min
Instrumental analysis time ∼2 min ∼15 min
Total time per sample ∼2 min ∼40 min
Measurement mode Non-destructive, quasi real time Destructive, off-line
Quantification accuracy Indirect estimation (PLSR); R2 = 0.92, RMSE = 0.38 mg/kg Direct chemical quantification; reference method
Role in residue analysis High throughput screening / ranking Confirmatory and regulatory analysis
Typical equipment cost ∼102 USD ∼105 USD
Open in a new tab

These features make the proposed approach well suited as a front-end screening tool for rapid pesticide residue assessment, where speed, throughput, and low equipment cost are critical. For industrial implementation, system performance could be further improved by reducing surface halation using polarizing filters and by integrating multiple cameras and light sources with automated image acquisition and processing, enabling high-throughput, quasi-real-time, and continuous non-destructive inspection on production lines.

Although identifying specific fluorescent compounds was beyond the scope of this study, the observed correlation between fluorescence intensity and cyenopyrafen residue underscores the usefulness of co-formulated solvents as practical residue indicators. In commercial suspension concentrate products such as STARMITE®, regulatory control of the declared active ingredient content constrains batch-to-batch variation, supporting the practical applicability of formulation-specific calibration models. In addition, the temporal stability of formulation-associated fluorescence signals warrants further quantitative investigation through extended observation periods and larger sample sets, toward the long-term reliability of fluorescence-based sensing platforms.

Future work may focus on constructing simplified model formulations or reference solvent systems that do not correspond to proprietary commercial products, combined with complementary analytical techniques such as GC–MS, to further elucidate the origin of the fluorescence signal and its interaction mechanisms. In parallel, establishing a comprehensive database of fluorescence properties for pesticides and agricultural products would support the rational selection of excitation and emission wavelengths and improve the specificity and accuracy of fluorescence-based detection platforms. On this basis, integration of machine learning with spectral imaging data could further enhance residue classification and quantitative estimation in real-time applications.

5. Conclusions

This study developed a 280 nm UV-excited fluorescence imaging system to estimate cyenopyrafen residues on strawberry fruits. Fluorescence characteristics were examined using excitation–emission matrix (EEM) spectroscopy, revealing that cyenopyrafen itself is non-fluorescent, whereas distinct fluorescence originates from solvent-associated components in the commercial formulation. This feature enables indirect quantitative detection of residues. The partial least squares regression (PLSR) model using samples in the 0 to 5 mg/kg range achieved a root mean square error (RMSE) of 0.38 mg/kg and a coefficient of determination (R2) of 0.92, with a detection limit of 0.44 mg/kg, meeting the regulatory requirements of Japan and Korea. Given that the pesticide types used in farms and processing facilities are typically fixed and known in advance, this method supports practical quantitative estimation. The system shows strong potential for integration into production lines to enable rapid, non-destructive and cost-effective tools for pesticide residue monitoring.

CRediT authorship contribution statement

Zichen Huang: Writing – original draft, Visualization, Software, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization. Tetsuhito Suzuki: Writing – review & editing, Supervision, Resources, Project administration. Panintorn Prempree: Writing – review & editing, Validation, Methodology, Formal analysis. Ken Abamba Omwange: Software, Resources, Methodology, Investigation, Formal analysis, Data curation. Kazunori Ninomiya: Writing – review & editing, Supervision, Resources. Yoshito Saito: Writing – review & editing, Methodology, Investigation. Takahiro Hayashi: Writing – review & editing, Validation, Supervision, Investigation. Ryohei Nakano: Writing – review & editing, Supervision, Methodology, Conceptualization. Tetsuya Nakazaki: Writing – review & editing, Project administration, Investigation, Conceptualization. Siyao Chen: Writing – review & editing, Supervision, Methodology, Data curation. Naoshi Kondo: Supervision, Resources, Project administration, Investigation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgment

This research was supported by the Zhejiang Provincial Natural Science Foundation of China (Grant No. LQN25C130006) and the Grant-in-Aid for JSPS Fellows (Project No. 21F21397). The authors would like to thank Prof. Garry John Piller of Kyoto University for proof-reading this manuscript. We thank Nissan Chemicals Co. for their advice in this study. We thank the staff members Mr. Fumio Kishida of Experimental Farm of Graduate School of Agriculture, Kyoto University for managing the cultivation of the materials.

Footnotes

Appendix A

Supplementary data to this article can be found online at https://doi.org/10.1016/j.fochx.2026.103665.

Appendix A. Supplementary data

Supplementary material 1
mmc1.docx (115.6KB, docx)

Data availability

Data will be made available on request.

References

  1. Al Riza D.F., Widodo S., Purwanto Y.A., Kondo N. Combined fluorescence-transmittance imaging system for geographical authentication of patchouli oil. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2019;218:155–160. doi: 10.1016/j.saa.2019.04.009. [DOI] [PubMed] [Google Scholar]
  2. Ayala-Zavala J.F., Wang S.Y., Wang C.Y., González-Aguilar G.A. Effect of storage temperatures on antioxidant capacity and aroma compounds in strawberry fruit. LWT - Food Science and Technology. 2004;37(7):687–695. doi: 10.1016/j.lwt.2004.03.002. [DOI] [Google Scholar]
  3. Cheng X., Kondo N., Ogawa Y., Shiigi T., Widodo S., Watanabe Y., Fujiura T. A halation reduction method for high quality images of tomato fruits in greenhouse. Engineering in Agriculture, Environment and Food. 2015;8(4):200–206. doi: 10.1016/j.eaef.2015.10.001. [DOI] [Google Scholar]
  4. Consolidated federal laws of Canada, Pest Control Products Act 2020. https://laws-lois.justice.gc.ca/eng/acts/P-9.01/page-1.html
  5. Cortés V., Rodríguez A., Blasco J., Rey B., Besada C., Cubero S., Salvador A., Talens P., Aleixos N. Prediction of the level of astringency in persimmon using visible and near-infrared spectroscopy. Journal of Food Engineering. 2017;204:27–37. doi: 10.1016/j.jfoodeng.2017.02.017. [DOI] [Google Scholar]
  6. Dhakal S., Li Y., Peng Y., Chao K., Qin J., Guo L. Prototype instrument development for non-destructive detection of pesticide residue in apple surface using Raman technology. Journal of Food Engineering. 2014;123:94–103. doi: 10.1016/j.jfoodeng.2013.09.025. [DOI] [Google Scholar]
  7. Dong C., Ye Y., Zhang J., Zhu H., Liu F. Detection of Thrips defect on Green-Peel Citrus using hyperspectral imaging technology combining PCA and B-spline lighting correction method. Journal of Integrative Agriculture. 2014;13(10):2229–2235. doi: 10.1016/S2095-3119(13)60671-1. [DOI] [Google Scholar]
  8. Food and Agricultural Materials Inspection Center (FAMIC). (2005). Cyenopyrafen Pesticide Abstracts. https://www.acis.famic.go.jp/syouroku/cyenopyrafen/index.htm.
  9. Food Safety Commission of Japan . In: Food Health Impact Assessment of Cyenopyrafen (Evaluation Report (Pesticide) No. kya2010110002) Food Safety Commission of Japan, editor. 2011. https://www.fsc.go.jp/fsciis/evaluationDocument/show/kya20101110002 [Google Scholar]
  10. Food Safety Commission of Japan Chemical hazards toxicological evaluation database [database summary and risk assessment reference]. Food Safety Commission of Japan. 2023. https://www.fsc.go.jp/english/evaluationreports/
  11. Forouzangohar M., Kookana R.S., Forrester S.T., Smernik R.J., Chittleborough D.J. Midinfrared spectroscopy and chemometrics to predict diuron sorption coefficients in soils. Environmental Science & Technology. 2008;42(9):3283–3288. doi: 10.1021/es702750d. [DOI] [PubMed] [Google Scholar]
  12. Hayashi S., Yamamoto S., Saito S., Ochiai Y., Kamata J., Kurita M., Yamamoto K. Field operation of a movable strawberry-harvesting robot using a travel platform. Japan Agricultural Research Quarterly: JARQ. 2014;48(3) doi: 10.6090/jarq.48.307. [DOI] [Google Scholar]
  13. Hikawa-Endo M. Improvement in the shelf-life of Japanese strawberry fruits by breeding and post-harvest techniques. The Horticulture Journal. 2020;89(2):Article 2. doi: 10.2503/hortj.UTD-R008. [DOI] [Google Scholar]
  14. Hu Y., Ma B., Wang H., Li Y., Zhang Y., Yu G. Non-destructive detection of different pesticide residues on the surface of Hami melon classification based on tHBA-ELM algorithm and SWIR hyperspectral imaging. Foods. 2023;12(9):Article 9. doi: 10.3390/foods12091773. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Huang Z., Omwange K.A., Saito Y., Kuramoto M., Kondo N. Monitoring strawberry (Fragaria × ananassa) quality changes during storage using UV-excited fluorescence imaging. Journal of Food Engineering. 2023;353 doi: 10.1016/j.jfoodeng.2023.111553. [DOI] [Google Scholar]
  16. Huang Z., Omwange K.A., Tsay L.W.J., Saito Y., Maai E., Yamazaki A.…Kondo N. UV excited fluorescence image-based non-destructive method for early detection of strawberry (Fragaria × ananassa) spoilage. Food Chemistry. 2022;368 doi: 10.1016/j.foodchem.2021.130776. [DOI] [PubMed] [Google Scholar]
  17. Huang Z., Takemoto T., Omwange K.A., Saito Y., Kuramoto M., Kondo N. Macroscopic and microscopic characterization of fluorescence properties of multiple sweet pepper cultivars (Capsicum annuum L.) using excitation-emission matrix and UV induced fluorescence imaging. Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy. 2023;288 doi: 10.1016/j.saa.2022.122094. [DOI] [PubMed] [Google Scholar]
  18. Jamshidi B., Mohajerani E., Jamshidi J. Developing a Vis/NIR spectroscopic system for fast and non-destructive pesticide residue monitoring in agricultural product. Measurement. 2016;89:1–6. doi: 10.1016/j.measurement.2016.03.069. [DOI] [Google Scholar]
  19. Kabir M.H., El-Aty A.M.A., Kim S.-W., Rahman M.M., Chung H.S., Lee H.S.…Shim J.-H. Decline pattern and risk assessment of cyenopyrafen in different varieties of Asian pear using liquid chromatography and tandem mass spectrometry. Food Science and Biotechnology. 2017;26(2):537–543. doi: 10.1007/s10068-017-0074-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Kanmani R., Maheswar R., Sureshkumar A., Mugilan A., Praveena V. Non-destructive approach to detect pesticides in fruits and vegetables using IoT technology. International Conference on Computer Communication and Informatics (ICCCI) 2020;2020:1–8. doi: 10.1109/ICCCI48352.2020.9104166. [DOI] [Google Scholar]
  21. Kim Y.-J., Kim J.-H., Kwon Y.-S., Song J.-W., Seo J.-S. Residual characteristics and monitoring of Cyenopyrafen and Cyflumetofen in strawberries for export. Korean Journal of Environmental Agriculture. 2017;36(4):279–287. doi: 10.5338/KJEA.2017.36.4.35. [DOI] [Google Scholar]
  22. Lakowicz J.R. Springer US; 2006. Principles of fluorescence spectroscopy. [DOI] [Google Scholar]
  23. Lewis K.A., Tzilivakis J., Warner D.J., Green A. An international database for pesticide risk assessments and management. Human and Ecological Risk Assessment: An International Journal. 2016;22(4):1050–1064. doi: 10.1080/10807039.2015.1133242. [DOI] [Google Scholar]
  24. Li Y., Xu J., Zhao X., He H., Zhang C., Zhang Z. The dissipation behavior, household processing factor and risk assessment for cyenopyrafen residues in strawberry and mandarin fruits. Food Chemistry. 2021;359 doi: 10.1016/j.foodchem.2021.129925. [DOI] [PubMed] [Google Scholar]
  25. Liu S., Kou H., Mu B., Wang J., Zhang Z. Dietary risk evaluation of tetraconazole and bifenazate residues in fresh strawberry from protected field in North China. Regulatory Toxicology and Pharmacology. 2019;106:1–6. doi: 10.1016/j.yrtph.2019.04.008. [DOI] [PubMed] [Google Scholar]
  26. MAFE Determination of cyenopyrafen content [Online post] 2020. https://ppqs.gov.in/sites/default/files/cyenopyrafen.pdf
  27. Myers, D. (2020). Surfactant science and technology. John Wiley & Sons.
  28. Ohkouchi T., Tsuji K. Basic technology and recent trends in agricultural formulation and application technology. Journal of Pesticide Science. 2022;47(4):155–171. doi: 10.1584/jpestics.D22-055. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Rajan C., .S., Gupta B.D. Surface plasmon resonance based fiber-optic sensor for the detection of pesticide. Sensors and Actuators. B, Chemical. 2007;123(2):661–666. doi: 10.1016/j.snb.2006.10.001. [DOI] [Google Scholar]
  30. Saylan Y., Akgönüllü S., Çimen D., Derazshamshir A., Bereli N., Yılmaz F., Denizli A. Development of surface plasmon resonance sensors based on molecularly imprinted nanofilms for sensitive and selective detection of pesticides. Sensors and Actuators B: Chemical. 2017;241:446–454. doi: 10.1016/j.snb.2016.10.017. [DOI] [Google Scholar]
  31. Search engine for MRLs of Agricultural Chemicals in Foods 2010. http://db.ffcr.or.jp/front/food_group_detail?id=10100
  32. Sikorska E., Włodarska K., Khmelinskii I. Application of multidimensional and conventional fluorescence techniques for classification of beverages originating from various berry fruit. Methods and Applications in Fluorescence. 2020;8(1) doi: 10.1088/2050-6120/ab6367. [DOI] [PubMed] [Google Scholar]
  33. Suzuki T., Ogawa Y., Kondo N. Characterization of pesticide residue, cis-permethrin by terahertz spectroscopy. Engineering in Agriculture, Environment and Food. 2011;4(4):90–94. doi: 10.1016/S1881-8366(11)80007-8. [DOI] [Google Scholar]
  34. Tan H., Gu Y., Liu S., Zhang H., Li X., Zeng D. Rapid residue determination of Cyenopyrafen in Citrus Peel, pulp, and whole fruit using ultra-performance liquid chromatography/tandem mass spectrometry. Food Analytical Methods. 2018;11(8):2123–2130. doi: 10.1007/s12161-018-1197-y. [DOI] [Google Scholar]
  35. Williams P., Dardenne P., Flinn P. Tutorial: Items to be included in a report on a near infrared spectroscopy project. Journal of Near Infrared Spectroscopy. 2017;25(2):85–90. doi: 10.1177/0967033517702395. [DOI] [Google Scholar]
  36. You J.-S., Gwak H.-M., Chang H.-R. Comparison of residue patterns for systemic and non-systemic pesticides in strawberry. Korean Journal of Environmental Agriculture. 2020;39(4):305–311. doi: 10.5338/KJEA.2020.39.4.36. [DOI] [Google Scholar]
  37. Yu H., Xu M., Chen Y., Wu H., Luo Y., Li B. Synthesis and acaricidal activity of cyenopyrafen and its geometric isomer. Arkivoc. 2012;2012(6):26–34. doi: 10.3998/ark.5550190.0013.603. [DOI] [Google Scholar]
  38. Zhang J., Wang X., Yu O., Tang J., Gu X., Wan X., Fang C. Metabolic profiling of strawberry (Fragaria×ananassa Duch.) during fruit development and maturation. Journal of Experimental Botany. 2011;62(3):1103–1118. doi: 10.1093/jxb/erq343. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary material 1
mmc1.docx (115.6KB, docx)

Data Availability Statement

Data will be made available on request.

Articles from Food Chemistry: X are provided here courtesy of Elsevier

RESOURCES