Validation study on light scattering changes in kiwifruit during postharvest storage using time-resolved transmittance spectroscopy

Visible and near-infrared spectroscopy has been well studied for characterizing the organic compounds in fruit and vegetables from pre-harvest to late harvest. However, due to the challenge of decoupling of optical properties, the relationship between the collected samples’ spectral data and their properties, especially their mechanical properties (e.g., firmness, hardness, and resilience) is hard to understand. This study developed a time-resolved transmittance spectroscopic method to validate the light scattering changing characteristics in kiwifruit during shelf-life and in cold storage conditions. The experimental results demonstrated that the reduced scattering coefficient (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mu }_{s}^{\prime}$$\end{document}μs′) of 846 nm inside kiwifruit decreased steadily during postharvest storage and is more evident under shelf-life than in cold storage conditions. Moreover, the correlation between the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mu }_{s}^{\prime}$$\end{document}μs′ and the storage time was confirmed to be much higher than that using the external color indexes measured using a conventional colorimeter. Furthermore, employing time-resolved profiles at this single wavelength, an efficacious mathematical model has been successfully formulated to classify the stages of kiwifruit softening, specifically early, mid-, and late stages. Notably, classification accuracies of 84% and 78% were achieved for the shelf-life and cold storage conditions, respectively.


Sample preparation
In this study, 'Hayward' kiwifruit (Actinidia chinensis var.deliciosa) under pre-climacteric conditions were commercially purchased from a local fruit market in the Wakayama Prefecture of Japan.and such kiwifruit are provided around February each year for fruit ripening experience at home.The skin of this species is thin but has hair and is irregular, which makes it challenging to use conventional reflection spectroscopic methods for quality evaluation.

TRTS measurement system
Figure 1 shows the three main components of the TRTS system, which include an 846 nm picosecond pulsed laser at 70 ps pulse width and 151 mW peak power (PLP-10; Hamamatsu Photonics Co., Hamamatsu, Japan), a synchronous delay generator, and a streak camera with a time resolution of 10.3 ps (C5680; Hamamatsu Photonics Co., Hamamatsu, Japan).First, a circular laser beam on the sample of approximately 1.5 mm in diameter was adjusted.Then, the transmitted photons exiting the sample were collected using a step-index fiber with a diameter of 300 μm and a numerical aperture of 0.22 ± 0.02 (A5760-02; Hamamatsu Photonics Co., Hamamatsu, Japan), placed in contact with the sample.Finally, the time variation of the transmitted radiation intensity was recorded using the streak camera.It can register the time signal of the transmitted light through a photo-counting sensor and a time-acquisition board.The instrument response function (IRF) was measured using a couple of neutral density filters with 1% transmission ratio.The samples' TRP was acquired along the equator line through the sample's skin using a time range of 5 ns, and photon counting was performed for 60 s.Measurements on aqueous solutions of intralipid were done to provide evidence of the performance of the developed TRTS system in recent work 31 .

Check the 846-nm light absorption and scattering characteristics of sliced kiwifruit samples
Three samples at almost the same commercial maturity stage were first measured using the designed TRTS system by slicing them.Their cross-sectional microscopic images were then taken using a bright-field microscope (PrimoStar, Carl Zeiss Microscopy Co., Japan).

Evaluate the efficacy of the µ ′ s in monitoring ripening in comparison with conventional color indexes
A total of 110 samples belong to the same batch were prepared.The firmness, SSC, and pH of 10 samples were measured immediately after arriving at the laboratory (as 0 day).To compare the quality changes under different postharvest storage conditions, 50 kiwifruit samples were randomly selected and stored under shelf-life (22 °C ± 2 °C with 65% RH) condition.Their time-resolved pulse (TRP) spectra were non-destructively measured by the developed TRTS system.Then, each ten individual kiwifruits were randomly picked at 2-day intervals from 1 day, the TRP spectra, firmness, SSC, and pH were measured on each occasion until 7 days.To estimate the light scattering changes at 846 nm during postharvest storage, the last ten samples were non-destructively measured by the developed TRTS system on each occasion until 13 days, when one of the samples had visually rotted.Finally, their firmness, SSC, and pH values were measured.
Meanwhile, 50 remained samples were stored under the same cold (4 °C ± 2 °C with 75% RH) condition.As the same data measurement steps as shelf-life condition, their TRP spectra were non-destructively measured by the developed TRTS system.Then, every ten samples were randomly picked at 7-day intervals from 7 days.Their TRP spectra, firmness, SSC, and pH were measured for up to 1 month.In comparison, the last ten samples were non-destructively measured using the same TRTS system until 49 days, when one was severely damaged.Finally, their firmness, SSC, and pH values were measured.Notably, to reduce the temperature-induced spectral variations, the samples were stored at room temperature conditions for approximately 6 h before conducting the measurements.Other pre-treatments (e.g., the use of 1-methylcyclopropene) were not used.

Non-destructive measurement of external color and diameter
A portable colorimeter (SC-10 precision colorimeter, 3nh Technology Co., Ltd., Shenzhen, China) was employed to ascertain the skin color of the kiwifruit's equatorial region after each TRP measurement.The possibilities for expressing the color were selected as L * , a * , b * , a */ b * , Chroma = √ a * 2 + b * 2 , or Hue = tan −1 (b */ a * ).Subsequently, the sample diameter was measured using a digital caliper (0.01 mm accuracy) to correct the optical path effects in the TRP.

Destructive measurement of firmness, SSC, and pH reference values
Following various postharvest storage durations, MT firmness assessments were administered on the identical equatorial region previously employed for the measurement of TRP.The sample's skin was first removed, and the pulp was destructively measured using a 5-mm-diameter plunger (Fudoh Rheo Meter, Rheotech, Inc., Tokyo, Japan) at a speed of 1 mm/s.Then, the fresh juice of the same equatorial area was extracted for traditional SSC measurement using a Brix refractometer (IPR-201, Spittz, Atago Co., Ltd., Tokyo, Japan), pH measurements were recorded using a pH meter (LAQUAtwin-pH-22B, Horiba Advanced Techno Co., Ltd., Kyoto, Japan).

Time-resolved profile analysis
The measured raw TRP spectral data were first smoothed by a Savitzky-Golay finite impulse response smoothing filter (polynomial order, 2; frame length, 5).Principal component analysis (PCA) was then used to extract the characteristic changes in the TRP spectra.The PCA is a non-parametric, unsupervised tool for reducing the dimensionality of spectral data, increasing interpretability while minimizing information loss simultaneously [38][39][40] .The PC loadings were used to generate the PC scores, which were then utilized to validate the correlation between the storage time of kiwifruit (i.e., the softening process) and the estimated µ ′ s using the curve-fitting method.The values of µ a and µ ′ s were computed by fitting the convolution between the IRF and the model solution 41 .Theoretically, the number of photons arriving at the detector per unit area per unit time can be expressed as follows: where ρ is the distance from the light incident center, which was set as 0 for the transmittance measurement in this study; t is time; s is sample thickness; v is the speed of light in a vacuum dividing by the sample's index of refraction (n); A and n were set as 2.58 and 1.34, respectively; seven dipoles (m = 0, ± 1, ± 2, ± 3) were used.In this study, the entire TRP was used for fitting.The fitting range for µ ′ s was set from 0.1 to 10 mm −1 ; and from 0.0001 to 0.1 mm −1 for µ a , because the µ ′ s was significantly larger than µ a at the 846 nm light wavelength.The fitting method was trust-region-reflective.

Develop a mathematical model to classify early, mid-, and late kiwifruit softening using the single 846-nm TRP spectra
To achieve the initial value (i.e., kiwifruit initial quality variation) correction purpose, the TRP difference spectra were obtained by subtracting the TRP spectra collected at 0 day from the spectra that were measured with the firmness measurements.PCA was then used to extract the characteristic changes in the TRP difference spectra.Early, mid-, and late kiwifruit softening stages were classified based on the measured firmness values.The representative PC scores were further used to classify the three softening stages that under shelf-life and cold storage conditions, respectively, via the support vector machine (SVM) method 42 .Five-fold cross-validation was used to against overfitting.This study used Matlab (The MathWorks Inc., Natick, MA, USA) for data analysis and image processing.s difference between the half and quarter sizes was relatively significant, suggesting that light scattering in the core parts was stronger than in the flesh part.This could be because the central core parts are fibrous (Fig. 3), and a high difference in the refractive index between the core and the flesh.

846-nm light absorption and scattering characteristics of sliced kiwifruit samples
The TRP difference spectra of the 80 samples subjected cold storage are depicted in Fig. 4A.It is apparent from the spectra that the transmitted light exhibited an increase in intensity (up-shift) and a decrease in temporal duration (left-shift) as the storage time progressed.Previous studies have indicated some characteristic changes during the fruit-ripening process, including cell wall depolymerization and water-soluble pectin increases [43][44][45][46][47] .Furthermore, Newman and Redgwell 48 used nuclear spin relaxation experiments to show that kiwifruit's firmness was proportional to the content of rigid non-cellulosic matter in the cell wall.These factors could lead to an overall trend decrease in light scattering despite the expected increase in light scattering caused by water loss.
(1) Additionally, PC2 loading exhibited relatively high absolute values at approximately 1 and 1.5 ns, suggesting a potential influence from the speed of transmitted light.In contrast, the cumulative contribution rate of PC3 loading was extremely low, underscoring its utility in capturing variance from the original input spectra.Figure 4C shows the scatter plot of the first three PC scores, effectively demonstrating the significant changes observed in the TRP difference spectra after a 3-week period of cold storage.

Statistics of external color, weight loss, and diameter during postharvest storage
Figure 5 shows the boxplot of the outside color change of the ten kiwifruits during (A) shelf-life and in (B) cold storage conditions, respectively.The whiskers indicate the most extreme data points not considered outliers that are plotted individually using the '×' marker symbol.Tables 3 and 4 summarize the correlation coefficients (r) between the color parameters and the storage time.Previous studies 7,8 indicated that under storage time, the L * value of fresh-cut kiwifruit slices tended to decrease with an increase in a * values.This study demonstrates that the color changes on the kiwifruit skin were not noticeable enough, showing the correlation to storage time with large fluctuations.It suggests the limitations of using conventional color-based methods to monitor the postharvest decay of kiwifruit non-destructively.

Statistics of firmness, SSC, and pH reference values
Overall, the time and temperature of storage fundamentally influence the diminution of firmness and pH of the kiwifruit samples.Figure 6 shows the changing tendency of firmness, SSC, and pH of the kiwifruit samples under (A) shelf-life and (B) cold storage conditions, respectively.The rates of the decrease in firmness and the increase in pH were different at low temperatures.In contrast, the SSC was less changing, which is not in agreement with most previous studies, in which SSC increased with storage time 49,50 .This could be because the purchased kiwifruit samples initially had variability in the degree of SSC values, and their values did not change as significantly as the other two properties during the storage.

Investigation of 846-nm light absorption and scattering characteristics of intact kiwifruit samples during postharvest storage
Figure 7A shows the TRP at the 846 nm wavelength of a representative sample under the shelf-life condition.It illustrates that longer storage time contributed to increased transmitted light intensity and decreased transmission time.The full width at half maximum of the IRF was 159.83 ps.It should be noted that light illumination was adjusted to be stronger than that used for the measurements of sliced samples.Figure 7B shows the TRP collected from another kiwifruit that under the cold storage condition, with the same tendency seen under the shelf-life condition, longer storage time contributed to an increase in transmitted light intensity and a decrease in transmission time.However, the speed of such changes was slower than that under the shelf-life condition.This agrees with the tendency of measured firmness values (Fig. 6), suggesting the importance of accessing the ripening process under different storage conditions.Figure 8 shows one example of separating the light scattering effects from that of light absorption by fitting the TRP (black empty circles), which is convolved with the IRF, and the analytical solution of the diffusion approximation of the transport equation (red solid line).Tables 1 and 2 summarize the estimated μ a and µ ′ s values of every ten kiwifruit during shelf-life and cold storage, respectively.The fluctuations in the range of µ ′ s among the ten samples could be mainly caused by the initial quality variation.Overall, the experimental results  www.nature.com/scientificreports/demonstrate that the µ ′ s inside kiwifruit tends to decrease during postharvest storage and is more evident under shelf-life than in cold storage conditions.
Tables 3 and 4 also summarize the r values between the estimated values of μ a and µ ′ s and the storage time of every ten kiwifruit under shelf-life and cold storage conditions, respectively.Overall, there was an extremely unstable correlation between the μ a and the storage time, suggesting that the μ a at 846 nm has little relation to the   www.nature.com/scientificreports/fruit softening process.In contrast, fruit storage time and the µ ′ s (846 nm) of the same fruit that was negatively correlated (the longer the time, the lower the µ ′ s ) for both storage conditions.Numerous research investigations have demonstrated a correlation between the µ ′ s properties of fruit and textural attributes, including factors like firmness, cell size, and so on 31,51 .In the present study, the µ ′ s values and firmness of kiwifruit flesh exhibited a simultaneous reduction over the storage period.Comparable positive associations between µ ′ s and firmness were observed in 'Golden Delicious' and 'Granny Smith' apples, as indicated by Cen et al. 22 .However, Rowe et al. reported contrasting correlations for 'Royal Gala' apples with the µ ′ s range of 550-900 nm 52 , while Qin and Lu identified contrasting associations for 'Golden Delicious' apples at µ ′ s of 780 nm 53 , and Vanoli et al. found similar trends for 'Braeburn' apples at µ ′ s of 800 nm 15 .This could be due to the change of flesh texture from firm, crispy and juicy at harvest to mealy at the end of shelf life 54 .Since the µ ′ s was found to be predominantly influenced by both porosity and pore surface density, highlighting their significance as primary physical parameters for predicting scattering behavior 55,56 .In addition to the expected increase in light scattering caused by water loss, www.nature.com/scientificreports/ it is crucial to acknowledge that other factors can counterintuitively lead to a decrease in light scattering.These factors include the depolymerization of cell walls 48 , the loss of rigidity within pectic domains, and the presence of weakly bound water, which exhibits thermodynamically behavior similar to pure water 49 , could reduce the light scattering inside kiwifruit during post-harvest storage.Such phenomena agree with Burdon and Clark's study 57 , which indicates that fruit water content decreases with decreased water potential.Slight moisture loss can cause subtle quality changes in color and texture, and critical moisture loss can cause severe damage to turgidity, firmness, flavor, and nutrition 58 .Furthermore, Taglienti et al. showed that free water becomes the prevailing fraction of the total aqueous protons in kiwifruit samples stored at 20 °C but does not significantly alter during 0 °C storage, even after several weeks.Since the r values of the µ ′ s were lower than those of the PC 1 scores, suggesting the TRTS measurement of the intact kiwifruit, the curve-fitting method should be further improved.Nevertheless, this experiment indicated that the use of µ ′ s at 846 nm was more robust than the use of color parameters or μ a at the same wavelength to monitor the kiwifruit softening process during postharvest storage.To further develop a nondestructive fruit softening progress prediction method based on the light scattering phenomena, the TRP difference spectra were first classified into three classes based on their measured firmness values (Fig. 9A 1 : shelf-life and B 1 : cold storage).It is evident that the softer samples tend to have higher light intensity with faster transmitted time, i.e., lower µ ′ s .Figure 9A 2 ,B 2 show their PC1, PC2, and PC7 loadings.The accumulated contribution rate of the three PC scores was approximately 66.05% and 75.29%, respectively.The scatter plots of the three PC scores are shown in Fig. 9A 3 ,B 3 .The three PC scores exhibited significant variations that played a crucial role in effectively classifying the stages of kiwifruit softening, namely early, mid-, and late stages.
The classification results of the three softening classes of kiwifruit using the SVM and employing a five-fold cross-validation approach are presented in Fig. 10.The PC scores were utilized as representative features for training the SVM classification model based on the TRP spectra.Notably, classification accuracies of 84% and 78% were achieved for the shelf-life and cold storage conditions, respectively.It demonstrated the usefulness of utilizing the light scattering information for monitoring the kiwifruit softening stages.

Conclusion
This study developed a TRTS measurement system that mainly includes a picosecond pulsed laser with a wavelength of 846 nm and a streak camera.The light scattering characteristics of the kiwifruit samples during the softening (i.e., their mechanical properties) were compared.Overall, the µ ′ s exhibited a decreasing trend over time for both the shelf-life and cold postharvest storage conditions, with a more pronounced effect observed under the former condition.Furthermore, a successful mathematical model was developed to effectively classify the softening stages of early, mid-, and late using a single wavelength of 846 nm.
The experimental findings are anticipated to have significant implications in two key aspects.Firstly, they contribute to enhancing our comprehension of the intricate relationship between the spectral data obtained from the collected samples and their corresponding mechanical properties.Secondly, these results hold potential for driving advancements in the field of portable measurement systems, paving the way for the development of robust firmness prediction models that can operate independently of conventional multivariate statistical analysis methods.This progress is crucial for promoting efficient and reliable assessments of fruit firmness in various practical applications.Further experimental investigations are necessary to explore the variations in light scattering through the utilization of diverse data collection methods, encompassing a wide range of fruit types and postharvest storage conditions.

Figure 1 .
Figure 1.Main components of the developed TRTS measurement system.

Figure
Figure2Ashows the TRP collected by cutting the same kiwifruit sample.The intact fruit has the lowest light intensity, and the quarter size without peel (i.e., the pulp) has the highest value.Figure2B,C show the estimated µ ′ s and μ a of the total three samples by fitting their TRP.Their firmness values were 23.52, 22.05, and 21.08 N/ cm 2 , and their diameters were 51.7, 51.8, and 50.2 mm.Overall, the estimated μ a was much lower than the µ ′ s , indicating that the light at 846 nm could penetrate and considerably scatter in the whole kiwifruit.The µ ′ s difference between the half and quarter sizes was relatively significant, suggesting that light scattering in the core parts was stronger than in the flesh part.This could be because the central core parts are fibrous (Fig.3), and a high difference in the refractive index between the core and the flesh.The TRP difference spectra of the 80 samples subjected cold storage are depicted in Fig.4A.It is apparent from the spectra that the transmitted light exhibited an increase in intensity (up-shift) and a decrease in temporal duration (left-shift) as the storage time progressed.Previous studies have indicated some characteristic changes during the fruit-ripening process, including cell wall depolymerization and water-soluble pectin increases[43][44][45][46][47] .Furthermore, Newman and Redgwell 48 used nuclear spin relaxation experiments to show that kiwifruit's firmness was proportional to the content of rigid non-cellulosic matter in the cell wall.These factors could lead to an overall trend decrease in light scattering despite the expected increase in light scattering caused by water loss.

Figure 3 .
Figure 3. Microscopic images of different parts of the same kiwifruit.

Figure 4 .
Figure 4. (A) TRP difference spectra of the 80 samples under cold storage, (B) the first three PC loadings, (C) scatter plot of the first three PC scores.

Figure 5 .
Figure 5. Outside color change of the ten kiwifruits during (A) shelf-life and in (B) cold storage conditions, respectively.

Figure 6 .
Figure 6.Changing tendency of firmness, SSC, and pH under (A) shelf-life and (B) cold storage conditions, respectively.

Figure 7 .
Figure 7. TRF at wavelength 846 nm collected from the kiwifruit during (A) shelf-life and in (B) cold storage conditions, respectively.

Figure 8 .
Figure 8.Estimated μ a and µ ′ s values by fitting the TRP of the same kiwifruit during cold storage for (A) 7 and (B) 49 days, respectively.

Figure 9 .
Figure 9. TRP difference spectral data (A 1 shelf-life and B 1 cold storage), with their PC1, PC2, and PC7 loadings (A 2 ,B 2 ), and the three PC scores of the TRP difference spectra (A 3 ,B 3 ).The number in each scatter plot show the firmness reference values.

Figure 10 .
Figure 10.Prediction results for the softening stage of kiwifruit under (A) shelf-life and (B) cold storage conditions, respectively.In each matrix, each row corresponds to the predicted class, while each column represents the actual class (class 1 indicates the samples with the highest firmness values, followed by class 2 and class 3).

Table 1 .
Estimated μ a and µ ′ s values of the ten kiwifruit during shelf-life storage.

Table 2 .
Estimated μ a and µ ′ s values of the ten kiwifruit during cold storage.

Table 3 .
Correlation between the measured parameters and storage time of every ten kiwifruit during shelflife storage.

Table 4 .
Correlation between the measured parameters and storage time of every ten kiwifruit during cold storage.