Monitoring of banana’s optical properties by laser light backscattering imaging technique during drying

Banana drying is an important process that used to extend the shelf life and increase the marketability of the dried banana. However, this process can lead to changes in weight loss, firmness, and color, which may influence consumer acceptance. As a result, it is crucial to monitor these changes to maintain the desired quality. Therefore, the aim of this study was to assess the quality of sliced bananas during the drying process by simultaneously monitoring their optical and physical properties using laser light backscatter imaging, near-infrared spectroscopy, and electrical impedance spectroscopy techniques. Banana sliced were prepared with 10 mm thickness and immersed into 4% ascorbic acid solution and water as treated and control samples dried at 50 °C for 6 h. The parameters measured were weight loss, color, firmness, NIR absorbance in the range from 740 to 1700 nm and electrical impedance in the frequency range from 30 kHz to 1 MHz. Absorption, reduced scattering and diffusion coefficients, penetration depth and full width at half maximum (FWHM) were computed on the LLBI signal at wavelengths of 532, 635, 780, 808, 850 and 1064 nm. The results showed that both the drying time and the ascorbic acid treatment and their interaction affected the measured values. The strong NIR absorption spectrum changes observed at wavelengths of 1064 and 1416 nm. The least squares partial regression model (PLSR) was performed with high accuracy for weight loss (%) and relaxation time (ms) at a coefficient of determination (R2) of 0.940 and 0.945 with a mean square error (RMSE) of 3.748 and 0.001, respectively. The electrical impedance spectral changes were found in the frequency range from 60 Hz to 1 MHz. The most sensitive laser wave lengths to evaluate optical properties were 532, 635, 780 and 1064 nm. Therefore, laser backscatter imaging together with NIR spectroscopy and impedance spectroscopy is a promising technique to assess the quality of sliced bananas during the drying process.


Introduction
Bananas are one of the most important tropical fruits, accounting for approximately 15% of the world's total fresh fruit production [1][2][3]. They are a rich source of carbohydrates, vitamins (vitamin B6 and C), minerals (potassium and magnesium), antioxidants, and dietary fiber [4,5]. According to Kader's recommendation in 2012, bananas are perishable and cannot stored for extended periods after harvest unless they are stored between 10 and 14 °C with a relative humidity of 85-95% [6]. On the other hand, according to a review by Al-Dairi et al. in 2023, the post-harvest loss of banana is up to 50% of the total volume of production of bananas [2]. This loss has a negative impact on the banana industry and the economy of producing countries.
Drying bananas is a process that involves removing moisture from the product, leading to an extended shelf life and increased marketability of the dried product [7][8][9]. Additionally, it provides long-term storage and facilitates transportation [10,11]. Nowadays, studies have shown the application of different drying technologies such as freeze drying, vacuum drying, microwave drying, convective hot air drying, and heat pump drying [12][13][14][15][16]. The color, weight loss, and firmness of dried fruit are critical properties that affect consumer acceptability [17]. Various pre-treatments such as lemon juice, ascorbic acid, l-cysteine, and phosphoric acid have investigated in different studies to prevent browning and shrinkage during the drying process [18][19][20]. However, the changes in the quality parameters are due to chemical reactions such as the Maillard reaction and enzymatic activity [21]. Hence, temperature and drying time are key factors affecting the color, texture, and chemical composition of the dried bananas [8,12,15]. Therefore, quality assessment techniques are crucial for monitoring the quality of the final dried product.
The application of non-destructive techniques such as laser backscattering imaging, near-infrared (NIR) spectroscopy, and electrical impedance spectroscopy can provide valuable information about the quality of dried bananas. However, the simultaneous application of these techniques can give a 1 3 more comprehensive understanding. Despite this, studies on monitoring banana quality during the drying process using these techniques have limited in the literature. The aim of this study was to assess the quality of bananas during the drying process by monitoring physical and optical properties using non-destructive techniques such as electrical impedance spectroscopy, near-infrared (NIR) spectroscopy, and laser light backscatter imaging (LLBI). The results obtained from each technique were compared, and the observations were validated with commonly used reference methods.

Sample preparation and pretreatment
Five kilograms of banana (Musa cavendishii L.) were purchased from retail (SPAR Magyarország Kereskedelmi Kft, Budapest, Hungary). They were packed in polyethylene plastic bags and transported to the laboratory within 30 min. The bananas used in the experiment were graded and selected based on uniformity of size, color, and freedom from external defects. The selected bananas were then manually peeled and sliced into 10 mm thick cylindrical discs using a dual-blade parallel sample preparation tool (Type SP/TB, Stable Micro Systems, UK). The sliced samples were randomly divided into two groups: control and ascorbic acid treated groups. Each group contained 70 pieces. The control and treatment groups were soaked in water and in 4% ascorbic acid solution for 15 min, respectively.

Drying process
The prepared samples subjected to convective hot air drying in a Venticell 222 oven (MMM Medcenter Einrichtungen GmbH, Germany) for 6 h at 50 °C with 100% ventilation flow. Ten slices were taken hourly from each group for measurement. The initial mass of each slice was measured using a Mettler AE200 analytical balance (Mettler Toledo, USA) with 0.01 g accuracy before being subjected to the drying process. The reference quality attributes were determined using traditional methods and are presented below.

Weight loss
The mass of each slice during the 6 h drying time was measured using a Mettler AE200 analytical balance (Mettler Toledo, USA) with the accuracy of 0.01 g. Weight loss was calculated as difference of the current and the initial weight relative to the initial value.

Firmness
The maximum cutting force (N) was used to express the firmness of the banana samples. A cutting test was performed on the samples using a Warner-Bratzler Rectangle Slot Blade (HDP/WBR) on a SMS TA-XTplus texture analyzer (Stable Micro Systems, Surrey, UK) at a test speed of 1 mm/s. The force required to penetrate the sliced banana was recorded every 0.01 s.

Color
The color parameters (L*, a*, b*) were measured using a ColorLite sph850 spectrophotometer (ColorLite GmbH, Germany). Before the sample measurement, the spectrophotometer was calibrated using a BAM CERAM-White and a white working standard etalons. The device was then positioned perpendicular to the surface of the slices for scanning. Chroma (C*) and hue (h*) were calculated from the extracted a* and b* values as color property indices, using the following formula: where a* is the red-green axis, b* is yellow-blue axis.

Impedance spectroscopy
Banana slices were placed between the two stainless steel parallel capacitor plates (HP16451, Hewlett-Packard, City, USA). The diameter of electrodes was 35 mm. The magnitude and phase of the electrical impedance were measured in the frequency range from 30 Hz to 1 MHz using an LCR meter (HP4284A, Hewlett-Packard, City, USA). The measured spectra were open-short corrected according to the procedure suggested by Repo et al. [39] to eliminate stray capacitance and inductance. The impedance spectrum of a banana disc was modeled using an equivalent circuit with a single resistance distribution circuit element (DCE) (Fig. 1). The circuit model for corrected spectra was approached using a combination of resistance R 1 and distributed circuit elements, as shown in formula (2). The electrical impedance of the samples is a complex value that includes both resistance and capacitance. Resistive pathways across the tissues are associated with the real part of the impedance, while capacitive pathways are associated with the imaginary part of the impedance [29].
Resistances R and R 1 , relaxation time τ, exponent ψ, i = √ −1 is the imaginary unit and = 2 f is the angular and f is the measuring frequency.

Near infrared (NIR) spectroscopy
Reflectance spectra in the wavelength range 740-1700 nm with the resolution of 2 nm were recorded using the MetriNIR benchtop spectrophotometer (MetriNIR Research, Development and Service Co., Budapest, Hungary). The samples were placed in a ∅40 mm glass cuvette and two consecutive spectra were taken for each sample with 90° rotation between. For the NIR spectra, the pre-processing of the data included the calculation of the standard normal variate (SNV) and the 1st and 2nd derivatives. They used to reduce scattering caused by instrumental noise and variations [6]. The most important wavelengths regarding the prediction of quality parameters have also been determined. The pretreated NIR spectral data were analyzed using principal component analysis (PCA) and partial least squares regression (PLSR). The performance parameters of PLSR, including root mean square error (RMSE), coefficient of determination (R 2 ), and ratio of prediction to deviation (RPD) were determined using R software (4.1.2 R Foundation for Statistical Computing, Vienna, Austria).

Laser beam system
The laser beam imaging system was assembled with a 12 bit/ pixel monochrome CMOS camera (MV1-D1312, Photon Focus AG, Switzerland) and lenses optimized for the camera spectral range of 320-1080 nm. Images were captured with the resolution of 0.1127 mm/pixel and picture size of 512 × 512 pixel. Laser diodes emitting at 532, 635, 780, 808, 850, and 1064 nm. The incident angle of laser beams was adjusted to 15° and beams were focused within a circular area of Ø1 mm. Image acquisition was performed in a dark chamber to avoid  scattering light from other sources and increase the signalto-noise ratio. The optical parameters of absorption, reduced scattering, total and effective attenuation coefficients, were extracted using Light Scatter software [40].
From the extracted total and effective attenuation coefficients, the diffusion coefficient and light penetration depth were calculated as follows:

Line laser system
In this study, a line laser imaging system developed to introduce a new measurement system to monitor the quality parameters of banana slices during the drying process. The system consisted of a dark chamber equipped with portable digital CCD camera (SONY, DSC-W800, Tokyo, Japan) and a zoom lens (F3.2 and a focal length of 26-130 mm). A single laser module of 635 nm (1 mW, 1 mm line thickness) was used to illuminate the sample. The distance between the camera lens and the sample was set at 27 cm, and the angle of incidence of the line laser was set at 20° to minimize distortion and direct reflections from the point of incidence. Digital images with the resolution of 0.03125 mm/pixel and picture size of 5152 × 3864 pixel were taken from each sample. The redgreen-blue (RGB) images were converted to grey scale, and then full-width half maximum-value (FWHM) was extracted from the intensity profile using Scilab (version 6.1.1) software (Fig. 2).

Statistical analysis
A two-way multivariate analysis of variance (MANOVA) was conducted to analyze the quality parameters of banana samples. Significance level was adjusted to p < 0.05 and confidence interval of 95% was calculated. Pearson's correlation coefficient (r) calculated to assess the relationship between measured parameters. The normality of the data was evaluated according to skewness and kurtosis.

Weight loss
During a 6-h drying, both the ascorbic acid treated, and control samples of sliced bananas experienced an increase in weight loss. This weight loss occurred as result of the evaporation of moisture within the bananas due to heat and mass transfer [16]. Additionally, enzymatic reactions may occur during the drying process, causing changes in the composition of the bananas and contributing to further weight loss [8]. Furthermore, although the treated samples showing lower mean weight loss values at each drying hour interval compared to the control group, the weight loss increased over time (Fig. 3). This suggests that while the ascorbic acid  treatment had a positive effect in reducing weight loss, it was not able to completely prevent weight loss during the drying process [41]. The statistical analysis indicates that both the drying time and treatment have a significant influence on the weight loss ( Table 1). The significant difference in weight loss between each drying time interval for both the treated and control sample groups, as evidenced by Tukey's post hoc test results, showed that the dynamics of moisture removal during the drying process. The change in weight loss reflect the progressive removal of moisture from the samples with increasing drying time [42].

Firmness
The maximum force required to cut banana slices with a cutting probe increases with drying time, which is attributed to changes in firmness (Fig. 4). This increase is primarily due to moisture loss. When moisture evaporates the sliced banana shrinks and becomes a more compact and dense structure [7,9,41]. This structural change makes the banana slices tougher and more resistant to cutting. Additionally, starch retrogradation occurs, referring to the rearrangement and recrystallization of starch molecules, leading to the formation of a more rigid and less soluble structure. This starch retrogradation contributes to the increased firmness of the dried banana slices [43]. Furthermore, the force required for cutting also increases due to Maillard reactions. These reactions can lead to the formation of a crust-like layer on the surface of the banana slices, further adding to their resistance to cutting [18,44]. Statistical analysis showed that maximum force of both the control and ascorbic acid treated groups increased with drying time (Table1). According to the Tukey's post hoc test result, there was a significant difference in maximum force required between each drying time interval for both sample groups. However, there was no significant difference between the first and second drying hour in the treated sample groups. This could be due to the role of ascorbic acid pretreatment in maintaining organic compounds and minimizing reactions [18,20]. Figure Table 2). The increase in hue intensity suggests that the color of the banana slices became more pronounced and vibrant as the drying proceeds. This change in hue intensity is due to alterations in the concentration of carotenoid pigments [45]. Moreover, the use of ascorbic acid resulted in a lower change in the color of sliced bananas compared to the control groups. This suggests that ascorbic acid can accelerate the anti-browning effect by inhibiting the activity of enzymes that cause browning and acting as antioxidants to prevent oxidation [46]. According to the Tukey post-hoc test results, there was a significant difference in lightness between each drying time intervals for both the treated and control sample groups. This could be due to the decrease in moisture content and structural changes leading to an increase in the concentration of light absorbing compounds that can contribute to a darker appearance and therefore reduced lightness levels. The color of the dried slices shifts to a yellow tone [16]. Figure 6 illustrates the near infrared (NIR) raw spectra (a), normalized spectra using SNV (b), first derivative(c), and second derivatives (d). The higher absorption peak spectrum observed in the near-infrared (NIR) spectra at wavelengths around 1000, 1200, 1400 nm, and 1600 nm would be due to various molecular vibrations and interactions within the sample (Fig. 6). The variations in absorption at these specific wavelengths are related to specific chemical bonds and functional groups present in the sample. In this case, the observed variations can be explained by the presence of OH (hydroxyl) groups and C-H (carbon-hydrogen) bonds in the sample. The absorption peak at around 1000 nm is associated with the OH stretching of the second overtone. This means that the OH groups in the sample are vibrating in a way that absorbs light at this wavelength. Similarly, the absorption peaks at 1200 nm and 1400 nm would be due to a combination of the first overtone of OH stretching and OH bending bands. These vibrational modes involve the stretching and bending of the OH groups, resulting in absorption of light at these wavelengths. The absorption peak at 1600 nm can be attributed to a combination of OH stretching and C-H stretching. This suggests that both OH groups and C-H bonds engage in molecular vibrations that absorb light at this wavelength [23,24]. Principal component analysis revealed that the variations in the NIR absorption spectra found in the water absorption wavelength band because of the drying process. The NIR data variation is illustrated by score plots (Fig. 7a) and loading charts (Fig. 7b). The first three principal components explain 99% of the variability in the datasets which are the dominant features at the water absorption wavelength (Fig. 7b). The NIR absorption peaks at 1356, 1416, and 1646 nm increased with increasing drying time, whereas the NIR absorption at 1064 nm decreased with time. This suggests that the increase in absorption is due to molecular vibrations, while the decrease is due to a reduction in water content. Water forms strong hydrogen bonds with ions, organic monomers, and polymers that can affect the water absorption bands in the near-infrared spectrum [47]. Table 3 shows the results of statistical analysis (analysis of variance) of NIR absorbance at selected wavelengths to assess the influence of drying time, ascorbic acid treatment and their interaction. Drying time and ascorbic acid had an extremely significant effect on NIR absorbance 1064 (F value = 426.663 and 26.185, respectively at p < 0.001; correlation with drying time, r = − 0.90), 1416 nm (F value = 1040.204 and 16.539, respectively at p < 0.001; correlation with drying time, r = 0.95) and 1646 nm (F value = 746.554 and 12.705 respectively at p < 0.001; correlation with drying time, r = 0.95). The chemical composition of the samples can change during drying. This can lead to shifts or variations in the absorption spectra observed in the NIR analysis. In addition, the structural degradation of the samples during drying can affect their optical properties [24,26]. Changes in the physical structure, such as moisture loss, shrinkage, or formation of new chemical bonds, can influence the interaction of light with the sample [23,25]. These structural changes can contribute to changes in the absorption spectra at the analyzed wavelengths. Furthermore, the water matrix plays a crucial role in the NIR absorption measurements. As the drying proceeds, the water content in the samples decreases, which affects the molecular environment and the interactions within the sample. Since water has strong absorption properties in the NIR region, changes in the water matrix can affect the overall absorption spectra of the samples [26].

NIR spectroscopy
According to the results of Pearson's correlation analysis, strong positive correlations were observed among the NIR absorbance values at different wavelengths. Specifically, there was a strong positive correlation between NIR absorbance at 1356 nm and 1416 nm (r = 0.99), as well as between NIR absorbance at 1356 nm and 1646 nm (r = 1.0). Additionally, a strong positive correlation was found between NIR absorbance at 1416 nm and 1646 nm (r = 0.99). These findings indicate that changes in absorbance at one wavelength are likely to be associated with similar changes in absorbance at the other wavelengths. Furthermore, the results showed that the NIR absorbance values at 1356 nm, 1416 nm, and 1646 nm exhibited strong positive correlations with weight loss (r = 0.89, r = 0.85, r = 0.89, respectively) and maximum force (r = 0.92, r = 0.94, r = 0.93, respectively), indicating significant relationships between these variables. This suggests that as the absorbance values at these wavelengths increase, there is a tendency for weight loss and maximum force to increase as well. On the other hand, a negative correlation was observed between NIR absorbance at 1356 nm, 1416 nm, and 1646 nm, and color parameters such as lightness (r = − 0.96, r = − 0.93, r = − 0.96, respectively) and hue (r = − 0.74, r = − 0.69, r = − 0.72, respectively). These correlation coefficients indicate that as the NIR absorbance values increase, the corresponding lightness and hue values tend to decrease.
Derivatives were the best NIR spectral preprocessing method to predict the quality parameters using parietal least square regression (PLSR) model. This is because they enhance spectral resolution, eliminate unwanted noise, and reduce baseline variations ( Table 4).

Electrical impendence spectroscopy
The changes in the magnitude of impedance (a, b), and phase angle (c, d) for both treated and control sliced bananas illustrated in Fig. 9. The measurements were taken in the frequency ranges of 30 Hz-1 MHz with a 1-h interval for 6 h. The impedance magnitude of both control and treated samples decreases as the measured frequency and drying time increase. This means that as the frequency of measurement and the duration of drying increase, the impedance magnitude decreases for both control and treated samples. Additionally, the maximum phase angle spectrum was observed at low frequencies initially. However, as the drying time increased, the phase angle spectrum shifted towards higher frequencies. In terms of a comparison between the ascorbic acid treated and control samples, the impedance magnitude of the treated samples was lower than that of the control once. This suggests that the ascorbic acid influenced reducing the impedance magnitude compared to the untreated samples (Fig. 9a, d). This is because of the electrical impedance is determined by cellular components and tissue structure in different frequency ranges [27,48]. Moreover, their cellular and structural situations can vary due to the influence of drying temperature, moisture content, and frequency [49,50]. During the drying process, sliced bananas showed a steady decrease in the real-and imaginary impedance components over the 60-1 MHz frequency range. This decrease in real impedance can be attributed to the presence of charges forming different double layers (Stern layer, Helmholtz layer) around the electrode, and resulting in electrode polarization and electrode impedance. Furthermore, structural breakdown during drying due to electrical conduction in biological tissues depends on the ion content and the ion mobility of the tissue causes change in impedance. On the other hand, the low-frequency range had a high imaginary impedance value, resulting in an extremely large reactance. This created a high impedance across the cell that prevented the passage of an electrical signal [28,51,52].
The changes in electrical impedance parameters of sliced bananas at a 95% confidence interval, including the initial (a) and tissue (b) resistances, distribution exponent variable (c), and relaxation time (d) were depicted in Fig. 10. The change in all these parameters indicates water loss, chemical and structural degradation of the samples during drying process. Hence, water plays a crucial role in maintaining the structural and functional integrity of biological membrane [29]. The effect of drying time and ascorbic acid treatment was found to be significant for both the initial (F-value = 1053.74 and 2297.123, respectively, p < 0.001; correlation with drying time, r = − 0.79) and tissue (F-value = 165.34 and 41.297, respectively, p < 0.001; correlation with drying time, r = − 0.76) resistances according to Table 5. The treated samples exhibited lower electrical resistance and shorter relaxation times compared to the control samples. This suggests that ascorbic acid plays a protective role in maintaining the cellular structure and organic compounds. Furthermore, the tissue resistance demonstrated a strong positive correlation with relaxation time (r = 0.88) and the exponent variable (r = 0.79). Additionally, it showed a significant negative correlation with weight loss (r = − 0.78) and maximum force (r = − 0.62). Conversely, it exhibited a positive correlation .585, respectively, p < 0.001; correlation with drying time, r = − 0.89) indicates that as the drying time increases, the relaxation time decreases, implying a reduced ability of the tissue to relax and redistribute electrical charges. Similarly, the decrease in the exponent variable suggests an impaired ability of the tissue to exhibit complex electrical behavior. These findings also suggest a limited potential for the redistribution of dielectric current and the occurrence of tissue disorder [48].

Laser light backscattering imaging (LLBI)
The study found that the optical properties of both control and ascorbic acid samples, such as absorption coefficient (µ a ), reduced scattering coefficient (µ′ s ), diffusion coefficient (D), and light penetration depth (h) significantly influenced by drying time ( Table 6). The most sensitive wavelengths for the optical parameters of sliced bananas were 532 nm, 635 nm, 780 nm, and 1064 nm. The mean absorption and reduced scattering coefficient and for diffusion coefficient and penetration depth at the selected wavelength presented in Figs. 11 and 12. The mean absorption coefficient increased with drying time for both ascorbic acid-treated and control groups at wavelengths of 532 nm and 635 nm (Fig. 11a, b), possibly due to moisture loss, pigment and structural degradation [53][54][55]. However, the treated samples showed a lower absorption coefficient than the control groups, suggesting that the ascorbic acid treatment retained the changes in the structure, water content, and color of the sliced bananas depending on the enzymatic and non-enzymatic chemical reactions [46]. At 532 nm wavelengths, the absorption coefficient exhibited a strong positive correlation with weight loss (r = 0.71) and maximum force (r = 0.94). In contrast, it showed a negative correlation with color [lightness (r = − 0.85)]. Moreover, at 635 nm, the absorption coefficient had a strong positive correlation with weight loss (r = 0.82) and maximum force (r = 0.86), but showed a negative correlation with color [lightness and hue (r = − 0.88 and r = − 0.65, respectively)], and electrical impedance parameters [tissue resistances, and relaxation time (r = − 0.58 and r = − 0.53, respectively)]. These findings indicate that the absorption coefficient is closely related to the physical and chemical changes occurring during the drying process of sliced bananas and can serve as an indicator of the quality attributes of dried banana slices [55,56].
The mean reduced scattering coefficient decreased with increasing drying time for both sample groups at 635 nm and 780 nm, which is supported by the Pearson correlation (r = − 0.95 and r = − 0.73, respectively). This finding indicates that the structural properties of the banana samples, such as density, particle size, and cell structure, played a role in light scattering [32]. In addition, the samples treated with ascorbic acid showed a smaller decrease in the scatter coefficient compared to the control group (Fig. 11c,  d), suggesting that the treatment helped to preserve the internal structure of the banana slices. At 635 nm, the reduced scattering coefficient showed a strong negative correlation with weight loss (r = − 0.96) and maximum force (r = − 0.78), while exhibiting a positive correlation with color [lightness and hue (r = 0.92 and r = 0.67, respectively)] and electrical impedance parameters [tissue resistances and relaxation time (r = 0.78 and r = 0.78,   . These findings suggest that a decrease in the reduced scattering coefficient coincides with an increase in weight loss and texture parameters, limiting photon transport within the tissue [56,57]. Figure 12 illustrates the changes in the diffusion coefficient and light penetration depth at different wavelengths. As the drying time increased, the mean diffusion coefficient showed an increase for both sample groups at 532 nm and 1064 nm. The most significant difference was observed between the control and treated samples at 532 nm (F-value = 6619.153 and 269.092, respectively, p < 0.001; correlation with drying time, r = 0.74) and 1064 nm (F-value = 1133.706 and 6.551, respectively, p < 0.001; correlation with drying time, r = 0.91). In the first three drying hours, the treated samples exhibited a higher diffusion coefficient at 532 nm, but a lower one in the last 3 h (Fig. 12a). Additionally, during the first 5 drying hours, the treated samples demonstrated a higher light diffusion coefficient at 1064 nm. This could be due to decline moisture, surface structure and chemical composition degradation of the slices [8,57]. On the other hand, according to Pearson's correlation analysis, the diffusion coefficient was positively correlated with weight loss (r = 0.64) as well as maximum force (r = 0.58). Additionally, at laser wavelength of 1064 nm, it was positively correlated with weight loss (r = 0.82), and maximum force required (r = 0.90). However, it exhibited negative correlations with impedance parameters: tissue resistance and relaxation time (r = − 0.62 and r = − 0.57, respectively).
The increase in mean light penetration depth at both 532 nm and 1064 nm with drying time during banana drying can be attributed to changes in the optical properties of the banana slices during the drying process (Fig. 12c, d).
In the initial stages of drying, compared to dry tissue, the sliced banana contains more water with a higher refractive index, resulting in greater light scattering and absorption [56]. However, as the drying time proceeds, the moisture content in the banana slices gradually decreases and the refractive index of the tissue approaches that of the dry state. This transition results in a reduction in light scattering and absorption, allowing the light to penetrate deeper into the banana slices [58]. The effect of drying time was significant on the light penetration depth of banana slices and showed a strong correlation at both wavelengths (r = 0.92 at 1064 nm and r = 0.85 at 532 nm) ( Table 6).
The light penetration depth at 532 nm wavelengths exhibited a strong positive correlation with weight loss (r = 0.76) and maximum force (r = 0.64). In contrast, it showed a negative correlation with color [lightness and hue (r = − 0.81, r = − 0.63, respectively)]. Moreover, at 1064 nm, it had a strong positive correlation with weight loss (r = 0.83) and maximum force (r = 0.77), but showed a negative correlation with color [lightness and hue (r = − 0.85 and r = − 0.83 respectively)], and electrical impedance parameters [tissue resistances, and relaxation time (r = − 0.65 and r = − 0.64, respectively)]. Figure 13 indicates the backscattering images of the sliced banana at different drying time intervals. Each image corresponds to the light intensity and scatter during drying of sliced bananas at 635 nm. The bright blue part in the line represented the light intensity, which decreased as drying time continued. Moreover, the scattering around the incident point decreased as the drying time proceeded. This is due the scattering of photons in the visible wavelength range is influenced by cell structure and moisture content of the fruit tissue [36,53,59].
The intensity profile of the samples varies over the 6-h drying period (Fig. 14a). As the distance from the center of the peak increased, the intensity decreased. The full width at half maximum (FWHM) of the intensity profile for sample decreased with drying time. The treated sample groups had higher FWHM than control groups during the whole experiment (Fig. 14b). This could be due to changes in the moisture content and surface structure during the drying process [56]. Both the drying time and ascorbic acid treatment had a significant effect on the half-maximum fullwidth (F = 9.44 × 10 29 and 1.825 × 10 29 , p < 0.001; correlation with drying time, r = − 0.71) as presented in Table 6. The FWHM at 635 nm has a strong negative correlation with weight loss (r = − 0.78), while showing a strong positive correlation with the color parameter; lightness (r = 0.68) and impedance parameters; tissue resistance and relaxation time (r = 0.64 and r = 0.86, respectively).

Correlations between optical and impedance parameters
The laser optical absorption coefficient at 535 nm showed a strong positive correlation with the NIR absorption at 1416 nm (r = 0.89) and an inverse correlation with the impedance parameter; exponent variable (r = − 0.59). In addition, the absorption coefficient at 635 nm showed a strong positive correlation with the NIR absorption at 1416 nm (r = 0.96), but a negative correlation with the impedance parameter; exponent variable (r = − 0.75). This suggests that these wavelengths can be used to monitor changes in the physical and structural properties of banana during drying.
The reduced scattering coefficient at 635 nm showed an inverse correlation with the NIR absorption at wavelengths of 1064 nm (r = − 0.81). However, it showed positive correlations with impedance parameters including tissue resistance (r = 0.78), relaxation time (r = 0.78), and exponent variable (r = 0.87). Furthermore, the reduced scattering coefficient at 780 nm showed a negative correlation with NIR readings at 1064 nm (r = − 0.70) and a positive correlation with the impedance parameter relaxation time (r = 0.74).
The laser diffusion coefficient at 1064 nm showed a positive correlation (r = 0.95) with the NIR absorption at 1416 nm, but an inverse correlation with impedance parameters, especially tissue resistances (r = − 0.62). The penetration depth of the laser light at 532 nm showed an inverse correlation with the NIR absorption at 1064 nm (r = − 0.76) and with the impedance parameter; exponent variable (r = − 0.74). Furthermore, light penetration depth at 1064 nm was positively correlated with NIR absorbance at 1356 nm (r = 0.90) but was inversely correlated with impedance parameters such as tissue resistance (r = − 0.68) and exponent variables (r = − 0.93). This would be due to the interaction of light with the constituents of banana tissue such as carbohydrates and cellular structures as the moisture content changes during drying, resulting in variations in the depth of penetration [56][57][58][59].

Conclusions
This study investigated the effects of drying time and ascorbic acid treatment were significant in weight loss, color, firmness of dried banana, as well as its impedance, and optical parameters. Ascorbic acid-treated samples had better structural integrity compared to control. The changes in NIR absorbance at certain wavelengths, particularly 1356 nm, 1416 nm, and 1646 nm, strongly correlated with weight loss and texture parameters. A partial least square regression model accurately predicted changes in weight loss (R 2 = 0.940 RMSE = 3.748%) and relaxation time (R 2 = 0.945, RMSE = 0.001 ms). The impedance magnitude of both treated and control samples decreased with frequency and drying time and the treated samples showed higher impedance magnitude. The spectral changes were found in the frequency range from 60 Hz to 1 MHz. The most sensitive wavelengths for the optical parameters were 532 nm, 635 nm, 780 nm, and 1064 nm. The absorption coefficient at 532 nm and 635 nm, diffusion coefficient and light penetration depth at 532 and 1064 nm increased with drying time. However, the mean reduced scattering coefficient at 780 nm and 635 nm decreased in both sample groups. The full width at half maximum (FWHM) of laser intensity profile at 635 nm negatively correlated with weight loss and positively correlated with color and impedance parameters. Therefore, laser backscattering imaging along with NIR spectroscopy and impedance spectroscopy are promising techniques for evaluating the quality changes of banana.
Based on the results, ascorbic acid treatment can be used to improve the quality of dried bananas. NIR spectroscopy, particularly at wavelengths of 1064, 1356 nm, 1416 nm and 1646, laser backscatter imaging (LLBI) at sensitive laser wavelengths of 532 nm, 635 nm, 780 nm and 1064 nm and electrical impedance in the frequency range from 60 Hz to 1 MHz can used to monitor changes in weight loss, firmness, and color of dried bananas.