Characterization and classification of selective Indian chickpea cultivars based on physical and hydration properties, using image and multivariate analysis

Chickpea (Cicer arietinum) is an important leguminous crop, widely consumed for its high protein content, mainly in Asian countries. India is the largest chickpea‐producing country in the world. Chickpea seeds vary in shape and size depending on the cultivars. Physical properties play an important role in designing the equipment for storage and processing. Thus, the present study was carried out with the objective to characterize selected Indian chickpea cultivars based on their physical properties and hydration characteristics. Hydration is an important preprocessing treatment in most processes like germination, cooking, and fermentation, so the water uptake characteristics (water absorption capacity, swelling index, equilibrium moisture content, and weight gain) vary depending on the size and shape of the seeds. The increase in the length of chickpea seeds after hydration was found to be in the range of 18.18% to 32.08% for different cultivars, whereas the percentage increase in width ranged from 25.62% to 45.45%. The maximum increase in sphericity was 7.39%. Weight gain (%) after hydration was observed to be ranging from 98.87 ± 0.64% to 113.53 ± 0.60%. Comparative characteristics of selective chickpea cultivars based on their physical and hydration properties were studied and classify them using multivariate analysis. Principal component analysis and hierarchical cluster analysis (HCA) for multivariate data may be used to classify the chickpea cultivars. First three principal components explained the 90% variation, with PC1 and PC2 explaining the 73.03% variation. Using HCA, all the cultivars were classified into three clusters based on the similarities in their properties. The three‐dimensional (3‐D) image analysis approach was also employed to study the physical characteristics before and after soaking. It can be a helpful technique for process automation as it is less time consuming.

imaging technique can be employed to automate and validate data obtained from the manual method.  image analysis can be used as an alternate method to save time, and the data obtained can be further employed in simulation processes (Kaur et al., 2021).
Multivariate analysis is used to analyze the variance in physical characteristics among various Chickpea cultivars and classify them accordingly. Cluster analysis is widely used to divide cultivars into groups depending on similarity or variance. The principal component (PC) analysis (PCA) is the most commonly used multivariate analysis method.
The present study was carried out to determine the dimensional, gravimetric, and hydration characteristics of selected Indian chickpea cultivars. Cultivars were classified based on physical and hydration characteristics using PCA and hierarchical cluster analysis (HCA). Image analysis was used to validate the data of dimensional analysis.

| MATERIAL AND METHODS
Different Indian chickpea cultivars of desi type were procured from Punjab Agriculture University (PAU), Ludhiana. Seeds were cleaned to remove extraneous matter and stored in airtight containers under refrigerated conditions. The eleven cultivars used in the present study were Pant Gram 5, GNG-2144, GNG-1581, PBG-8, PBG-5, RSG-963, GPF-2, PDG-4, PDG-3, GNG-2171 2.1 | Physical properties Dimensional characteristics of 11 Indian chickpea cultivars before and after soaking were estimated in terms of three principal dimensions, namely, length (L), breadth (B), and thickness (T). Dial-type vernier calipers with the least count of 0.02 mm were used to measure the dimensional parameters. These principal dimensions were further used for calculating the deriver parameters such as geometric mean diameter (GMD), arithmetic mean diameter (AMD), surface area (SA), aspect ratio (AR), and sphericity (SPH) (Milani et al., 2007). GMD, AR, and SPH were calculated using the relationship given by (Mohsenin, 2020): Sphericity and aspect ratio are mainly used to describe the shape of the seeds.
The surface area (SA) of chickpea seeds was calculated using the following equation, given by (McCabe et al., 1986) SA ¼ πD 2 g Gravimetric properties such as bulk density, true density, and porosity were also evaluated. The weight of chickpea seeds was measured using an electronic weighing balance having milligram accuracy. The weighed sample was then transferred into a measuring cylinder, and bulk volume was noted. Bulk density was calculated as True density was measured using the liquid displacement method, where the known weight of the sample is transferred into a measuring cylinder containing a known volume of a non-absorbing solvent such as toluene. A change in the solvent volume is called the true volume of seeds.
Porosity is an important parameter and is calculated using the bulk density and true density values, using the following equation:

| Hydration characteristics
Seed hydration characteristics were measured in terms of hydration capacity (HC), swelling index (SI), and equilibrium moisture content (EMC) of seeds. The initial moisture content of chickpea seeds was measured using the hot air oven method. Soaking of seeds was carried out at 30 C in proportional-integral-derivative (PID) controlled water bath (MTPH-049, Micro-technologies) with a precision of ±0.1 C until equilibrium was attained (based on previous studies). The time period from the starting of soaking to the time after which the weight of seeds become constant was measured as time to reach equilibrium (Kaur & Prasad, 2022). The moisture content at the point after which there is no significant change in the weight of seeds is termed as saturation point or equilibrium, and EMC can be calculated as where M e % ð Þ= equilibrium moisture content.
W e ð Þ = Weight of grains at equilibrium (g) Hydration capacity was measured as the gain in weight of seeds at the equilibrium stage: The swelling index of chickpea seeds is calculated as: The effect of soaking on the physical characteristics of chickpea cultivars was also evaluated. Percentage change in various parameters was calculated.

| 3-D scanning
Nondestructive 3-D imaging technique was also used to measure the dimensional parameters of chickpea seeds before and after soaking.
The data obtained from the manual method was validated using 3-D imaging. A turntable and 3-D scanning system (Range Vision Spectrum, Italy) was used to generate the polygonal mesh models by crosslinking the group of points, which is further used to construct the 3-D models using SolidWorks software.

| Statistical analysis
A multivariate statistical analysis approach was used to classify the chickpea cultivars. The PCA was used to visualize the relationship between data and the samples. PCA transforms the independent variable into PCs and score plots of PCs are used to overview the similarities or differences among the samples. Similar samples lie closer to each other in the score plot (Ansari et al., 2021). HCA was done for clustering the eleven chickpea cultivars based on their similarities. Multivariate analysis was performed using Origin pro 8.5 software.

| Physical characteristics of chickpea cultivars and effect of soaking
The dimensional and gravimetric properties of different chickpea cultivars are summarized in were found to be similar and was higher among other cultivars. The porosity of PBG-8 was found to be maximum.
During hydration, there is an increase in dimensional parameters of chickpea cultivars (Table S1). The average increase in length of chickpea seeds was around 24.11 ± 4.00%, and an average increase in breadth and thickness was found to be 36.04 ± 6.82% and 31.04 ± 5.00%, respectively. The highest increase in length and thickness was found to be in the case of PBG-7, whereas GNG-2144 showed a maximum increase in breadth. The increase in the size of chickpea seeds during soaking is due to the expansion of the seed structure that was shrunk during drying. The percentage increase in breadth and thickness was more as compared with the increase in length of all the chickpea cultivars. This might be due to the different cell arrangements in different directions (Baryeh, 2002). Similar results were observed during the soaking of chickpea seeds at different temperatures (Kaur & Prasad, 2021a). Derived parameters, namely AMD and GMD, also followed a similar trend. PDG-3 showed maximum sphericity value after soaking. The sphericity of PDG-3 was 79.99 ± 2.79% before soaking and increased to 85.59 ± 2.34% after soaking. In terms of percentage increase, GNG-2144 showed a 7.39% increase in sphericity, whereas it was a 6.93% increase in the case of PDG-3. The average increase in surface area was about 70.31%, with a maximum in the case of PBG-7. During soaking, the volume of seeds increases, resulting in a decrease in the true density of chickpea seeds. An increase in volume can be attributed to the swelling of starch and protein molecules. The porosity of seeds also decreases due to hydration.

| Hydration characteristics
Hydration, also known as soaking, is a mass transfer phenomenon that is attributed to the difference in the water activity, which is the main governing force for this process. During chickpea seeds' hydration, water absorption occurs initially through the hilum and then the seed coat. Absorption of water by the seed structure leads to weight gain and expansion of the seed structure. The stage after which there is no increase in the weight of seeds is termed as equilibrium stage and moisture content at this point is measured as equilibrium moisture content. Hydration capacity (%) was observed to be ranging from

| Validation of dimensional parameters before and after soaking, using 3-D scanning
The data obtained for dimensional parameters using verniers calipers was compared with that obtained using a 3-D scanner (Table S2). Figure 1 shows the representative images of unsoaked and soaked chickpea seeds obtained using 3-D scanning. The values obtained using 3-D scanning were slightly higher than that estimated using the manual method, but the variation was less than 5%. Variations in the values of length and thickness as measured by 3-D scanning were greatly reduced after soaking. Thus, it can be employed as a technique for process automation, as it is a more precise method. This technique has been previously used for morphological studies of various biomaterials (Karasik et al., 2018;Uyar & Erdo gdu, 2009). The irregular shape of chickpea seeds makes it difficult to characterize them using manual methods, but it can be simplified using 3-D image processing.
Due to more irregular seed structure during the initial stage (unsoaked), it was observed that variation in surface area was more for both methods whereas variation among methods was reduced three times for surface area after soaking. This is because the smoothness of the seed surface increases due to hydration. 3-D scanning techniques can improve geometrical modeling and process simulation studies (Uyar & Erdo gdu, 2009).
T A B L E 1 Physical and hydration characteristics of chickpea cultivars in terms of mean ± SD, minimum value, and maximum value Abbreviations: AMD, arithmetic mean diameter; AR, aspect ratio; B, breadth; BD, bulk density; EMC, equilibrium moisture content; GMD, geometric mean diameter; HC, hydration capacity; L, length; POR, porosity; SA, surface area; SI, swelling index; SPH, sphericity; T, thickness; TD, true density.

| Multivariate analysis
The correlation matrix of physical and hydration characteristics is shown in Figure 2. All the dimensional parameters (length, breadth, and thickness) are positively correlated with GMD, AMD, and SA. The surface area was found to have a highly positive correlation with AMD. The aspect ratio was negatively correlated with the length of seeds. As indicated in the figure, larger circles with red color indicate a highly positive correlation, whereas blue-colored circles indicate a strong negative correlation. The swelling index is positively correlated with hydration capacity and equilibrium moisture content. In another study, a significant correlation was found between hydration coefficient and water absorption index (ÖZaktan, 2021). Seed dimensional parameters are negatively correlated with bulk and true density.
F I G U R E 1 Representative figure of threedimensional (3-D) scanning of chickpea seeds (PBG-7) before and after soaking.
F I G U R E 2 Correlation matrix for physical and hydration characteristics of chickpea cultivars.
F I G U R E 3 Eigen values and cumulative variance for all principal components.
F I G U R E 4 Biplots of chickpea cultivars using principal component analysis. Figure 3, the first three PCs explained more than 90% variation, and the components having eigenvalues of more than one are taken into consideration because the values less than 1, account for lesser variability (Girden & Kabacoff, 2010). The first PC (PC1) explained 42.95% variation, followed by the second PC (PC2), which explained 30.08% variation, followed by PC3, explaining 19.47% of the variation, and altogether the three PCs explained 92.50% variation. In PCA, the loading plot describes the relationship between variables, and the score plot depicts the distribution of samples. Figure

| CONCLUSION
The variability and relationship between selected Indian chickpea cultivars were evaluated in the present study. Classification of chickpea cultivars was done based on their physical and hydration characteristics. The hydration behavior of the seeds was cultivar dependent. The three major dimensions (L, B, and T) increased significantly after soaking, but the increase in breadth and thickness was more as compared with the increase in length for all the chickpea cultivars. 3-D image analysis technique can be effectively used as an alternative method to conventional techniques to estimate the physical parameters of chickpea seeds before and after soaking, as they are irregular in shape, and it becomes difficult to characterize them manually. This technique is less time consuming, more precise, and can find a potential role in process automation. A multivariate analysis approach was used to classify the chickpea cultivars, and the grouping of these cultivars was carried out using HCA. Chickpea cultivars were categorized into three clusters: the first cluster comprised of three F I G U R E 5 Dendrogram of cluster analysis for chickpea cultivars.
cultivars; the second cluster consisted of six cultivars; and the remaining two cultivars were placed in third cluster, depending on the similarity in the physical or hydration properties.