The samples of this study originated from bilateral sections of the PON of 30 humans. The brain samples belonged to histological collection from the department of Anatomy of the University of Novi Sad, and the Ethics committee approved the further investigation (01-3367/1). The samples were cut into 2 cm × 1.25 cm, which corresponded to dimensions of the human PON and treated with a modified Golgi method of silver impregnation by Kopsch-Bubenaite [17]. The previous publication [18] explained this procedure in detail. The tissue samples were cut on a sliding microtome, in a horizontal plane relative to the anatomical position of the PON, with a thickness of 70 µm.

The incisions of the PON were analyzed by optical microscope “Leica DC 100” (Leica Microsystems, Wetzlar, Germany) using the system magnification of 400× [19]. Adequate impregnated glial cells were evident in a series of optical sections (horizontal cross-sectional images), while any problematic cells were excluded from the sample [7]. All identified glial cells were permanently recorded and transformed into digital images using a digital camera “Leica DC 100” with accompanying software program “Digital Camera System” (Leica Microsystems, Heerbrugg, Switzerland). Any glial cell was recovered by analysis of several images ranging from 3 to 12 in each set [19] [20]. Finally, each cross-sectional image was taken as color images (RGB) and saved in TIFF format (Figure 2(A)).

The digitized images, i.e., horizontal sections, were loaded into Image J (http://rsbweb.nih.gov/ij, National Institute of Health, Bethesda, MD). Then, an anatomist (20+ years of experience) made further editing using various subcommands of the package. Firstly, the image of the entire neuron was reconstructed by subcommand “Z Project”, where all optical sections for a given cell merged into a single image. Then, the RGB image was converted into grayscale [19], and each image was converted into a binary format (Figure 2(B)) using the subcommand “Threshold” by adjusting the contrast. Meanwhile, the specialists reconstructed the trajectory of the soma and each extension [20] and removed all remains of artifacts (subcommand: “File: Delete”). Also, from each cell, anatomists digitally separated the body and attachments. Thus, the investigated sample consisted of 294 (precisely 588) binary images of the PON astrocytes.

The morphology was quantified using seven parameters that describe four image features (size, shape, complexity, and homogeneity). These parameters have been

calculated using the box-counting method [21] [22] [23] [24], either by subcommand of Image J (“Fractal Box count”) or by the FracLac plugin [25].

The monofractal parameters estimated the 2D projection of the glial soma and the glia. Concretely, 1) the surface area with ((D_bin)_s) and ((D_bin)_g), 2) the shape with ((D_out)_s) and ((D_out)_g), 3) the complexity with ((D_skel)_g and 4) the homogeneity with (L_s) and (L_g). The (D_bin)_s, L_s, (D_bin)_g and L_g were calculated either from a binary image of glial soma or the glia (Figure 2(B)) [26].

Also, the (D_out)_s and (D_out)_g were calculated from the outline image of glial soma or the glia (Figure 2(C)) with a one-pixel broad outline was generated in Image J: “Process-Binary-Outline” [27]. Finally, the (D_skel)_g) estimated on the glia’s skeleton image (Figure 2(D)) when all extensions were reduced to a single-pixelwide line in Image J: “Process-Binary-Skeletonize” [27].

The study has investigated the difference between cell types with a two-tailed t-test for independent samples. In contrast, statistical significance between groups was examined by one-way ANOVA with a Bonferroni post hoc test [28]. A value of p < 0.05 was considered statistically significant. These statistical procedures were performed by the SPSS software package v23 (IBM SPSS Statistics, Chicago, IL, USA, demo version).

The sample size indicated that the laws of nonparametric statistics should perform a presentation and further analysis. Thus, Table 1 showed the mean values (and standard errors) of seven monofractal parameters for two types of astrocytes and the corresponding p-value, obtained by the Bonferroni post hoc test.

The first investigation of Table 1 showed that protoplasmic type had higher means of shape and size than fibrous. However, only two parameters showed a significant difference: p < 0.001 for the size ((D_bin)_g) and p < 0.05 for the shape

((D_out)_g). The parameters of the complexity and homogeneity showed the opposite conclusion. Furthermore, in this case, two parameters were statistically different (p < 0.001): one for the complexity ((D_skel)_g) and one for the homogeneity (L_g). Finally, the main result of this analysis was that the monofractal parameters performed on the image of the whole-cell more accurately distinguish the morphology of astrocytes than on the body (Table 1). Also, the two astrocyte types were statistically different in all four morphological properties (size, shape, complexity and homogeneity). Figure 3 showed these types, from the sample analyzed, with values of monofractal parameters close to the mean values from Table 1.

The investigated pool of images belonged to samples of the human PON aged from 35 to 90 years. In age distribution, three age groups were recognized: group A (the second group of maturation), group B (the first group of aging) and group C (the second group of aging). The first group consisted of 104 images (35 - 60 years), the second of 84 images (61 - 75 years) and the third of 100 images (76 - 90 years). Again, all groups had more than 50 images, and morphological differences were investigated by one-way ANOVA followed by Bonferroni post hoc test. Therefore, the statistical evaluation results between the age groups in both types of astrocytes were presented in Table 2.

The table clearly showed that the most significant differences existed for the protoplasmic than for the fibrous type (11 vs. 5). In fibrous type, only three characteristics differed: the (D_bin)_s in two pairs A-C and B-C, the (D_out)_g in A-B and the L_g in two pairs A-B and A-C. Conversely, in the protoplasmic type, six parameters showed significant differences: four properties of the glia ((D_bin)_g, (D_out)_g, (D_skel)_g and L_g) and two of the soma ((D_out)_s and L_s). Also, A-B had the most significant differences (5), while pairs A-C and B-C had the same differences (3). Finally, Table 2 clearly showed that images of the glia had more differences between groups (three and eight) than images of the body (two and three).

In recent times, various studies utilized numerous qualitative studies to classify neurons and glial cells in CNS [6]. However, their approaches are mainly based on the inspection of many structural parameters [15] [26]. This study evaluates the morphology of two types of astrocytes, analyzing the binary image’s four characteristics (size, shape, complexity and homogeneity), using only monofractal parameters.

The location in the CNS and the size are two significant characteristics among the fibrous and protoplasmic types [4]. The fibrous type can be found throughout white matter, while protoplasmic in the grey matter and first type had a larger size (in the ocular plane) than the second. Also, the qualitative analysis of protoplasmic type postulated short, thick and branched extensions [4]. They extend symmetrically and give them the appearance of star-shaped flakes [5] [6]. The fibrous type also had thin and long sections (generally unbranched) that extend radially between nerve fibers and asymmetrical shape in the two dimensions. Earlier qualitative studies also investigated the number of primary extensions, their direction and branching but did not establish apparent differences in types [6]. Other morphological differences, however, have not been quantitatively confirmed. On the other hand, the results of this study clearly showed the difference between types in all four features of the image (Table 1). Furthermore, while there was no significant difference between the bodies, this difference was observed between pictures of the whole cells (Table 1).

As for the size, this study had a surprising result: the protoplasmic type had a larger size than the fibrous (Table 1). However, this result is not supported by the example shown in Figure 3, and anyone can explain this outcome by knowing the true nature of D_bin. This parameter is related to the size, but it quantifies the space-filling property of the image. The D_out showed a similar result, and its extreme values are reported previously [24]. Thus, the reader can conclude that protoplasmic astrocytes have a shape closer to an ellipse than fibrous ones. On the other hand, fibrous astrocytes have higher values of complexity and homogeneity (Table 1). However, this is not seen in Figure 3, but, knowing the meaning of D_skel and Λ, it is easy to conclude that fibrous astrocytes have a greater curvature of the extensions and space than the protoplasmic ones.

The starting point of this part of the investigation was the conclusion that the effect of aging was most clearly evident on astrocyte extensions [8] [29]. Also, the early period of aging quantified the cell complexity and branch extension [4]. Thus, we analyzed the mean values in three age groups and the morphological changes of two types with age. Our results clearly showed a distinct difference in protoplasmic cells compared with fibrous (Table 2).

The most pronounced changes were observed between the second group of maturation and the first group of aging (Table 2). All four features of the image of protoplasmic astrocytes differed significantly, while the difference in cell bodies was observed only in size and shape. The body size, shape, and homogeneity differed substantially between the second group of maturation and the second group of aging. At the same time, the complexity of the whole cell was quite different (Table 2). Finally, the size of the body, the size, shape and homogeneity of the glia differed between the first and second groups of aging (Table 2). In the end, these results support previous qualitative and semi-quantitative findings [30] [31] [32] during the aging process.