*OPTION 1: To use commercial software for image analysis with established protocols which the user can adapt to the marker specific color.*

Example 1: Algorithm for the fundamental substance (GAGs). Each cylinder was analyzed individually with Aperio Positive Pixel Count Algorithm on Aperio ScanScope software by selecting each cylinder outline in the whole section image. This algorithm was used with the default parameters and counted the number of pixels belonging to a given staining intensity, being the intensity of each pixel the average between the values of red, green, and blue (RGB) intensities. It recognizes Alcian blue stained pixels which have average intensities over 221. Nuclear fast red stained pixels were detected with weak, middle, and strong intensities painted in yellow, orange, and brown following default parameters (**Figure 5**). The remaining cus‐ tomizable parameters were set by default.

**Figure 5.** Segmentation obtained with the positive pixel count algorithm on a salivary gland. **A)** Original image**. B)** Mark‐up image after segmentation. Alcian blue staining is marked in blue and nuclear fast red is marked in yellow, orange or maroon, depending on the intensity of staining.

Example 2: Algorithm for the immune system cells. Each cylinder was analyzed individually with the Nuclear Quant algorithm on Panoramic Viewer software (3D Histech) by selecting each cylinder outline in the whole section image. This algorithm was used with the default parameters and counted the number of cells presenting a given staining intensity. Stained cells were detected with weak, middle, and strong intensities painted in yellow, orange, and maroon (**Figure 6**).

**Figure 6.** Segmentation of a neuroblastic tumor sample stained with IHC anti‐CD45. **A)** Original image. **B)** Mark‐up image after segmentation. Immune system cells are marked in yellow, orange or maroon, depending on the intensity of staining blue and hematoxylin is marked in blue.

*OPTION 2: To use commercial software for image analysis which allows the user to configure personal image analysis protocols or macros capable of recognizing and describing the specific color and shape of the element of interest.*

Example: Algorithm for the reticulin fibers. The grading of fibrosis in general and namely of reticulin fibers is of main interest in bone marrow pathologies [86] and fibrous diseases of the liver [87]. For this reason, different methods for the quantification of reticular fibrosis have been developed, some of them consisting in automated morphometry [88–93]. However, these methods quantify the percentage of stained area and we considered that not only the amount of fibers, but also the morphometric features were relevant, given that these are usually subjectively assessed. Image Pro‐Plus software (Media Cybernetics), which enables the design of specific algorithms, was used to analyze the fibrous component. An image of every cylinder was extracted in a separate JPEG‐quality 80 image from the whole section scan. An algorithm capable of specifically detecting the reticulin fibers and measure their amount, size, and shape was customized. Image J is a free software providing similar options. An example of the segmentation process in a control tissue (kidney) is provided by **Figure 7**.

Tumor Microenvironment Heterogeneity: A Review of the Biology Masterpiece, Evaluation Systems, and Therapeutic Implications http://dx.doi.org/10.5772/62479 273

**Figure 7.** Segmentation process on a kidney tissue. **A)** Original image. **B)** Image after segmentation. The reticulin fibers recognized by the algorithm are marked‐up in red.

**Figure 5.** Segmentation obtained with the positive pixel count algorithm on a salivary gland. **A)** Original image**. B)** Mark‐up image after segmentation. Alcian blue staining is marked in blue and nuclear fast red is marked in yellow,

Example 2: Algorithm for the immune system cells. Each cylinder was analyzed individually with the Nuclear Quant algorithm on Panoramic Viewer software (3D Histech) by selecting each cylinder outline in the whole section image. This algorithm was used with the default parameters and counted the number of cells presenting a given staining intensity. Stained cells were detected with weak, middle, and strong intensities painted in yellow, orange, and maroon

**Figure 6.** Segmentation of a neuroblastic tumor sample stained with IHC anti‐CD45. **A)** Original image. **B)** Mark‐up image after segmentation. Immune system cells are marked in yellow, orange or maroon, depending on the intensity

*OPTION 2: To use commercial software for image analysis which allows the user to configure personal image analysis protocols or macros capable of recognizing and*

Example: Algorithm for the reticulin fibers. The grading of fibrosis in general and namely of reticulin fibers is of main interest in bone marrow pathologies [86] and fibrous diseases of the liver [87]. For this reason, different methods for the quantification of reticular fibrosis have been developed, some of them consisting in automated morphometry [88–93]. However, these methods quantify the percentage of stained area and we considered that not only the amount of fibers, but also the morphometric features were relevant, given that these are usually subjectively assessed. Image Pro‐Plus software (Media Cybernetics), which enables the design of specific algorithms, was used to analyze the fibrous component. An image of every cylinder was extracted in a separate JPEG‐quality 80 image from the whole section scan. An algorithm capable of specifically detecting the reticulin fibers and measure their amount, size, and shape was customized. Image J is a free software providing similar options. An example of the

*describing the specific color and shape of the element of interest.*

segmentation process in a control tissue (kidney) is provided by **Figure 7**.

orange or maroon, depending on the intensity of staining.

272 Composition and Function of the Extracellular Matrix in the Human Body

of staining blue and hematoxylin is marked in blue.

(**Figure 6**).

#### *OPTION 3: To design a personal application capable of solving commercial softwares lacks.*

Example: Algorithm for the vascular component. A common feature of tumor vessels studies is that the researchers focus on microvessel density overlooking other parameters that might be significant, such as the size and shape of the microvessels [94]. Studies have revealed the importance of the size and shape of blood vessels in, for instance, laryngeal tumors [95]. The morphometric tool Angiopath closes blood vessels with discontinuous endothelial layer, recognizes all blood vessels and classifies them in six categories corresponding to different types of vessels, differentiated by their largest diameter or length. Density (density and occupied area), size (area, width, length, and perimeter), and shape (perimeter‐ratio, shape index, branching, aspect, roundness, and deformity) parameters are extracted [96, 97]. An example of the segmentation process is provided in **Figure 8**.

**Figure 8.** Segmentation process on a neuroblastic tumor sample immunostained with CD31. **A)** Original image. **B)** Im‐ age after segmentation. Note that the big blood vessel with an interrupted staining of the endothelial cells surrounding the vascular lumen (asterisk) has been closed, thus providing morphometric measures.

#### *3.3.2.3. Spatial distribution of microenvironmental components*

Topological network analysis and the graph theory in combination with Vonoroi tessellations [98] have recently been found to be useful in the diagnosis of muscular dystrophies and neurogenic atrophies, in the classification of neuromuscular disease or to model the progres‐ sion of dementia [99–102]. All the generated information is subject to capture relevant information about the organization of different tissue markers.

#### *3.3.2.4. Texture analysis*

Another novel approach to cancer research is the texture analysis of different tissue images and machine‐learning methods to train automated algorithms which find different patterns in the tissue capable of discriminating between prognostic groups, among other variables, resulting in computer‐aided diagnosis tools [103–106].
