Table 4 Summary of the image processing procedure, including sequential processing steps, applied operations, and extracted quantitative features.

1

Input data preparation

Sequential image frames acquired at predefined experimental time points were subjected to analysis. Each frame was treated as an independent sample for further processing.

2

Definition of regions of interest (ROIs)

For each frame, regions of interest were defined to include the skin area exposed to niacin and a control skin area not exhibiting a reaction. When necessary, background regions were also identified to allow their exclusion from further analysis.

3

Determination of baseline parameters

Individual baseline skin color parameters were established based on the control skin regions. These baseline values served as a reference for subsequent calculations.

4

Skin area segmentation

The regions designated for analysis were segmented using edge-based and texture-based filtering techniques. To improve the quality of segmentation masks, morphological operations including dilation, erosion, opening, and closing were applied, enabling effective separation of the ROIs from the background.

5

Manual correction of segmentation

Due to natural variability in skin coloration and the possibility of weak or visually subtle erythema, manual correction of the ROI boundaries was performed in ambiguous cases.

6

Preservation of original color information

Image analysis was conducted using the original color data without reducing the images to a single intensity channel, in order to preserve the full spectral information describing the skin response.

7

Extraction of color space components

From the segmented ROIs, color components of the RGB space (R, G, B) were extracted. Subsequently, the images were converted to the HSV color space to obtain the H, S, and V parameters.

8

Calculation of statistical descriptors

For each color channel within the ROI and for each time point, a set of statistical measures describing the distribution of pixel intensities was calculated, including the mean, median, mode, minimum, maximum, and range.

9

Normalization with respect to control skin

To reduce the influence of individual differences in baseline skin tone and subtle variations in lighting conditions, differential features were calculated as the difference between values obtained in the reaction area and the corresponding values in the control skin area within the same frame.

10

Feature aggregation and data export

All calculated parameters were compiled into a structured feature table including RGB and HSV components, statistical descriptors, and time points. The resulting dataset was exported to a spreadsheet file for further statistical analysis and model development.