Relative Humidity Shifts During Evaporation Enhance the Donor Differentiation Potential of Serum and Plasma Dried...

The droplet evaporation method (DEM) is an analytical approach in which substances dissolved or suspended in a liquid self-assemble into characteristic patterns upon evaporation. Applied to blood and blood-derived fluids, DEM has shown potential for diagnostics. In a previous study, we identified optimal temperature and relative humidity (rH) conditions for generating two main pattern features in dried serum and plasma droplets: crystalline structures (favored at high rH) and peripheral cracks (favored at low rH). The aim of the present study was to investigate whether sequential exposure to two humidity conditions, first 45% rH and then 15% rH, during the evaporation phase of a single experiment could combine both structural features and thereby enhance donor-specific pattern characteristics. Droplets of serum and plasma from eight healthy donors were evaporated under three conditions: constant 45% rH, constant 15% rH, and a changing condition (45% rH for 2 h, followed by 15% rH until completely dry). Evaporation temperatures were 24.5 °C for serum and 30.5 °C for plasma. Computerized texture (GLCM) and fractal (LCFD, MFD, BCFD, lacunarity) image analysis was used to correlate image features with serum and plasma parameters and contents. The donor differentiation was performed by a convolutional neural network (CNN) trained on the droplet pattern images. The changing-humidity condition produced patterns combining the central crystalline structures characteristic of the high-rH condition with the peripheral cracks characteristic of the low-rH condition and yielded the highest CNN-based donor classification accuracy for both fluids, 84.4% for serum, 83.3% for plasma, compared with 66.2-80.8% under the stable conditions. Selected image-evaluation parameters further correlated with the blood derivate contents and parameters including glucose, total protein, triglycerides, homocysteine, urea and the age of the donor. These results suggest that sequential humidity exposure substantially enriches the structural information content of DEM patterns and the amount of donor-specific information that can be extracted from them, with potential implications for diagnostic applications.