Exosome Structures Supported by Machine Learning Can Be Used as a Promising Diagnostic Tool.

Principal component analysis (PCA) as a machine-learning technique could serve in disease diagnosis and prognosis by evaluating the dynamic morphological features of exosomes via Cryo-TEM-imaging. This hypothesis was investigated after the crude isolation of similarly featured exosomes derived from the extracellular vehicles (EVs) of immature dendritic cells (IDCs) JAWSII. It is possible to identify functional molecular groups by FTIR, but the unique physical and morphological characteristics of exosomes can only be revealed by specialized imaging techniques such as cryo-TEM. On the other hand, PCA has the ability to examine the morphological features of each of these IDC-derived exosomes by considering software parameters such as various membrane projections and differences in Gaussians, Hessian, hue, and class to assess the 3D orientation, shape, size, and brightness of the isolated IDC-derived exosome structures. In addition, Brownian motions from nanoparticle tracking analysis of EV IDC-derived exosomes were also compared with EV IDC-derived exosome images collected by scanning electron microscopy and confocal microscopy. Sodium-Dodecyl-Sulphate-Polyacrylamide-Gel-Electrophoresis (SDS-PAGE) was performed to separate the protein content of the crude isolates showing that no considerable protein contamination occurred during the crude isolation technique of IDC-derived-exosomes. This is an important finding because no additional purification of these exosomes is required, making PCA analysis both valuable and novel.