Spherical particles are routinely monitored and described by hydrodynamic diameters determined, e.g., by light scattering techniques. Non-spherical particles such as prolate ellipsoids require alternative techniques to characterize particle size as well as particle shape. In this study, oligo(ε-caprolactone) (oCL) based micronetwork (MN) particles with a shape-shifting function based on their shape-memory capability were programmed from spherical to prolate ellipsoidal shape aided by incorporation and stretching in a water-soluble phantom matrix. By applying light microscopy with automated contour detection and aspect ratio analysis, differences in characteristic aspect ratio distributions of non-crosslinked microparticles (MPs) and crosslinked MNs were detected when the degrees of phantom elongation (30-290%) are increased. The thermally induced shape recovery of programmed MNs starts in the body rather than from the tips of ellipsoids, which may be explained based on local differences in micronetwork deformation. By this approach, fascinating intermediate particle shapes with round bodies and two opposite sharp tips can be obtained, which could be of interest, e.g., in valves or other technical devices, in which the tips allow to temporarily encage the switchable particle in the desired position.