Optical coherence tomography angiography (OCTA) enables three-dimensional, noninvasive, contrast-free visualization of vascular networks at different tissue depths and allows in vivo assessment of skin microvasculature. However, imaging artifacts degrade volumetric quality and compromise vascular segmentation and quantitative analysis. This project aims to analyze and optimize OCTA reconstruction algorithms for skin imaging and implement strategies to remove the main artifacts. The aim is to develop a processing pipeline that improves automatic vascular segmentation. The study was conducted on a dataset including controlled skin-mimicking phantoms and in vivo human skin volumes. Two types of phantoms were used: a five-parallel-vessel phantom and a two-layer complex vascular phantom designed to mimic depth-dependent vascular distributions.