Shape and size are factors that influence the way a medical compression stocking exerts pressure on the leg. To create a correct medical device, not only the mechanical properties of the textile structure are relevant but also the way the knitting pattern is being created, the latter being influenced by the morphological shape of the wearer. In this study, we present a comprehensive shape analysis framework for a dataset of human legs, focusing on quantifying the shape variations within our population, leveraging both classical and modern computational techniques. Initially, we process 1D and 2D shape descriptors with spectral clustering methods to identify morphological groups for human legs. This work also presents a methodology for extracting a 3D shape descriptor with promising comparability factors alone. Our future work aims to take the 3D leg as a polygonal mesh, introduce it in known mesh convolutional neural networks, and determine a final set of morphotypes. This could enhance our understanding of leg shape and aid in the design of well-adapted medical compression stockings.