Following lower limb amputation, amputees are trained to walk with a prosthesis. The loss of a lower limb deprives them of essential somatosensory information, which is one of the causes of the difficulties of walking with a prosthesis. We here explored whether a solution to this lack of somatosensory feedback could come from natural sensations of the phantom limb, present in most amputees, instead of from substitutive technologies. Indeed, it is known that phantom sensations can be modulated by (i) global mechanical characteristics of the prosthesis socket, and (ii) locally applying a stimulus on an area of the residual limb. The purpose of this pilot study was to verify the feasibility of influencing phantom sensations via such socket modifications in a participant with transfemoral amputation. Four prosthetic interface conditions were studied: a rigid and a semi-rigid socket, each one with and without a focal pressure increase on a specific area of the residual limb. The results show that phantom sensations during walking were different according to the 4 interface conditions. The participant had more vivid phantom sensations in his foot and calf of which some varied as a function of the gait phases. Preliminary gait analysis with wearable sensors shows that these modifications were accompanied by changes in some gait spatiotemporal parameters. This preliminary study of single case demonstrates that phantom sensations can be modulated by the prosthetic interface and can provide natural somatosensory information dynamically varying with gait phases. Although this needs to be confirmed for a larger population of lower limb amputees, it already encourages non-painful phantom sensations to be considered early during the rehabilitation of lower limb amputees.