Exploration of vision transformer models in medical images synthesis