Deep evidential fusion with uncertainty quantification and reliability learning for multimodal medical image segmentation | Litlas