Fig. 1: The architecture of MUDI.

A preparation stage: surgery feature profiles \({{\boldsymbol{F}}}_{{\mathcal{S}}}\) are estimated; an application stage: patient’s feature vector X and surgeon’s preference vector α are fed.

A preparation stage: surgery feature profiles \({{\boldsymbol{F}}}_{{\mathcal{S}}}\) are estimated; an application stage: patient’s feature vector X and surgeon’s preference vector α are fed.