Project description and goals

During the last decades, statistical shape models (SSMs) became an emerging tool to capture natural shape variability within a given class of objects. The majority of the state-of-the-art and publicly available models are built for the human face and are typically used for face recognition, animation, and single-shot 3D face reconstruction (known as inverse rendering). However, apart from faces, SSMs were also successfully learned from anatomical objects such as bones and organs.

Despite the popularity of SSMs in the aforementioned areas, only very few works can be found attempting to construct an SSM from 3D breast scans. As such, to date, no publicly available 3D SSM of the female breast exists.

The primary goal of this project is to build a well-performing and expressive 3D SSM of the female breast. This also includes the development of robust methods for rigid and non-rigid registration of 3D breast scans. As a secondary goal, this project aims to evaluate to which extent some of the well-studied applications of SSMs can be transferred into the breast shape domain and how the constructed model could be used for breast surgery simulation.

As part of this project, we developed the Regensburg Breast Shape Model (RBSM) – the first publicly available 3D SSM of the female breast built from 110 breast scans, acquired in a standing position. It can be freely downloaded from our website.

Publications

Maximilian Weiherer, Andreas Eigenberger, Bernhard Egger, Vanessa Brébant, Lukas Prantl, Christoph Palm.
Learning the shape of female breasts: an open-access 3D statistical shape model of the female breast built from 110 breast scans.
Vis Comput (2022). https://doi.org/10.1007/s00371-022-02431-3