Although the popularity of technologies such as 3d printing have exploded in recent years, maxillofacial surgeons and engineering experts have been working together since the mid 1990's to develop methods of creating physical models derived from medical scan data. This picture shows a milled model, which was commissioned by the team at Morriston Hospital, Swansea in 1997. Whilst the fidelity of reproduction has increased with high resolution 3d printing/additive manufacturing methods, the principles of going from CT slices to physical model remain relatively unchanged since then.
Early software technology such as Materialise (Belgium) released software capable of taking CT image slices, segmenting anatomy based on Hounsfield units and re-stacking them to form a 3d model back in the mid 1990's. Early versions didn't include an undo function (much to my personal frustration!), but proved an extremely powerful way for surgeons to understand procedures before getting into the operating theatre. Computer power and software capability is evolving rapidly now. We're seeing increased levels of segmentation automation and even design automation, which dramatically speeds up processes that were traditionally extremely time-consuming. However, my experience tells me that human intervention is still required to satisfy a safe and effective output.