The team is the asset. This is how it distils expertise. From research into software, from teams into agents.
01Published science
Everything here rests on work we published, peer-reviewed, in the field's own journals.
Segmentation. A raw CT scan becomes every vertebra, labelled and separated — a deep-learning method that beat every other at the spine segmentation challenge.
Korez, Ibragimov, Likar, Pernuš, Vrtovec. A framework for automated spine and vertebrae interpolation-based detection and model-based segmentation. IEEE TMI, 36(7), 2017.
Yao, Burns, … Vrtovec, … Li. A multi-center milestone study of clinical vertebral CT segmentation. Computerized Medical Imaging and Graphics, 49, 2016.
Korez, Likar, Pernuš, Vrtovec. Segmentation of pathological spines in CT images using a two-way CNN and a collision-based model. MSKI 2017 (LNCS 10734).
Pedicle-screw planning. Every screw's size and trajectory, computed — and proven against real surgical outcomes.
Knez, Likar, Pernuš, Vrtovec. Computer-assisted screw size and insertion trajectory planning for pedicle screw placement surgery. IEEE TMI, 35(6), 2016.
Knez, Mohar, Cirman, Likar, Pernuš, Vrtovec. Variability analysis of manual and computer-assisted preoperative thoracic pedicle screw placement planning. Spine, 43(21), 2018.
Knez, Nahle, Vrtovec, Parent, Kadoury. Computer-assisted pedicle screw trajectory planning using CT-inferred bone density. Medical Physics, 46(8), 2019.
04Agent-native
The app is an interpreter of a language the agent speaks — so the agent builds the tools itself. We hand it a vertebra, its atlas, and one sentence. It fits the atlas, frames every vertebra, measures the curve, corrects it, and packages the whole thing into a reusable tool. What took us years by hand, an agent did in seven minutes.

Where this goes
We distil expertise into software, into the teams we hand it to, and now into agents. Every field runs the same loop.