Multimodal transformers predict cancer therapy response from tumor mechanics
Styliana Georgiou, Triantafyllos Stylianopoulos, Chrysovalantis Voutouri · 2026
Read Multimodal transformers predict cancer therapy response from tumor mechanics by Styliana Georgiou, Triantafyllos Stylianopoulos, Chrysovalantis Voutouri (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13614918.txt.
Description
Whether a tumor will respond to treatment may be written in its stiffness as much as in its genes. Styliana Georgiou and two co-authors build a model that reads both: ultrasound shear-wave images showing tissue texture, plus a numeric stiffness measure, fused inside a transformer network. They train it on 1,578 baseline images from five mouse tumor models.
The model sorted tumors into responders, stable, and non-responders with about 92 percent accuracy on held-out data. Careful ablation tests showed the skill came from truly combining the two inputs, not from stiffness alone. It also generalized to a tumor type left out of training, at about 95 percent accuracy. Softer tumors tended to respond better. The authors call for human validation of mechanics as a biomarker. The study appears in 2026.
Catalog record
| CREATOR | Styliana Georgiou, Triantafyllos Stylianopoulos, Chrysovalantis Voutouri |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 88 KB |
| VERSION | 1.0 |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |