Newest papers

What Language Adds, and at What Cost: Four Controlled Probes of Pathology Vision-Language Models

Wenhao Zhang, Zhongliang Zhou, Yi Ping, Zirui Wang, Sheng Li · 2026 · arXiv

DOI: 10.48550/arXiv.2610.26805 · open-access PDF

Spanish

Los modelos de lenguaje de visión (VLM) prometen hacer que la patología computacional sea cuestionable y capaz de redactar informes, pero cuánto añade el lenguaje más allá de la columna vertebral de la visión no está claro, ya que casi cada sistema es evaluado por sus desarrolladores en conjuntos de pruebas de su elección. clasificamos la evidencia publicada por cómo se produjo y probamos cuatro afirmaciones con sondas controladas que reutilizan los puntos de control liberados y los datos públicos, variando una entrada a la vez.

Machine translation (Meta M2M100); the English original follows.

Abstract

Vision--language models (VLMs) promise to make computational pathology queryable and able to draft reports, but how much language adds beyond the vision backbone is unclear, because almost every system is evaluated by its developers on test sets of their own choosing. We grade published evidence by how it was produced and test four claims with controlled probes that reuse released checkpoints and public data, varying one input at a time. Report text adds prognostic signal: adding a frozen embedding of the stage-redacted report to a slide embedding raises the pooled C-index by 0.019 for progres

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange