MAIS: Memory-Attention for Interactive Segmentation

Mauricio Orbes-Arteaga 1, Oeslle Lucena 1, Sebastien Ourselin 1, M. Jorge Cardoso 1 · 2025

Read MAIS: Memory-Attention for Interactive Segmentation by Mauricio Orbes-Arteaga 1, Oeslle Lucena 1, Sebastien Ourselin 1, M. Jorge Cardoso 1 (2025) 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/arxiv-2505-07511.txt.

Description

Mauricio Orbes-Arteaga and co-authors present MAIS in 2025, a way to make interactive medical image outlining remember past help. In interactive segmentation, a model drafts an outline of an organ or tumor, and a user clicks to fix its errors. Top models treat each round of clicks as if it were the first, which wastes effort.

MAIS adds a memory that stores past clicks and past outlines. With that context, each new round builds on the last instead of starting over. The authors say the change lifts both speed and accuracy of corrections across many scan types. The work aims to cut the time experts spend tracing medical images by hand.

Catalog record

CREATORMauricio Orbes-Arteaga 1, Oeslle Lucena 1, Sebastien Ourselin 1, M. Jorge Cardoso 1
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE29 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-07511

Browse the full library · About the Tower of Babel