nnInteractive: Redefining 3D Promptable Segmentation

Fabian Isensee, Maximilian Rokuss, Lars Krämer, Stefan Dinkelacker, Ashis Ravindran, Florian Stritzke, Benjamin Hamm, Tassilo Wald, Moritz Langenberg, Constantin Ulrich, Jonathan Deissler, Ralf Floca, Klaus Maier-Hein · 2025

Read nnInteractive: Redefining 3D Promptable Segmentation by Fabian Isensee, Maximilian Rokuss, Lars Krämer, Stefan Dinkelacker, Ashis Ravindran, Florian Stritzke, Benjamin Hamm, Tassilo Wald, Moritz Langenberg, Constantin Ulrich, Jonathan Deissler, Ralf Floca, Klaus Maier-Hein (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-2503-08373.txt.

Description

Fabian Isensee and co-authors present a 2025 paper on interactive three-dimensional image segmentation for medicine and research. Tools inspired by the Segment Anything Model work well on flat pictures but struggle with volumetric scans, support few prompt types, and rarely plug into the viewers clinicians actually use. The team set out to close those gaps.

Their system, nnInteractive, accepts points, scribbles, boxes, and a new lasso prompt drawn in two dimensions, then returns a full 3D outline. It was trained on more than 120 volumetric datasets spanning CT, MRI, PET, and microscopy. Benchmarks put it well ahead of earlier methods, and it ships inside popular viewers such as Napari and MITK as free software.

Catalog record

CREATORFabian Isensee, Maximilian Rokuss, Lars Krämer, Stefan Dinkelacker, Ashis Ravindran, Florian Stritzke, Benjamin Hamm, Tassilo Wald, Moritz Langenberg, Constantin Ulrich, Jonathan Deissler, Ralf Floca, Klaus Maier-Hein
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE67 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-08373

Browse the full library · About the Tower of Babel