Rethinking deep learning in bioimaging through a data centric lens

Jiajun Cao; Jan Wenzel; Shanghang Zhang; Josephine Lampe; Hongxiao Wang; Jiachen Yao; Zhicheng Zhang; Shuo Zhao; Yu Zhou; Chao Chen; Markus Schwaninger; Jufeng Yang; Danny Z. Chen; Jianxu Chen

doi:10.1038/s44303-025-00092-0

Rethinking deep learning in bioimaging through a data centric lens

Jiajun Cao
, Jan Wenzel
, Shanghang Zhang
, Josephine Lampe
, Hongxiao Wang
, Jiachen Yao
, Zhicheng Zhang
, Shuo Zhao
, Yu Zhou
, Chao Chen
, Markus Schwaninger
, Jufeng Yang
, Danny Z. Chen
, Jianxu Chen

Peking University
University of Lübeck
DZHK (German Research Centre for Cardiovascular Research)
Capital Normal University
Stony Brook University
Nankai University
Nankai International Advanced Research Institute
Leibniz-Institut für Analytische Wissenschaften
Ruhr University Bochum
University of Notre Dame

Research output: Contribution to journal › Comment/debate

6 Scopus citations

Abstract

Deep learning has become essential in bioimaging for tasks. By examining data-centric strategies in general AI and revisiting existing deep learning methods in bioimaging, we describe a prototypical “BioData-Centric AI” framework. For AI users in bioimaging, this framework promotes a more practical approach beyond simply annotating large datasets or relying on a universal model. For method developers, it highlights key research directions to enhance AI toolboxes for the bioimaging community.

Original language	English
Article number	29
Journal	NPJ Imaging
Volume	3
Issue number	1
DOIs	https://doi.org/10.1038/s44303-025-00092-0
State	Published - Dec 2025

Access to Document

10.1038/s44303-025-00092-0

Cite this

@article{26962c7b005c4754a35fde5ebc22b132,

title = "Rethinking deep learning in bioimaging through a data centric lens",

abstract = "Deep learning has become essential in bioimaging for tasks. By examining data-centric strategies in general AI and revisiting existing deep learning methods in bioimaging, we describe a prototypical “BioData-Centric AI” framework. For AI users in bioimaging, this framework promotes a more practical approach beyond simply annotating large datasets or relying on a universal model. For method developers, it highlights key research directions to enhance AI toolboxes for the bioimaging community.",

author = "Jiajun Cao and Jan Wenzel and Shanghang Zhang and Josephine Lampe and Hongxiao Wang and Jiachen Yao and Zhicheng Zhang and Shuo Zhao and Yu Zhou and Chao Chen and Markus Schwaninger and Jufeng Yang and Chen, \{Danny Z.\} and Jianxu Chen",

note = "Publisher Copyright: {\textcopyright} The Author(s) 2025.",

year = "2025",

month = dec,

doi = "10.1038/s44303-025-00092-0",

language = "English",

volume = "3",

journal = "NPJ Imaging",

issn = "2948-197X",

number = "1",

}

TY - JOUR

T1 - Rethinking deep learning in bioimaging through a data centric lens

AU - Cao, Jiajun

AU - Wenzel, Jan

AU - Zhang, Shanghang

AU - Lampe, Josephine

AU - Wang, Hongxiao

AU - Yao, Jiachen

AU - Zhang, Zhicheng

AU - Zhao, Shuo

AU - Zhou, Yu

AU - Chen, Chao

AU - Schwaninger, Markus

AU - Yang, Jufeng

AU - Chen, Danny Z.

AU - Chen, Jianxu

PY - 2025/12

Y1 - 2025/12

N2 - Deep learning has become essential in bioimaging for tasks. By examining data-centric strategies in general AI and revisiting existing deep learning methods in bioimaging, we describe a prototypical “BioData-Centric AI” framework. For AI users in bioimaging, this framework promotes a more practical approach beyond simply annotating large datasets or relying on a universal model. For method developers, it highlights key research directions to enhance AI toolboxes for the bioimaging community.

AB - Deep learning has become essential in bioimaging for tasks. By examining data-centric strategies in general AI and revisiting existing deep learning methods in bioimaging, we describe a prototypical “BioData-Centric AI” framework. For AI users in bioimaging, this framework promotes a more practical approach beyond simply annotating large datasets or relying on a universal model. For method developers, it highlights key research directions to enhance AI toolboxes for the bioimaging community.

UR - https://www.scopus.com/pages/publications/105022305152

U2 - 10.1038/s44303-025-00092-0

DO - 10.1038/s44303-025-00092-0

M3 - Comment/debate

AN - SCOPUS:105022305152

SN - 2948-197X

VL - 3

JO - NPJ Imaging

JF - NPJ Imaging

IS - 1

M1 - 29

ER -

Rethinking deep learning in bioimaging through a data centric lens

Abstract

Access to Document

Other files and links

Fingerprint

Cite this