Peer-Reviewed Publication
J Imaging Inform Med2025December 16, 2025Journal Article
Pixel Tampering: Does Face Redaction Harm Medical AI Performance?
Eduardo M J M Farina1,2, Felipe A Matsuoka3,4, Gustavo Corradi1, Yosuke Yamagishi5, Masatoshi Abe6, Maximilian Pfeiffer7, Andrea S Souza8, Raquel Moreno8, Ivanei Bramati8, Fernanda Moll8, Almir Bitencourt9, Carlos Sacomani9, Soraia Quaranta Damião9, Rubens Chojniak9, Nitamar Abdala2, Rodrigo Ragazzini2, Henrique Carrete2, Paulo E A Kuriki10, Marcelo Straus Takahashi11, Nelson Caserta12, Cesar H Nomura13, Felipe C Kitamura2,14
1Dasa, São Paulo, Brazil.
2Universidade Federal de São Paulo, Unifesp, São Paulo, Brazil.
3Dasa, São Paulo, Brazil. felipe.matsu.2003@gmail.com.
4Universidade Federal de São Paulo, Unifesp, São Paulo, Brazil. felipe.matsu.2003@gmail.com.
5Division of Radiology and Biomedical Engineering, University of Tokyo, Tokyo, Japan.
6Department of Nephrology, Osaka University, Osaka, Japan.
7, Siegen, Germany, North Rhine-Westphalia.
8IDOR, São Paulo, Brazil.
9AC Camargo, São Paulo, Brazil.
10UT Southwestern Medical Center, Dallas, TX, USA.
11Department of Radiology, University of North Carolina at Chapel Hill, Chapel Hill, USA.
12Universidade Estadual de Campinas, São Paulo, Brazil.
13Instituto do Coração - InCor, São Paulo, Brazil.
14Bunkerhill Health, San Francisco, CA, USA.
Abstract
Balancing data sharing and patient privacy is essential in medical imaging. Face redaction tools anonymize head CTs by removing identifiable features, but their impact on deep learning (DL) model performance remains a concern. We present an open-source face redaction tool designed to enhance data-sharing security while preserving DL performance, validated through a Kaggle competition on age predic…
Create a free account to keep reading
Free members get 10 full research views every month across publications, clinical trials, FDA clearances, adverse events, and NIH grants. No credit card required.
Want unlimited research access? See Pro plans