Christoph J. Bender,1,2 Kris K. Dreher,1,2 Niklas Holzwarth,1,2 Marcel Knopp,1,2 Jan-Hinrich Nölke,1,2 Tom Rixhttps://orcid.org/0000-0002-0322-0370,1,2 Melanie Schellenberg,1,2,3 Julius Kempf,4,5 Werner Lang M.D.,4,5 Alexander Seitel,1,3 Ulrich Rother,4 Lena Maier-Hein1,2,3
1Deutsches Krebsforschungszentrum (Germany) 2Ruprecht-Karls-Univ. Heidelberg (Germany) 3Nationales Centrum für Tumorerkrankungen Dresden (Germany) 4Friedrich-Alexander-Univ. Erlangen-Nürnberg (Germany) 5Universitätsklinikum Erlangen (Germany)
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
This study delves into the largely uncharted domain of biases in photoacoustic imaging, spotlighting potential shortcut learning as a key issue in reliable machine learning. Our focus is on hardware variation biases. We identify device-specific traits that create detectable fingerprints in photoacoustic images, demonstrate machine learning's capability to use these discrepancies to determine the device that acquired the image, and highlight their potential impact on machine learning model predictions in downstream tasks, such as disease classification.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.
The alert did not successfully save. Please try again later.
Christoph J. Bender, Kris K. Dreher, Niklas Holzwarth, Marcel Knopp, Jan-Hinrich Nölke, Tom Rix, Melanie Schellenberg, Julius Kempf, Werner Lang M.D., Alexander Seitel, Ulrich Rother, Lena Maier-Hein, "Hardware-related biases in machine learning algorithms for photoacoustic image analysis," Proc. SPIE PC12842, Photons Plus Ultrasound: Imaging and Sensing 2024, PC128422T (13 March 2024); https://doi.org/10.1117/12.3000348