TY - GEN
T1 - Reliable Saliency Maps for Weakly-Supervised Localization of Disease Patterns
AU - Möller, Maximilian
AU - Kohl, Matthias
AU - Braunewell, Stefan
AU - Kofler, Florian
AU - Wiestler, Benedikt
AU - Kirschke, Jan S.
AU - Menze, Björn H.
AU - Piraud, Marie
N1 - Publisher Copyright:
© 2020, Springer Nature Switzerland AG.
PY - 2020
Y1 - 2020
N2 - Training convolutional neural networks with image-based labels leads to black-box image classification results. Saliency maps offer localization cues of class-relevant patterns, without requiring costly pixel-based labels. We show a failure mode for recently proposed weakly supervised localization models, e.g., models highlight the wrong input region, but classify correctly across all samples. Subsequently, we tested multiple architecture modifications, and propose two simple, but effective training approaches based on two-stage-learning and optional bounding box guidance, that avoid such misleading projections. Our saliency maps localize pneumonia patterns reliably and significantly better than gradCAM in terms of localization scores and expert radiologist’s ratings.
AB - Training convolutional neural networks with image-based labels leads to black-box image classification results. Saliency maps offer localization cues of class-relevant patterns, without requiring costly pixel-based labels. We show a failure mode for recently proposed weakly supervised localization models, e.g., models highlight the wrong input region, but classify correctly across all samples. Subsequently, we tested multiple architecture modifications, and propose two simple, but effective training approaches based on two-stage-learning and optional bounding box guidance, that avoid such misleading projections. Our saliency maps localize pneumonia patterns reliably and significantly better than gradCAM in terms of localization scores and expert radiologist’s ratings.
KW - Saliency maps
KW - Visualization
KW - Weakly-supervised
UR - http://www.scopus.com/inward/record.url?scp=85092920257&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-61166-8_7
DO - 10.1007/978-3-030-61166-8_7
M3 - Conference contribution
AN - SCOPUS:85092920257
SN - 9783030611651
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 63
EP - 72
BT - Interpretable and Annotation-Efficient Learning for Medical Image Computing - 3rd International Workshop, iMIMIC 2020, 2nd International Workshop, MIL3iD 2020, and 5th International Workshop, LABELS 2020, Held in Conjunction with MICCAI 2020, Proceedings
A2 - Cardoso, Jaime
A2 - Silva, Wilson
A2 - Cruz, Ricardo
A2 - Van Nguyen, Hien
A2 - Roysam, Badri
A2 - Heller, Nicholas
A2 - Henriques Abreu, Pedro
A2 - Pereira Amorim, Jose
A2 - Isgum, Ivana
A2 - Patel, Vishal
A2 - Zhou, Kevin
A2 - Jiang, Steve
A2 - Le, Ngan
A2 - Luu, Khoa
A2 - Sznitman, Raphael
A2 - Cheplygina, Veronika
A2 - Abbasi, Samaneh
A2 - Mateus, Diana
A2 - Trucco, Emanuele
PB - Springer Science and Business Media Deutschland GmbH
T2 - 3rd International Workshop on Interpretability of Machine Intelligence in Medical Image Computing, iMIMIC 2020, the 2nd International Workshop on Medical Image Learning with Less Labels and Imperfect Data, MIL3ID 2020, and the 5th International Workshop on Large-scale Annotation of Biomedical data and Expert Label Synthesis, LABELS 2020, held in conjunction with the 23rd International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2020
Y2 - 4 October 2020 through 8 October 2020
ER -