Reliable Saliency Maps for Weakly-Supervised Localization of Disease Patterns

Maximilian Möller, Matthias Kohl, Stefan Braunewell, Florian Kofler, Benedikt Wiestler, Jan S. Kirschke, Björn H. Menze, Marie Piraud

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Scopus citations

Abstract

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.

Original languageEnglish
Title of host publicationInterpretable 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
EditorsJaime Cardoso, Wilson Silva, Ricardo Cruz, Hien Van Nguyen, Badri Roysam, Nicholas Heller, Pedro Henriques Abreu, Jose Pereira Amorim, Ivana Isgum, Vishal Patel, Kevin Zhou, Steve Jiang, Ngan Le, Khoa Luu, Raphael Sznitman, Veronika Cheplygina, Samaneh Abbasi, Diana Mateus, Emanuele Trucco
PublisherSpringer Science and Business Media Deutschland GmbH
Pages63-72
Number of pages10
ISBN (Print)9783030611651
DOIs
StatePublished - 2020
Event3rd 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 - Lima, Peru
Duration: 4 Oct 20208 Oct 2020

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume12446 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference3rd 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
Country/TerritoryPeru
CityLima
Period4/10/208/10/20

Keywords

  • Saliency maps
  • Visualization
  • Weakly-supervised

Fingerprint

Dive into the research topics of 'Reliable Saliency Maps for Weakly-Supervised Localization of Disease Patterns'. Together they form a unique fingerprint.

Cite this