How Can Data Augmentation Improve Attribution Maps for Disease Subtype Explainability?

Elina Thibeau-Sutre, Jelmer M. Wolterink, Olivier Colliot, Ninon Burgos

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

86 Downloads (Pure)

Abstract

As deep learning has been widely used for computer aided-diagnosis, we wished to know whether attribution maps obtained using gradient back-propagation could correctly highlight the patterns of disease subtypes discovered by a deep learning classifier. As the correctness of attribution maps is difficult to evaluate directly on medical images, we used synthetic data mimicking the difference between brain MRI of controls and demented patients to design more reliable evaluation criteria of attribution maps. We demonstrated that attribution maps may mix the regions associated with different subtypes for small data sets while they could accurately characterize both subtypes using a large data set. We then proposed simple data augmentation techniques and showed that they could improve the coherence of the explanations for a small data set.
Original languageEnglish
Title of host publicationSPIE Medical Imaging 2023
Subtitle of host publicationImage Processing
EditorsOlivier Colliot, Ivana Isgum
PublisherSPIE
ISBN (Electronic)9781510660335
DOIs
Publication statusPublished - 3 Apr 2023
EventSPIE Medical Imaging 2023 - San Diego, United States
Duration: 19 Feb 202324 Feb 2023

Conference

ConferenceSPIE Medical Imaging 2023
Country/TerritoryUnited States
CitySan Diego
Period19/02/2324/02/23

Fingerprint

Dive into the research topics of 'How Can Data Augmentation Improve Attribution Maps for Disease Subtype Explainability?'. Together they form a unique fingerprint.

Cite this