First U-Net Layers Contain More Domain Specific Information Than the Last Ones

Boris Shirokikh, Ivan Zakazov, Alexey Chernyavskiy, Irina Fedulova, Mikhail Belyaev

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

3 Citations (Scopus)

Abstract

MRI scans appearance significantly depends on scanning protocols and, consequently, the data-collection institution. These variations between clinical sites result in dramatic drops of CNN segmentation quality on unseen domains. Many of the recently proposed MRI domain adaptation methods operate with the last CNN layers to suppress domain shift. At the same time, the core manifestation of MRI variability is a considerable diversity of image intensities. We hypothesize that these differences can be eliminated by modifying the first layers rather than the last ones. To validate this simple idea, we conducted a set of experiments with brain MRI scans from six domains. Our results demonstrate that 1) domain-shift may deteriorate the quality even for a simple brain extraction segmentation task (surface Dice Score drops from 0.85–0.89 even to 0.09); 2) fine-tuning of the first layers significantly outperforms fine-tuning of the last layers in almost all supervised domain adaptation setups. Moreover, fine-tuning of the first layers is a better strategy than fine-tuning of the whole network, if the amount of annotated data from the new domain is strictly limited.

Original languageEnglish
Title of host publicationDomain Adaptation and Representation Transfer, and Distributed and Collaborative Learning - 2nd MICCAI Workshop, DART 2020, and 1st MICCAI Workshop, DCL 2020, Held in Conjunction with MICCAI 2020, Proceedings
EditorsShadi Albarqouni, Spyridon Bakas, Konstantinos Kamnitsas, M. Jorge Cardoso, Bennett Landman, Wenqi Li, Fausto Milletari, Nicola Rieke, Holger Roth, Daguang Xu, Ziyue Xu
PublisherSpringer Science and Business Media Deutschland GmbH
Pages117-126
Number of pages10
ISBN (Print)9783030605476
DOIs
Publication statusPublished - 2020
Event2nd MICCAI Workshop on Domain Adaptation and Representation Transfer, DART 2020, and the 1st MICCAI Workshop on Distributed and Collaborative Learning, DCL 2020, held in conjunction with the 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)
Volume12444 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference2nd MICCAI Workshop on Domain Adaptation and Representation Transfer, DART 2020, and the 1st MICCAI Workshop on Distributed and Collaborative Learning, DCL 2020, held in conjunction with the Medical Image Computing and Computer Assisted Intervention, MICCAI 2020
Country/TerritoryPeru
CityLima
Period4/10/208/10/20

Keywords

  • CNN
  • Domain adaptation
  • MRI
  • Segmentation

Fingerprint

Dive into the research topics of 'First U-Net Layers Contain More Domain Specific Information Than the Last Ones'. Together they form a unique fingerprint.

Cite this