TUHH Open Research
Help
  • Log In
    New user? Click here to register.Have you forgotten your password?
  • English
  • Deutsch
  • Communities & Collections
  • Publications
  • Research Data
  • People
  • Institutions
  • Projects
  • Statistics
  1. Home
  2. TUHH
  3. Publications
  4. Combining Reconstruction-based Unsupervised Anomaly Detection with Supervised Segmentation for Brain MRIs
 
Options

Combining Reconstruction-based Unsupervised Anomaly Detection with Supervised Segmentation for Brain MRIs

Citation Link: https://doi.org/10.15480/882.14248
Publikationstyp
Conference Paper
Date Issued
2024-07
Sprache
English
Author(s)
Behrendt, Finn  
Medizintechnische und Intelligente Systeme E-1  
Bhattacharya, Debayan 
Medizintechnische und Intelligente Systeme E-1  
Maack, Lennart  
Medizintechnische und Intelligente Systeme E-1  
Krüger, Julia  
Opfer, Roland  
Schlaefer, Alexander  
Medizintechnische und Intelligente Systeme E-1  
TORE-DOI
10.15480/882.14248
TORE-URI
https://tore.tuhh.de/handle/11420/52965
First published in
Proceedings of machine learning research  
Number in series
250
Start Page
87
End Page
102
Citation
7th International Conference on Medical Imaging with Deep Learning, MIDL 2024
Contribution to Conference
7th International Conference on Medical Imaging with Deep Learning, MIDL 2024  
Publisher Link
https://openreview.net/forum?id=iWfUcg4FrD
Scopus ID
2-s2.0-85210625044
Peer Reviewed
true
In contrast to supervised deep learning approaches, unsupervised anomaly detection (UAD) methods can be trained with healthy data only and do not require pixel-level annotations, enabling the identification of unseen pathologies. While this is promising for clinical screening tasks, reconstruction-based UAD methods fall short in segmentation accuracy compared to supervised models. Therefore, self-supervised UAD approaches have been proposed to improve segmentation accuracy. Typically, synthetic anomalies are used to train a segmentation network in a supervised fashion. However, this approach does not effectively generalize to real pathologies. We propose a framework combining reconstruction-based and self-supervised UAD methods to improve both segmentation performance for known anomalies and generalization to unknown pathologies. The framework includes an unsupervised diffusion model trained on healthy data to produce pseudo-healthy reconstructions and a supervised Unet trained to delineate anomalies from deviations between input- reconstruction pairs. Besides the effective use of synthetic training data, this framework allows for weakly-supervised training with small annotated data sets, generalizing to unseen pathologies. Our results show that with our approach, utilizing annotated data sets during training can substantially improve the segmentation performance for in-domain data while maintaining the generalizability of reconstruction-based approaches to pathologies unseen during training
Subjects
Unsupervised Anomaly Detection | Diffusion Models | Brain MRI | Self Supervision
MLE@TUHH
DDC Class
006.3: Artificial Intelligence
610: Medicine, Health
510: Mathematics
Publication version
publishedVersion
Lizenz
https://creativecommons.org/licenses/by/4.0/
Loading...
Thumbnail Image
Name

225_Combining_Reconstruction_b.pdf

Type

Main Article

Size

2.59 MB

Format

Adobe PDF

TUHH
Weiterführende Links
  • Contact
  • Send Feedback
  • Cookie settings
  • Privacy policy
  • Impress
DSpace Software

Built with DSpace-CRIS software - Extension maintained and optimized by 4Science
Design by effective webwork GmbH

  • Deutsche NationalbibliothekDeutsche Nationalbibliothek
  • ORCiD Member OrganizationORCiD Member Organization
  • DataCiteDataCite
  • Re3DataRe3Data
  • OpenDOAROpenDOAR
  • OpenAireOpenAire
  • BASE Bielefeld Academic Search EngineBASE Bielefeld Academic Search Engine
Feedback