Please use this identifier to cite or link to this item:
Publisher DOI: 10.1007/s11548-021-02451-9
Title: Three-dimensional deep learning with spatial erasing for unsupervised anomaly segmentation in brain MRI
Language: English
Authors: Bengs, Marcel 
Behrendt, Finn 
Krüger, Julia 
Opfer, Roland 
Schlaefer, Alexander 
Keywords: 3D autoencoder;Anomaly;Brain MRI;Segmentation;Unsupervised
Issue Date: 12-Jul-2021
Publisher: Springer
Source: International Journal of Computer Assisted Radiology and Surgery 16 (9): 1413-1423 (2021-09)
Journal: International journal of computer assisted radiology and surgery 
Abstract (english): 
Purpose: Brain Magnetic Resonance Images (MRIs) are essential for the diagnosis of neurological diseases. Recently, deep learning methods for unsupervised anomaly detection (UAD) have been proposed for the analysis of brain MRI. These methods rely on healthy brain MRIs and eliminate the requirement of pixel-wise annotated data compared to supervised deep learning. While a wide range of methods for UAD have been proposed, these methods are mostly 2D and only learn from MRI slices, disregarding that brain lesions are inherently 3D and the spatial context of MRI volumes remains unexploited.

Methods: We investigate whether using increased spatial context by using MRI volumes combined with spatial erasing leads to improved unsupervised anomaly segmentation performance compared to learning from slices. We evaluate and compare 2D variational autoencoder (VAE) to their 3D counterpart, propose 3D input erasing, and systemically study the impact of the data set size on the performance. Results: Using two publicly available segmentation data sets for evaluation, 3D VAEs outperform their 2D counterpart, highlighting the advantage of volumetric context. Also, our 3D erasing methods allow for further performance improvements. Our best performing 3D VAE with input erasing leads to an average DICE score of 31.40% compared to 25.76% for the 2D VAE.

Conclusions: We propose 3D deep learning methods for UAD in brain MRI combined with 3D erasing and demonstrate that 3D methods clearly outperform their 2D counterpart for anomaly segmentation. Also, our spatial erasing method allows for further performance improvements and reduces the requirement for large data sets.
DOI: 10.15480/882.3761
ISSN: 1861-6429
Institute: Medizintechnische und Intelligente Systeme E-1 
Document Type: Article
More Funding information: This work was partially funded by Grant Number ZF4026303TS9.
Project: Projekt DEAL 
License: CC BY 4.0 (Attribution) CC BY 4.0 (Attribution)
Appears in Collections:Publications with fulltext

Files in This Item:
File Description SizeFormat
Bengs2021_Article_Three-dimensionalDeepLearningW.pdfVerlagsversion3,53 MBAdobe PDFView/Open
Show full item record

Google ScholarTM


Note about this record

Cite this record


This item is licensed under a Creative Commons License Creative Commons