DALSA : domain adaptation for supervised learning from sparsely annotated MR images / Michael Goetz, Christian Weber, Franciszek Binczyk, Joanna Polanska, Rafal Tarnawski, Barbara Bobek-Billewicz, Ullrich Koethe, Jens Kleesiek, Bram Stieltjes, and Klaus H. Maier-Hein

We propose a new method that employs transfer learning techniques to effectively correct sampling selection errors introduced by sparse annotations during supervised learning for automated tumor segmentation. The practicality of current learning-based automated tissue classification approaches is severely impeded by their dependency on manually segmented training databases that need to be recreated for each scenario of application, site, or acquisition setup. The comprehensive annotation of reference datasets can be highly labor-intensive, complex, and error-prone. The proposed method derives high-quality classifiers for the different tissue classes from sparse and unambiguous annotations and employs domain adaptation techniques for effectively correcting sampling selection errors introduced by the sparse sampling. The new approach is validated on labeled, multi-modal MR images of 19 patients with malignant gliomas and by comparative analysis on the BraTS 2013 challenge data sets. Compared to training on fully labeled data, we reduced the time for labeling and training by a factor greater than 70 and 180 respectively without sacrificing accuracy. This dramatically eases the establishment and constant extension of large annotated databases in various scenarios and imaging setups and thus represents an important step towards practical applicability of learning-based approaches in tissue classification..

Medienart:

E-Artikel

Erscheinungsjahr:

Jan. 2016

2016

Erschienen:

Jan. 2016

Enthalten in:

Zur Gesamtaufnahme - volume:35

Enthalten in:

IEEE transactions on medical imaging - 35(2016), 1, Seite 184-196

Sprache:

Englisch

Beteiligte Personen:

Götz, Michael [VerfasserIn]
Weber, Christian [VerfasserIn]
Binczyk, Franciszek [VerfasserIn]
Polanska, Joanna [VerfasserIn]
Tarnawski, Rafal [VerfasserIn]
Bobek-Billewicz, Barbara [VerfasserIn]
Köthe, Ullrich [VerfasserIn]
Kleesiek, Jens Philipp, 1977- [VerfasserIn]
Stieltjes, Bram [VerfasserIn]
Maier-Hein, Klaus H. [VerfasserIn]

Links:

Volltext [lizenzpflichtig]

Themen:

Algorithms
Automated tissue classification
Automated tumor segmentation
Automatic multi-modal segmentation
Biomedical MRI
Brain Neoplasms
Brain tumor segmentation
Compressed sensing
DALSA
Decision Trees
Domain adaptation
Domain adaptation techniques
Domain-adaptation-for-supervised-learning-from-sparsely-annotation
Glioma
Humans
Image Processing, Computer-Assisted
Image classification
Image segmentation
Labeling
Learning-based approach
MR Images
Machine Learning
Magnetic Resonance Imaging
Malignant gliomas
Medical image processing
Noise
Random forest
Sampling selection errors
Sparse annotations
Sparse sampling
Tissue classes
Training
Training data
Transfer learning
Transfer learning techniques
Tumors
Tumours
Vegetation

Anmerkungen:

Gesehen am 25.06.2020

Umfang:

13

doi:

10.1109/TMI.2015.2463078

funding:

Förderinstitution / Projekttitel:

PPN (Katalog-ID):

1702179869

Zugang & Verfügbarkeit




Zugehörige Publikationen/Bände

    Haven't found what you're looking for?