PuSH - Publikationsserver des Helmholtz Zentrums München

Kandemir, M.* ; Feuchtinger, A. ; Walch, A.K. ; Hamprecht, F.A.*

Digital pathology: Multiple instance learning can detect Barrett's cancer.

In: Proceedings (IEEE International Symposium on Biomedical Imaging, ISBI, 29. April - 02.May 2014, Beijing, China). Piscataway, NJ: IEEE, 2014. 1348-1351
Verlagsversion bestellen
We study diagnosis of Barrett's cancer from hematoxylin & eosin (H & E) stained histopathological biopsy images using multiple instance learning (MIL). We partition tissue cores into rectangular patches, and construct a feature vector consisting of a large set of cell-level and patch-level features for each patch. In MIL terms, we treat each tissue core as a bag (group of instances with a single group-level ground-truth label) and each patch an instance. After a benchmarking study on several MIL approaches, we find that a graph-based MIL algorithm, mi-Graph [1], gives the best performance (87% accuracy, 0.93 AUC), due to its inherent suitability to bags with spatially-correlated instances. In patch-level diagnosis, we reach 82% accuracy and 0.89 AUC using Bayesian logistic regression. We also pursue a study on feature importance, which shows that patch-level color and texture features and cell-level features all have significant contribution to prediction.
Weitere Metriken?
Zusatzinfos bearbeiten [➜Einloggen]
Publikationstyp Artikel: Konferenzbeitrag
Schlagwörter Cancer Diagnosis ; Histopathological Tissue Imaging ; Multiple Instance Learning
ISBN 9781467319591
Konferenztitel IEEE International Symposium on Biomedical Imaging, ISBI
Konferzenzdatum 29. April - 02.May 2014
Konferenzort Beijing, China
Konferenzband Proceedings
Quellenangaben Band: , Heft: , Seiten: 1348-1351 Artikelnummer: , Supplement: ,
Verlag IEEE
Verlagsort Piscataway, NJ
Begutachtungsstatus