Classification of COPD with multiple instance learning

Veronika Cheplygina; Lauge Emil Borch Laurs Sørensen; David Tax; Jesper Johannes Holst Pedersen; Marco Loog; Marleen de Bruijne

doi:10.1109/ICPR.2014.268

Classification of COPD with multiple instance learning

Veronika Cheplygina, Lauge Emil Borch Laurs Sørensen, David Tax, Jesper Johannes Holst Pedersen, Marco Loog, Marleen de Bruijne

22 Citations (Scopus)

Abstract

Chronic obstructive pulmonary disease (COPD) is a lung disease where early detection benefits the survival rate. COPD can be quantified by classifying patches of computed tomography images, and combining patch labels into an overall diagnosis for the image. As labeled patches are often not available, image labels are propagated to the patches, incorrectly labeling healthy patches in COPD patients as being affected by the disease. We approach quantification of COPD from lung images as a multiple instance learning (MIL) problem, which is more suitable for such weakly labeled data. We investigate various MIL assumptions in the context of COPD and show that although a concept region with COPD-related disease patterns is present, considering the whole distribution of lung tissue patches improves the performance. The best method is based on averaging instances and obtains an AUC of 0.742, which is higher than the previously reported best of 0.713 on the same dataset. Using the full training set further increases performance to 0.776, which is significantly higher (DeLong test) than previous results.

Original language	English
Title of host publication	22nd International Conference on Pattern Recognition (ICPR) 2014
Number of pages	6
Publisher	IEEE
Publication date	4 Dec 2014
Pages	1508-1513
ISBN (Electronic)	978-1-4799-5208-3
DOIs	https://doi.org/10.1109/ICPR.2014.268
Publication status	Published - 4 Dec 2014
Event	International Conference on Pattern Recognition - Stockholm, Sweden Duration: 24 Aug 2014 → 28 Aug 2014

Conference

Conference	International Conference on Pattern Recognition
Country/Territory	Sweden
City	Stockholm
Period	24/08/2014 → 28/08/2014

Series	International Conference on Pattern Recognition
ISSN	1051-4651

Access to Document

10.1109/ICPR.2014.268

Classification of COPD with Multiple Instance LearningFinal published version, 209 KB

Cite this

Classification of COPD with multiple instance learning. / Cheplygina, Veronika; Sørensen, Lauge Emil Borch Laurs; Tax, David et al.

22nd International Conference on Pattern Recognition (ICPR) 2014. IEEE, 2014. p. 1508-1513 (International Conference on Pattern Recognition).

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Cheplygina, V, Sørensen, LEBL, Tax, D, Pedersen, JJH, Loog, M & de Bruijne, M 2014, Classification of COPD with multiple instance learning. in 22nd International Conference on Pattern Recognition (ICPR) 2014. IEEE, International Conference on Pattern Recognition, pp. 1508-1513, International Conference on Pattern Recognition, Stockholm, Sweden, 24/08/2014. https://doi.org/10.1109/ICPR.2014.268

@inproceedings{9342944bd632445b828d3659598f6139,

title = "Classification of COPD with multiple instance learning",

abstract = "Chronic obstructive pulmonary disease (COPD) is a lung disease where early detection benefits the survival rate. COPD can be quantified by classifying patches of computed tomography images, and combining patch labels into an overall diagnosis for the image. As labeled patches are often not available, image labels are propagated to the patches, incorrectly labeling healthy patches in COPD patients as being affected by the disease. We approach quantification of COPD from lung images as a multiple instance learning (MIL) problem, which is more suitable for such weakly labeled data. We investigate various MIL assumptions in the context of COPD and show that although a concept region with COPD-related disease patterns is present, considering the whole distribution of lung tissue patches improves the performance. The best method is based on averaging instances and obtains an AUC of 0.742, which is higher than the previously reported best of 0.713 on the same dataset. Using the full training set further increases performance to 0.776, which is significantly higher (DeLong test) than previous results.",

author = "Veronika Cheplygina and S{\o}rensen, {Lauge Emil Borch Laurs} and David Tax and Pedersen, {Jesper Johannes Holst} and Marco Loog and {de Bruijne}, Marleen",

year = "2014",

month = dec,

day = "4",

doi = "10.1109/ICPR.2014.268",

language = "English",

series = "International Conference on Pattern Recognition",

publisher = "IEEE",

pages = "1508--1513",

booktitle = "22nd International Conference on Pattern Recognition (ICPR) 2014",

note = "International Conference on Pattern Recognition, ICPR ; Conference date: 24-08-2014 Through 28-08-2014",

}

TY - GEN

T1 - Classification of COPD with multiple instance learning

AU - Cheplygina, Veronika

AU - Sørensen, Lauge Emil Borch Laurs

AU - Tax, David

AU - Pedersen, Jesper Johannes Holst

AU - Loog, Marco

AU - de Bruijne, Marleen

PY - 2014/12/4

Y1 - 2014/12/4

N2 - Chronic obstructive pulmonary disease (COPD) is a lung disease where early detection benefits the survival rate. COPD can be quantified by classifying patches of computed tomography images, and combining patch labels into an overall diagnosis for the image. As labeled patches are often not available, image labels are propagated to the patches, incorrectly labeling healthy patches in COPD patients as being affected by the disease. We approach quantification of COPD from lung images as a multiple instance learning (MIL) problem, which is more suitable for such weakly labeled data. We investigate various MIL assumptions in the context of COPD and show that although a concept region with COPD-related disease patterns is present, considering the whole distribution of lung tissue patches improves the performance. The best method is based on averaging instances and obtains an AUC of 0.742, which is higher than the previously reported best of 0.713 on the same dataset. Using the full training set further increases performance to 0.776, which is significantly higher (DeLong test) than previous results.

AB - Chronic obstructive pulmonary disease (COPD) is a lung disease where early detection benefits the survival rate. COPD can be quantified by classifying patches of computed tomography images, and combining patch labels into an overall diagnosis for the image. As labeled patches are often not available, image labels are propagated to the patches, incorrectly labeling healthy patches in COPD patients as being affected by the disease. We approach quantification of COPD from lung images as a multiple instance learning (MIL) problem, which is more suitable for such weakly labeled data. We investigate various MIL assumptions in the context of COPD and show that although a concept region with COPD-related disease patterns is present, considering the whole distribution of lung tissue patches improves the performance. The best method is based on averaging instances and obtains an AUC of 0.742, which is higher than the previously reported best of 0.713 on the same dataset. Using the full training set further increases performance to 0.776, which is significantly higher (DeLong test) than previous results.

U2 - 10.1109/ICPR.2014.268

DO - 10.1109/ICPR.2014.268

M3 - Article in proceedings

T3 - International Conference on Pattern Recognition

SP - 1508

EP - 1513

BT - 22nd International Conference on Pattern Recognition (ICPR) 2014

PB - IEEE

T2 - International Conference on Pattern Recognition

Y2 - 24 August 2014 through 28 August 2014

ER -

Classification of COPD with multiple instance learning

Abstract

Conference

Access to Document

Fingerprint

Cite this