CARE: Class Attention to Regions of Lesion for Classification on Imbalanced Data

Jiaxin  Zhuang; Jiabin  Cai; Ruixuan  Wang; Jianguo Zhang; Weishi Zheng

CARE: Class Attention to Regions of Lesion for Classification on Imbalanced Data

Jiaxin Zhuang
, Jiabin Cai
, Ruixuan Wang
, Jianguo Zhang
, Weishi Zheng

Computing

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

57 Downloads (Pure)

Abstract

To date, it is still an open and challenging problem for intelligent diagnosis systems to effectively learn from imbalanced data, especially with large samples of common diseases and much smaller samples of rare ones. Inspired by the process of human learning, this paper proposes a novel and effective way to embed attention into the machine learning process, particularly for learning characteristics of rare diseases. This approach does not change architectures of the original CNN classifiers and therefore can directly plug and play for any existing CNN architecture. Comprehensive experiments on a skin lesion dataset and a pneumonia chest X-ray dataset showed that paying attention to lesion regions of rare diseases during learning not only improved the classification performance on rare diseases, but also on the mean class accuracy.

Original language	English
Title of host publication	Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning
Editors	M. Jorge Cardoso, Aasa Feragen, Ben Glocker, Ender Konukoglu, Ipek Oguz, Gozde Unal, Tom Vercauteren
Pages	588-597
Number of pages	10
Volume	102
Publication status	Published - 2019

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Attention
Imbalanced Data
Small Samples
Skin Lesion
Pneumonia Chest X-ray

Access to Document

Author Accepted ManuscriptAccepted author manuscript, 1.06 MBLicence: CC BY

Cite this

@inproceedings{a0b8796397f14e62a8a6752e9a2ba71f,

title = "CARE: Class Attention to Regions of Lesion for Classification on Imbalanced Data",

abstract = "To date, it is still an open and challenging problem for intelligent diagnosis systems to effectively learn from imbalanced data, especially with large samples of common diseases and much smaller samples of rare ones. Inspired by the process of human learning, this paper proposes a novel and effective way to embed attention into the machine learning process, particularly for learning characteristics of rare diseases. This approach does not change architectures of the original CNN classifiers and therefore can directly plug and play for any existing CNN architecture. Comprehensive experiments on a skin lesion dataset and a pneumonia chest X-ray dataset showed that paying attention to lesion regions of rare diseases during learning not only improved the classification performance on rare diseases, but also on the mean class accuracy.",

keywords = "Attention, Imbalanced Data, Small Samples, Skin Lesion, Pneumonia Chest X-ray",

author = "Jiaxin Zhuang and Jiabin Cai and Ruixuan Wang and Jianguo Zhang and Weishi Zheng",

year = "2019",

language = "English",

volume = "102",

pages = "588--597",

editor = "Cardoso, \{M. Jorge\} and Aasa Feragen and Ben Glocker and Ender Konukoglu and Ipek Oguz and Gozde Unal and Tom Vercauteren",

booktitle = "Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning",

}

CARE: Class Attention to Regions of Lesion for Classification on Imbalanced Data. / Zhuang, Jiaxin ; Cai, Jiabin ; Wang, Ruixuan et al.
Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning. ed. / M. Jorge Cardoso; Aasa Feragen; Ben Glocker; Ender Konukoglu; Ipek Oguz; Gozde Unal; Tom Vercauteren. Vol. 102 2019. p. 588-597.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - CARE

T2 - Class Attention to Regions of Lesion for Classification on Imbalanced Data

AU - Zhuang, Jiaxin

AU - Cai, Jiabin

AU - Wang, Ruixuan

AU - Zhang, Jianguo

AU - Zheng, Weishi

PY - 2019

Y1 - 2019

N2 - To date, it is still an open and challenging problem for intelligent diagnosis systems to effectively learn from imbalanced data, especially with large samples of common diseases and much smaller samples of rare ones. Inspired by the process of human learning, this paper proposes a novel and effective way to embed attention into the machine learning process, particularly for learning characteristics of rare diseases. This approach does not change architectures of the original CNN classifiers and therefore can directly plug and play for any existing CNN architecture. Comprehensive experiments on a skin lesion dataset and a pneumonia chest X-ray dataset showed that paying attention to lesion regions of rare diseases during learning not only improved the classification performance on rare diseases, but also on the mean class accuracy.

AB - To date, it is still an open and challenging problem for intelligent diagnosis systems to effectively learn from imbalanced data, especially with large samples of common diseases and much smaller samples of rare ones. Inspired by the process of human learning, this paper proposes a novel and effective way to embed attention into the machine learning process, particularly for learning characteristics of rare diseases. This approach does not change architectures of the original CNN classifiers and therefore can directly plug and play for any existing CNN architecture. Comprehensive experiments on a skin lesion dataset and a pneumonia chest X-ray dataset showed that paying attention to lesion regions of rare diseases during learning not only improved the classification performance on rare diseases, but also on the mean class accuracy.

KW - Attention

KW - Imbalanced Data

KW - Small Samples

KW - Skin Lesion

KW - Pneumonia Chest X-ray

UR - http://proceedings.mlr.press/v102/zhuang19a.html

UR - http://proceedings.mlr.press/v102/

M3 - Conference contribution

VL - 102

SP - 588

EP - 597

BT - Proceedings of The 2nd International Conference on Medical Imaging with Deep Learning

A2 - Cardoso, M. Jorge

A2 - Feragen, Aasa

A2 - Glocker, Ben

A2 - Konukoglu, Ender

A2 - Oguz, Ipek

A2 - Unal, Gozde

A2 - Vercauteren, Tom

ER -

CARE: Class Attention to Regions of Lesion for Classification on Imbalanced Data

Abstract

UN SDGs

Keywords

Access to Document

Other files and links

Fingerprint

Cite this