Joint Weakly and Semi-Supervised Deep Learning for Localization and Classification of Masses in Breast Ultrasound Images
- Authors
- Shin, Seung Yeon; Lee, Soochahn; Yun, Il Dong; Kim, Sun Mi; Lee, Kyoung Mu
- Issue Date
- Mar-2019
- Publisher
- Institute of Electrical and Electronics Engineers
- Keywords
- Breast ultrasound; convolutional neural networks; mass classification; mass localization; semi-supervised learning; weakly supervised learning
- Citation
- IEEE Transactions on Medical Imaging, v.38, no.3, pp 762 - 774
- Pages
- 13
- Indexed
- SCI
SCIE
SCOPUS
- Journal Title
- IEEE Transactions on Medical Imaging
- Volume
- 38
- Number
- 3
- Start Page
- 762
- End Page
- 774
- URI
- https://scholarworks.bwise.kr/erica/handle/2021.sw.erica/115185
- DOI
- 10.1109/TMI.2018.2872031
- ISSN
- 0278-0062
1558-254X
- Abstract
- We propose a framework for localization and classification of masses in breast ultrasound images. We have experimentally found that training convolutional neural network-based mass detectors with large, weakly annotated datasets presents a non-trivial problem, while overfitting may occur with those trained with small, strongly annotated datasets. To overcome these problems, we use a weakly annotated dataset together with a smaller strongly annotated dataset in a hybrid manner. We propose a systematic weakly and semi-supervised training scenario with appropriate training loss selection. Experimental results show that the proposed method can successfully localize and classify masses with less annotation effort. The results trained with only 10 strongly annotated images along with weakly annotated images were comparable to results trained from 800 strongly annotated images, with the 95% confidence interval (CI) of difference -3%-5%, in terms of the correct localization (CorLoc) measure, which is the ratio of images with intersection over union with ground truth higher than 0.5. With the same number of strongly annotated images, additional weakly annotated images can be incorporated to give a 4.5% point increase in CorLoc, from 80% to 84.50% (with 95% CIs 76%-83.75% and 81%-88%). The effects of different algorithmic details and varied amount of data are presented through ablative analysis. © 1982-2012 IEEE.
- Files in This Item
-
Go to Link
- Appears in
Collections - COLLEGE OF ENGINEERING SCIENCES > SCHOOL OF ELECTRICAL ENGINEERING > 1. Journal Articles
Items in ScholarWorks are protected by copyright, with all rights reserved, unless otherwise indicated.