TY - JOUR
T1 - Joint Learning for Pneumonia Classification and Segmentation on Medical Images
AU - Liu, Shaobo
AU - Zhong, Xin
AU - Shih, Frank Y.
N1 - Funding Information:
Dr. Shih held a Visiting Professor position at Princeton University, Columbia University, National Taiwan University, National Institute of Informatics, Tokyo, Conservatoire National Des Arts Et Métiers, Paris, and Nanjing University of Information Science and Technology, China. He is an internationally renowned scholar and currently serves as the Editor-in-Chief for the International Journal of Pattern Recognition and Arti¯cial Intelligence. He was the Editor-in-Chief for the International Journal of Multimedia Intelligence and Security. In addition, he is on the Editorial Boards of 12 international journals. He has served as a Steering Member, Session Chair and Committee Member for numerous professional conferences and workshops. He has received numerous grants from the National Science Foundation (NSF), NIH, NASA, Navy and Air Force and Industry. He has won the Research Initiation Award from NSF, the Outstanding Teaching Award and the Board of Overseers Excellence in Research Award from NJIT and the Best Paper Awards from journals and conferences.
Publisher Copyright:
© 2021 World Scientific Publishing Company.
PY - 2021/4
Y1 - 2021/4
N2 - Chest X-ray images are notoriously difficult to analyze due to the noisy nature. Automatic identification of pneumonia on medical images has attracted intensive study recently. In this paper, a novel joint-Task architecture that can learn pneumonia classification and segmentation simultaneously is presented. Two modules, including an image preprocessing module and an attention module, are developed to improve both the classification and segmentation accuracies. Results from the experiments performed on the massive dataset of the Radiology Society of North America have confirmed its superiority over the other existing methods. The classification test accuracy is improved from 0.89 to 0.95, and the segmentation model achieves an improved mean precision result of 0.58-0.78. Finally, two weakly supervised learning methods, class-saliency map and Grad-CAM, are used to highlight the corresponding pixels or areas which have significant influence on the classification model, such that the refined segmentation can focus on the correct areas with high confidence.
AB - Chest X-ray images are notoriously difficult to analyze due to the noisy nature. Automatic identification of pneumonia on medical images has attracted intensive study recently. In this paper, a novel joint-Task architecture that can learn pneumonia classification and segmentation simultaneously is presented. Two modules, including an image preprocessing module and an attention module, are developed to improve both the classification and segmentation accuracies. Results from the experiments performed on the massive dataset of the Radiology Society of North America have confirmed its superiority over the other existing methods. The classification test accuracy is improved from 0.89 to 0.95, and the segmentation model achieves an improved mean precision result of 0.58-0.78. Finally, two weakly supervised learning methods, class-saliency map and Grad-CAM, are used to highlight the corresponding pixels or areas which have significant influence on the classification model, such that the refined segmentation can focus on the correct areas with high confidence.
KW - Deep learning
KW - VGG16
KW - morphological neural networks
KW - pneumonia classification
KW - segmentation
UR - http://www.scopus.com/inward/record.url?scp=85097526205&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85097526205&partnerID=8YFLogxK
U2 - 10.1142/S0218001421570032
DO - 10.1142/S0218001421570032
M3 - Article
AN - SCOPUS:85097526205
SN - 0218-0014
VL - 35
JO - International Journal of Pattern Recognition and Artificial Intelligence
JF - International Journal of Pattern Recognition and Artificial Intelligence
IS - 5
M1 - 2157003
ER -