A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans

Quan Chen; Xiang Xu; Shiliang Hu; Xiao Li; Qing Zou; Yunpeng Li

doi:10.1117/12.2279021

A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans

Quan Chen, Xiang Xu, Shiliang Hu, Xiao Li, Qing Zou, Yunpeng Li

Radiation Medicine

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

33 Scopus citations

Abstract

Deep learning has shown a great potential in computer aided diagnosis. However, in many applications, large dataset is not available. This makes the training of a sophisticated deep learning neural network (DNN) difficult. In this study, we demonstrated that with transfer learning, we can quickly retrain start-of-the-art DNN models with limited data provided by the prostateX challenge. The training data consists of 330 lesions, only 78 were clinical significant. Efforts were made to balance the data during training. We used ImageNet pre-trained inceptionV3 and Vgg-16 model and obtained AUC of 0.81 and 0.83 respectively on the prostateX test data, good for a 4th place finish. We noticed that models trained for different prostate zone has different sensitivity. Applying scaling factors before merging the result improves the AUC for the final result.

Original language	English
Title of host publication	Medical Imaging 2017
Subtitle of host publication	Computer-Aided Diagnosis
Editors	Nicholas A. Petrick, Samuel G. Armato
ISBN (Electronic)	9781510607132
DOIs	https://doi.org/10.1117/12.2279021
State	Published - 2017
Event	Medical Imaging 2017: Computer-Aided Diagnosis - Orlando, United States Duration: Feb 13 2017 → Feb 16 2017

Publication series

Name	Progress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume	10134
ISSN (Print)	1605-7422

Conference

Conference	Medical Imaging 2017: Computer-Aided Diagnosis
Country/Territory	United States
City	Orlando
Period	2/13/17 → 2/16/17

Bibliographical note

Publisher Copyright:
© 2017 SPIE.

Keywords

Computer aided diagnosis (CADx)
Convolution neural Network (CNN)
Deep learning neural network (DNN)
Multiparameter MRI (mpMRI)
Prostate cancer
ProstateX challenge
Transfer learning
VGG

ASJC Scopus subject areas

Electronic, Optical and Magnetic Materials
Atomic and Molecular Physics, and Optics
Radiology Nuclear Medicine and imaging
Biomaterials

UN SDGs

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

Access to Document

10.1117/12.2279021

Cite this

Chen, Q., Xu, X., Hu, S., Li, X., Zou, Q., & Li, Y. (2017). A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans. In N. A. Petrick, & S. G. Armato (Eds.), Medical Imaging 2017: Computer-Aided Diagnosis Article 101344F (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10134). https://doi.org/10.1117/12.2279021

@inproceedings{5c20a3972f6443d89b969e5bb4a20cab,

title = "A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans",

abstract = "Deep learning has shown a great potential in computer aided diagnosis. However, in many applications, large dataset is not available. This makes the training of a sophisticated deep learning neural network (DNN) difficult. In this study, we demonstrated that with transfer learning, we can quickly retrain start-of-the-art DNN models with limited data provided by the prostateX challenge. The training data consists of 330 lesions, only 78 were clinical significant. Efforts were made to balance the data during training. We used ImageNet pre-trained inceptionV3 and Vgg-16 model and obtained AUC of 0.81 and 0.83 respectively on the prostateX test data, good for a 4th place finish. We noticed that models trained for different prostate zone has different sensitivity. Applying scaling factors before merging the result improves the AUC for the final result.",

keywords = "Computer aided diagnosis (CADx), Convolution neural Network (CNN), Deep learning neural network (DNN), Multiparameter MRI (mpMRI), Prostate cancer, ProstateX challenge, Transfer learning, VGG",

author = "Quan Chen and Xiang Xu and Shiliang Hu and Xiao Li and Qing Zou and Yunpeng Li",

note = "Publisher Copyright: {\textcopyright} 2017 SPIE.; Medical Imaging 2017: Computer-Aided Diagnosis ; Conference date: 13-02-2017 Through 16-02-2017",

year = "2017",

doi = "10.1117/12.2279021",

language = "English",

series = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",

editor = "Petrick, {Nicholas A.} and Armato, {Samuel G.}",

booktitle = "Medical Imaging 2017",

}

Chen, Q, Xu, X, Hu, S, Li, X, Zou, Q & Li, Y 2017, A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans. in NA Petrick & SG Armato (eds), Medical Imaging 2017: Computer-Aided Diagnosis., 101344F, Progress in Biomedical Optics and Imaging - Proceedings of SPIE, vol. 10134, Medical Imaging 2017: Computer-Aided Diagnosis, Orlando, United States, 2/13/17. https://doi.org/10.1117/12.2279021

A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans. / Chen, Quan; Xu, Xiang; Hu, Shiliang et al.
Medical Imaging 2017: Computer-Aided Diagnosis. ed. / Nicholas A. Petrick; Samuel G. Armato. 2017. 101344F (Progress in Biomedical Optics and Imaging - Proceedings of SPIE; Vol. 10134).

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

TY - GEN

T1 - A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans

AU - Chen, Quan

AU - Xu, Xiang

AU - Hu, Shiliang

AU - Li, Xiao

AU - Zou, Qing

AU - Li, Yunpeng

PY - 2017

Y1 - 2017

N2 - Deep learning has shown a great potential in computer aided diagnosis. However, in many applications, large dataset is not available. This makes the training of a sophisticated deep learning neural network (DNN) difficult. In this study, we demonstrated that with transfer learning, we can quickly retrain start-of-the-art DNN models with limited data provided by the prostateX challenge. The training data consists of 330 lesions, only 78 were clinical significant. Efforts were made to balance the data during training. We used ImageNet pre-trained inceptionV3 and Vgg-16 model and obtained AUC of 0.81 and 0.83 respectively on the prostateX test data, good for a 4th place finish. We noticed that models trained for different prostate zone has different sensitivity. Applying scaling factors before merging the result improves the AUC for the final result.

AB - Deep learning has shown a great potential in computer aided diagnosis. However, in many applications, large dataset is not available. This makes the training of a sophisticated deep learning neural network (DNN) difficult. In this study, we demonstrated that with transfer learning, we can quickly retrain start-of-the-art DNN models with limited data provided by the prostateX challenge. The training data consists of 330 lesions, only 78 were clinical significant. Efforts were made to balance the data during training. We used ImageNet pre-trained inceptionV3 and Vgg-16 model and obtained AUC of 0.81 and 0.83 respectively on the prostateX test data, good for a 4th place finish. We noticed that models trained for different prostate zone has different sensitivity. Applying scaling factors before merging the result improves the AUC for the final result.

KW - Computer aided diagnosis (CADx)

KW - Convolution neural Network (CNN)

KW - Deep learning neural network (DNN)

KW - Multiparameter MRI (mpMRI)

KW - Prostate cancer

KW - ProstateX challenge

KW - Transfer learning

KW - VGG

UR - http://www.scopus.com/inward/record.url?scp=85020281132&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85020281132&partnerID=8YFLogxK

U2 - 10.1117/12.2279021

DO - 10.1117/12.2279021

M3 - Conference contribution

AN - SCOPUS:85020281132

T3 - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

BT - Medical Imaging 2017

A2 - Petrick, Nicholas A.

A2 - Armato, Samuel G.

T2 - Medical Imaging 2017: Computer-Aided Diagnosis

Y2 - 13 February 2017 through 16 February 2017

ER -

A transfer learning approach for classification of clinical significant prostate cancers from mpMRI scans

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this