Generalization of prostate cancer classification for multiple sites using deep learning

Research output: Chapter in Book/Report/Conference proceedingPaper in conference proceeding

Standard

Generalization of prostate cancer classification for multiple sites using deep learning. / Arvidsson, Ida; Overgaard, Niels Christian; Marginean, Felicia Elena; Krzyzanowska, Agnieszka; Bjartell, Anders; Astrom, Kalle; Heyden, Anders.

2018 IEEE 15th International Symposium on Biomedical Imaging, ISBI 2018. Vol. 2018-April IEEE Computer Society, 2018. p. 191-194.

Research output: Chapter in Book/Report/Conference proceedingPaper in conference proceeding

Harvard

Arvidsson, I, Overgaard, NC, Marginean, FE, Krzyzanowska, A, Bjartell, A, Astrom, K & Heyden, A 2018, Generalization of prostate cancer classification for multiple sites using deep learning. in 2018 IEEE 15th International Symposium on Biomedical Imaging, ISBI 2018. vol. 2018-April, IEEE Computer Society, pp. 191-194, 15th IEEE International Symposium on Biomedical Imaging, ISBI 2018, Washington, United States, 2018/04/04. https://doi.org/10.1109/ISBI.2018.8363552

APA

CBE

MLA

Arvidsson, Ida et al. "Generalization of prostate cancer classification for multiple sites using deep learning". 2018 IEEE 15th International Symposium on Biomedical Imaging, ISBI 2018. IEEE Computer Society. 2018, 191-194. https://doi.org/10.1109/ISBI.2018.8363552

Vancouver

Arvidsson I, Overgaard NC, Marginean FE, Krzyzanowska A, Bjartell A, Astrom K et al. Generalization of prostate cancer classification for multiple sites using deep learning. In 2018 IEEE 15th International Symposium on Biomedical Imaging, ISBI 2018. Vol. 2018-April. IEEE Computer Society. 2018. p. 191-194 https://doi.org/10.1109/ISBI.2018.8363552

Author

RIS

TY - GEN

T1 - Generalization of prostate cancer classification for multiple sites using deep learning

AU - Arvidsson, Ida

AU - Overgaard, Niels Christian

AU - Marginean, Felicia Elena

AU - Krzyzanowska, Agnieszka

AU - Bjartell, Anders

AU - Astrom, Kalle

AU - Heyden, Anders

PY - 2018/5/23

Y1 - 2018/5/23

N2 - Deep learning has the potential to drastically increase the accuracy and efficiency of prostate cancer diagnosis, which would be of uttermost use. Today the diagnosis is determined manually from H&E stained specimens using a light microscope. In this paper several different approaches based on convolutional neural networks for prostate cancer classification are presented and compared, using three different datasets with different origins. The issue that algorithms trained on a certain site might not generalize to other sites, due to for example inevitable stain variations, is highlighted. Two different techniques to overcome this complication are compared; by training the networks using color augmentation and by using digital stain separation. Furthermore, the potential of using an autoencoder to get a more efficient downsampling is investigated, which turned out to be the method giving the best generalization. We achieve accuracies of 95% for classification of benign versus malignant tissue and 81% for Gleason grading for data from the same site as the training data. The corresponding accuracies for images from other sites are in average 88% and 52% respectively.

AB - Deep learning has the potential to drastically increase the accuracy and efficiency of prostate cancer diagnosis, which would be of uttermost use. Today the diagnosis is determined manually from H&E stained specimens using a light microscope. In this paper several different approaches based on convolutional neural networks for prostate cancer classification are presented and compared, using three different datasets with different origins. The issue that algorithms trained on a certain site might not generalize to other sites, due to for example inevitable stain variations, is highlighted. Two different techniques to overcome this complication are compared; by training the networks using color augmentation and by using digital stain separation. Furthermore, the potential of using an autoencoder to get a more efficient downsampling is investigated, which turned out to be the method giving the best generalization. We achieve accuracies of 95% for classification of benign versus malignant tissue and 81% for Gleason grading for data from the same site as the training data. The corresponding accuracies for images from other sites are in average 88% and 52% respectively.

KW - Autoencoder

KW - Convolutional neural network

KW - Digital stain separation

KW - Gleason grade

KW - Prostate cancer

UR - http://www.scopus.com/inward/record.url?scp=85048138355&partnerID=8YFLogxK

U2 - 10.1109/ISBI.2018.8363552

DO - 10.1109/ISBI.2018.8363552

M3 - Paper in conference proceeding

AN - SCOPUS:85048138355

VL - 2018-April

SP - 191

EP - 194

BT - 2018 IEEE 15th International Symposium on Biomedical Imaging, ISBI 2018

PB - IEEE Computer Society

T2 - 15th IEEE International Symposium on Biomedical Imaging, ISBI 2018

Y2 - 4 April 2018 through 7 April 2018

ER -