Fine-Tuning and training of densenet for histopathology image representation using TCGA diagnostic slides.

Autor: Riasatian A; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Babaie M; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada. Electronic address: mbabaie@uwaterloo.ca., Maleki D; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Kalra S; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Valipour M; School of Computer Science, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Hemati S; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Zaveri M; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Safarpoor A; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Shafiei S; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Afshari M; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Rasoolijaberi M; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Sikaroudi M; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Adnan M; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Shah S; Huron Digital Pathology, 1620 King Street North, St. Jacobs, ON, Canada., Choi C; Huron Digital Pathology, 1620 King Street North, St. Jacobs, ON, Canada., Damaskinos S; Huron Digital Pathology, 1620 King Street North, St. Jacobs, ON, Canada., Campbell CJ; Department of Pathology and Molecular Medicine, McMaster University, Hamilton, Canada., Diamandis P; Laboratory Medicine and Pathobiology, University of Toronto, ON, Canada., Pantanowitz L; Department of Pathology, University of Pittsburgh Medical Center, PA, USA., Kashani H; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada., Ghodsi A; School of Computer Science, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada; Vector Institute, 661 University Ave Suite 710, Toronto, ON, Canada., Tizhoosh HR; Kimia Lab, University of Waterloo, 200 University Ave. W., Waterloo, ON, Canada; Vector Institute, 661 University Ave Suite 710, Toronto, ON, Canada. Electronic address: hamid.tizhoosh@uwaterloo.ca.
Jazyk: angličtina
Zdroj: Medical image analysis [Med Image Anal] 2021 May; Vol. 70, pp. 102032. Date of Electronic Publication: 2021 Mar 10.
DOI: 10.1016/j.media.2021.102032
Abstrakt: Feature vectors provided by pre-trained deep artificial neural networks have become a dominant source for image representation in recent literature. Their contribution to the performance of image analysis can be improved through fine-tuning. As an ultimate solution, one might even train a deep network from scratch with the domain-relevant images, a highly desirable option which is generally impeded in pathology by lack of labeled images and the computational expense. In this study, we propose a new network, namely KimiaNet, that employs the topology of the DenseNet with four dense blocks, fine-tuned and trained with histopathology images in different configurations. We used more than 240,000 image patches with 1000×1000 pixels acquired at 20× magnification through our proposed "high-cellularity mosaic" approach to enable the usage of weak labels of 7126 whole slide images of formalin-fixed paraffin-embedded human pathology samples publicly available through The Cancer Genome Atlas (TCGA) repository. We tested KimiaNet using three public datasets, namely TCGA, endometrial cancer images, and colorectal cancer images by evaluating the performance of search and classification when corresponding features of different networks are used for image representation. As well, we designed and trained multiple convolutional batch-normalized ReLU (CBR) networks. The results show that KimiaNet provides superior results compared to the original DenseNet and smaller CBR networks when used as feature extractor to represent histopathology images.
Competing Interests: Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
(Copyright © 2021 The Author(s). Published by Elsevier B.V. All rights reserved.)
Databáze: MEDLINE
Externí odkaz: