Classification of colorectal tissue images from high throughput tissue microarrays by ensemble deep learning methods
Open Access
- 27 January 2021
- journal article
- research article
- Published by Springer Science and Business Media LLC in Scientific Reports
- Vol. 11 (1), 1-11
- https://doi.org/10.1038/s41598-021-81352-y
Abstract
Tissue microarray (TMA) core images are a treasure trove for artificial intelligence applications. However, a common problem of TMAs is multiple sectioning, which can change the content of the intended tissue core and requires re-labelling. Here, we investigate different ensemble methods for colorectal tissue classification using high-throughput TMAs. Hematoxylin and Eosin (H&E) core images of 0.6 mm or 1.0 mm diameter from three international cohorts were extracted from 54 digital slides (n = 15,150 cores). After TMA core extraction and color enhancement, five different flows of independent and ensemble deep learning were applied. Training and testing data with 2144 and 13,006 cores included three classes: tumor, normal or “other” tissue. Ground-truth data were collected from 30 ngTMA slides (n = 8689 cores). A test augmentation is applied to reduce the uncertain prediction. Predictive accuracy of the best method, namely Soft Voting Ensemble of one VGG and one CapsNet models was 0.982, 0.947 and 0.939 for normal, “other” and tumor, which outperformed to independent or ensemble learning with one base-estimator. Our high-accuracy algorithm for colorectal tissue classification in high-throughput TMAs is amenable to images from different institutions, core sizes and stain intensity. It helps to reduce error in TMA core evaluations with previously given labels.Funding Information
- Rising Tide, foundation for clinical cancer research (REF-36-361, REF-36-361)
- the Swiss Cancer League (Grant KFS-4427-02-2018)
This publication has 33 references indexed in Scilit:
- Stain Normalization using Sparse AutoEncoders (StaNoSA): Application to digital pathologyComputerized Medical Imaging and Graphics, 2016
- VE1 immunohistochemistry predictsBRAFV600E mutation status and clinical outcome in colorectal cancerOncotarget, 2015
- Global cancer statistics, 2012CA: A Cancer Journal for Clinicians, 2015
- Next-generation tissue microarray (ngTMA) increases the quality of biomarker studies: an example using CD3, CD8, and CD45RO in the tumor microenvironment of six different solid tumor typesJournal of Translational Medicine, 2013
- Regularized Negative Correlation Learning for Neural Network EnsemblesIEEE Transactions on Neural Networks, 2009
- A Decade of Tissue Microarrays: Progress in the Discovery and Validation of Cancer BiomarkersJournal of Clinical Oncology, 2008
- Computational Pathology Analysis of Tissue Microarrays Predicts Survival of Renal Clear Cell Carcinoma PatientsLecture Notes in Computer Science, 2008
- Color transfer between imagesIEEE Computer Graphics and Applications, 2001
- Tissue microarrays for high-throughput molecular profiling of tumor specimensNature Medicine, 1998
- Bagging predictorsMachine Learning, 1996