University of Oulu

Petäinen L, Väyrynen JP, Ruusuvuori P, Pölönen I, Äyrämö S, Kuopio T (2023) Domain-specific transfer learning in the automated scoring of tumor-stroma ratio from histopathological images of colorectal cancer. PLoS ONE 18(5): e0286270.

Domain-specific transfer learning in the automated scoring of tumor-stroma ratio from histopathological images of colorectal cancer

Saved in:
Author: Petäinen, Liisa1; Väyrynen, Juha P.2; Ruusuvuori, Pekka3,4,5;
Organizations: 1Faculty of Information Technology, University of Jyväskylä, Jyväskylä, Finland
2Cancer and Translational Medicine Research Unit, Medical Research Center, Oulu University Hospital, and University of Oulu, Oulu, Finland
3Faculty of Medicine and Health Technology, Tampere University, Tampere, Finland
4Cancer Research Unit, Institute of Biomedicine, University of Turku, Turku, Finland
5FICAN West Cancer Centre, Turku University Hospital, Turku, Finland
6Department of Education and Research, Hospital Nova of Central Finland, Jyväskylä, Finland
7Department of Biological and Environmental Science, University of Jyväskylä, Jyväskylä, Finland
8Department of Pathology, Hospital Nova of Central Finland, Jyväskylä, Finland
Format: article
Version: published version
Access: open
Online Access: PDF Full Text (PDF, 1.7 MB)
Persistent link:
Language: English
Published: Public Library of Science, 2023
Publish Date: 2023-08-24


Tumor-stroma ratio (TSR) is a prognostic factor for many types of solid tumors. In this study, we propose a method for automated estimation of TSR from histopathological images of colorectal cancer. The method is based on convolutional neural networks which were trained to classify colorectal cancer tissue in hematoxylin-eosin stained samples into three classes: stroma, tumor and other. The models were trained using a data set that consists of 1343 whole slide images. Three different training setups were applied with a transfer learning approach using domain-specific data i.e. an external colorectal cancer histopathological data set. The three most accurate models were chosen as a classifier, TSR values were predicted and the results were compared to a visual TSR estimation made by a pathologist. The results suggest that classification accuracy does not improve when domain-specific data are used in the pre-training of the convolutional neural network models in the task at hand. Classification accuracy for stroma, tumor and other reached 96.1% on an independent test set. Among the three classes the best model gained the highest accuracy (99.3%) for class tumor. When TSR was predicted with the best model, the correlation between the predicted values and values estimated by an experienced pathologist was 0.57. Further research is needed to study associations between computationally predicted TSR values and other clinicopathological factors of colorectal cancer and the overall survival of the patients.

see all

Series: PLoS one
ISSN: 1932-6203
ISSN-E: 1932-6203
ISSN-L: 1932-6203
Volume: 18
Issue: 5
Article number: e0286270
DOI: 10.1371/journal.pone.0286270
Type of Publication: A1 Journal article – refereed
Field of Science: 3122 Cancers
Funding: This study is one part of AI Hub Central Finland project that has received funding from the Council of Tampere Region ( (Decision number: A75000) and Leverage from the EU 2014–2020, funded by European Regional Development Fund (ERDF) ( The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Copyright information: © 2023 Petäinen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.