Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion

The increased workload in pathology laboratories today means automated tools such as artificial intelligence models can be useful, helping pathologists with their tasks. In this paper, we propose a segmentation model (DRU-Net) that can provide a delineation of human non-small cell lung carcinomas an...

Full description

Saved in:
Bibliographic Details
Main Authors: Soroush Oskouei, Marit Valla, André Pedersen, Erik Smistad, Vibeke Grotnes Dale, Maren Høibø, Sissel Gyrid Freim Wahl, Mats Dehli Haugum, Thomas Langø, Maria Paula Ramnefjell, Lars Andreas Akslen, Gabriel Kiss, Hanne Sorger
Format: Article
Language:English
Published: MDPI AG 2025-05-01
Series:Journal of Imaging
Subjects:
Online Access:https://www.mdpi.com/2313-433X/11/5/166
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850126992315252736
author Soroush Oskouei
Marit Valla
André Pedersen
Erik Smistad
Vibeke Grotnes Dale
Maren Høibø
Sissel Gyrid Freim Wahl
Mats Dehli Haugum
Thomas Langø
Maria Paula Ramnefjell
Lars Andreas Akslen
Gabriel Kiss
Hanne Sorger
author_facet Soroush Oskouei
Marit Valla
André Pedersen
Erik Smistad
Vibeke Grotnes Dale
Maren Høibø
Sissel Gyrid Freim Wahl
Mats Dehli Haugum
Thomas Langø
Maria Paula Ramnefjell
Lars Andreas Akslen
Gabriel Kiss
Hanne Sorger
author_sort Soroush Oskouei
collection DOAJ
description The increased workload in pathology laboratories today means automated tools such as artificial intelligence models can be useful, helping pathologists with their tasks. In this paper, we propose a segmentation model (DRU-Net) that can provide a delineation of human non-small cell lung carcinomas and an augmentation method that can improve classification results. The proposed model is a fused combination of truncated pre-trained DenseNet201 and ResNet101V2 as a patch-wise classifier, followed by a lightweight U-Net as a refinement model. Two datasets (Norwegian Lung Cancer Biobank and Haukeland University Lung Cancer cohort) were used to develop the model. The DRU-Net model achieved an average of 0.91 Dice similarity coefficient. The proposed spatial augmentation method (multi-lens distortion) improved the Dice similarity coefficient from 0.88 to 0.91. Our findings show that selecting image patches that specifically include regions of interest leads to better results for the patch-wise classifier compared to other sampling methods. A qualitative analysis by pathology experts showed that the DRU-Net model was generally successful in tumor detection. Results in the test set showed some areas of false-positive and false-negative segmentation in the periphery, particularly in tumors with inflammatory and reactive changes. In summary, the presented DRU-Net model demonstrated the best performance on the segmentation task, and the proposed augmentation technique proved to improve the results.
format Article
id doaj-art-bca9d301753043cd818a7cce8f91c24a
institution OA Journals
issn 2313-433X
language English
publishDate 2025-05-01
publisher MDPI AG
record_format Article
series Journal of Imaging
spelling doaj-art-bca9d301753043cd818a7cce8f91c24a2025-08-20T02:33:47ZengMDPI AGJournal of Imaging2313-433X2025-05-0111516610.3390/jimaging11050166Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens DistortionSoroush Oskouei0Marit Valla1André Pedersen2Erik Smistad3Vibeke Grotnes Dale4Maren Høibø5Sissel Gyrid Freim Wahl6Mats Dehli Haugum7Thomas Langø8Maria Paula Ramnefjell9Lars Andreas Akslen10Gabriel Kiss11Hanne Sorger12Department of Circulation and Medical Imaging, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayDepartment of Clinical and Molecular Medicine, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayDepartment of Clinical and Molecular Medicine, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayDepartment of Circulation and Medical Imaging, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayDepartment of Clinical and Molecular Medicine, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayDepartment of Clinical and Molecular Medicine, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayDepartment of Pathology, St. Olavs Hospital, Trondheim University Hospital, NO-7030 Trondheim, NorwayDepartment of Pathology, St. Olavs Hospital, Trondheim University Hospital, NO-7030 Trondheim, NorwayDepartment of Health Research, SINTEF Digital, NO-7465 Trondheim, NorwayCentre for Cancer Biomarkers CCBIO, Department of Clinical Medicine, University of Bergen, NO-5007 Bergen, NorwayCentre for Cancer Biomarkers CCBIO, Department of Clinical Medicine, University of Bergen, NO-5007 Bergen, NorwayCenter for Innovation, Medical Devices and Technology, Research Department, St. Olavs Hospital, Trondheim University Hospital, NO-7491 Trondheim, NorwayDepartment of Circulation and Medical Imaging, Norwegian University of Science and Technology (NTNU), NO-7491 Trondheim, NorwayThe increased workload in pathology laboratories today means automated tools such as artificial intelligence models can be useful, helping pathologists with their tasks. In this paper, we propose a segmentation model (DRU-Net) that can provide a delineation of human non-small cell lung carcinomas and an augmentation method that can improve classification results. The proposed model is a fused combination of truncated pre-trained DenseNet201 and ResNet101V2 as a patch-wise classifier, followed by a lightweight U-Net as a refinement model. Two datasets (Norwegian Lung Cancer Biobank and Haukeland University Lung Cancer cohort) were used to develop the model. The DRU-Net model achieved an average of 0.91 Dice similarity coefficient. The proposed spatial augmentation method (multi-lens distortion) improved the Dice similarity coefficient from 0.88 to 0.91. Our findings show that selecting image patches that specifically include regions of interest leads to better results for the patch-wise classifier compared to other sampling methods. A qualitative analysis by pathology experts showed that the DRU-Net model was generally successful in tumor detection. Results in the test set showed some areas of false-positive and false-negative segmentation in the periphery, particularly in tumors with inflammatory and reactive changes. In summary, the presented DRU-Net model demonstrated the best performance on the segmentation task, and the proposed augmentation technique proved to improve the results.https://www.mdpi.com/2313-433X/11/5/166lung carcinomadigital pathologytumor segmentationdeep learningdata augmentation
spellingShingle Soroush Oskouei
Marit Valla
André Pedersen
Erik Smistad
Vibeke Grotnes Dale
Maren Høibø
Sissel Gyrid Freim Wahl
Mats Dehli Haugum
Thomas Langø
Maria Paula Ramnefjell
Lars Andreas Akslen
Gabriel Kiss
Hanne Sorger
Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion
Journal of Imaging
lung carcinoma
digital pathology
tumor segmentation
deep learning
data augmentation
title Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion
title_full Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion
title_fullStr Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion
title_full_unstemmed Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion
title_short Segmentation of Non-Small Cell Lung Carcinomas: Introducing DRU-Net and Multi-Lens Distortion
title_sort segmentation of non small cell lung carcinomas introducing dru net and multi lens distortion
topic lung carcinoma
digital pathology
tumor segmentation
deep learning
data augmentation
url https://www.mdpi.com/2313-433X/11/5/166
work_keys_str_mv AT soroushoskouei segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT maritvalla segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT andrepedersen segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT eriksmistad segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT vibekegrotnesdale segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT marenhøibø segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT sisselgyridfreimwahl segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT matsdehlihaugum segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT thomaslangø segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT mariapaularamnefjell segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT larsandreasakslen segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT gabrielkiss segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion
AT hannesorger segmentationofnonsmallcelllungcarcinomasintroducingdrunetandmultilensdistortion