Impact of Annotation Noise on Histopathology Nucleus Segmentation
Deep learning is often used for automated diagnosis support in biomedical image processing scenarios. Annotated datasets are essential for the supervised training of deep neural networks. The problem of consistent and noise-free annotation remains for experts such as pathologists. The variability wi...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
De Gruyter
2022-09-01
|
Series: | Current Directions in Biomedical Engineering |
Subjects: | |
Online Access: | https://doi.org/10.1515/cdbme-2022-1051 |
_version_ | 1811159446144942080 |
---|---|
author | Schilling Marcel P. Ahuja Niket Rettenberger Luca Scherr Tim Reischl Markus |
author_facet | Schilling Marcel P. Ahuja Niket Rettenberger Luca Scherr Tim Reischl Markus |
author_sort | Schilling Marcel P. |
collection | DOAJ |
description | Deep learning is often used for automated diagnosis support in biomedical image processing scenarios. Annotated datasets are essential for the supervised training of deep neural networks. The problem of consistent and noise-free annotation remains for experts such as pathologists. The variability within an annotator (intra) and the variability between annotators (inter) are current challenges. In clinical practice or biology, instance segmentation is a common task, but a comprehensive and quantitative study regarding the impact of noisy annotations lacks. In this paper, we present a concept to categorize and simulate various types of annotation noise as well as an evaluation of the impact on deep learning pipelines. Thereby, we use the multi-organ histology image dataset MoNuSeg to discuss the influence of annotator variability. We provide annotation recommendations for clinicians to achieve high-quality automated diagnostic algorithms. |
first_indexed | 2024-04-10T05:41:15Z |
format | Article |
id | doaj.art-a568d9279ca74186854406b7c3ef2d03 |
institution | Directory Open Access Journal |
issn | 2364-5504 |
language | English |
last_indexed | 2024-04-10T05:41:15Z |
publishDate | 2022-09-01 |
publisher | De Gruyter |
record_format | Article |
series | Current Directions in Biomedical Engineering |
spelling | doaj.art-a568d9279ca74186854406b7c3ef2d032023-03-06T10:24:51ZengDe GruyterCurrent Directions in Biomedical Engineering2364-55042022-09-018219720010.1515/cdbme-2022-1051Impact of Annotation Noise on Histopathology Nucleus SegmentationSchilling Marcel P.0Ahuja NiketRettenberger LucaScherr TimReischl MarkusInstitute for Automation and Applied Informatics, Karlsruhe Institute of Technology, Hermann-von-Helmholtz-Platz 1, 76344Eggenstein-Leopoldshafen, GermanyDeep learning is often used for automated diagnosis support in biomedical image processing scenarios. Annotated datasets are essential for the supervised training of deep neural networks. The problem of consistent and noise-free annotation remains for experts such as pathologists. The variability within an annotator (intra) and the variability between annotators (inter) are current challenges. In clinical practice or biology, instance segmentation is a common task, but a comprehensive and quantitative study regarding the impact of noisy annotations lacks. In this paper, we present a concept to categorize and simulate various types of annotation noise as well as an evaluation of the impact on deep learning pipelines. Thereby, we use the multi-organ histology image dataset MoNuSeg to discuss the influence of annotator variability. We provide annotation recommendations for clinicians to achieve high-quality automated diagnostic algorithms.https://doi.org/10.1515/cdbme-2022-1051instance segmentationannotator variabilitydeep learningimage processing |
spellingShingle | Schilling Marcel P. Ahuja Niket Rettenberger Luca Scherr Tim Reischl Markus Impact of Annotation Noise on Histopathology Nucleus Segmentation Current Directions in Biomedical Engineering instance segmentation annotator variability deep learning image processing |
title | Impact of Annotation Noise on Histopathology Nucleus Segmentation |
title_full | Impact of Annotation Noise on Histopathology Nucleus Segmentation |
title_fullStr | Impact of Annotation Noise on Histopathology Nucleus Segmentation |
title_full_unstemmed | Impact of Annotation Noise on Histopathology Nucleus Segmentation |
title_short | Impact of Annotation Noise on Histopathology Nucleus Segmentation |
title_sort | impact of annotation noise on histopathology nucleus segmentation |
topic | instance segmentation annotator variability deep learning image processing |
url | https://doi.org/10.1515/cdbme-2022-1051 |
work_keys_str_mv | AT schillingmarcelp impactofannotationnoiseonhistopathologynucleussegmentation AT ahujaniket impactofannotationnoiseonhistopathologynucleussegmentation AT rettenbergerluca impactofannotationnoiseonhistopathologynucleussegmentation AT scherrtim impactofannotationnoiseonhistopathologynucleussegmentation AT reischlmarkus impactofannotationnoiseonhistopathologynucleussegmentation |