Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint
Kloster M, Burfeid-Castellanos AM, Langenkämper D, Nattkemper TW, Beszteri B (2023)
PLOS ONE 18(2): e0272103.
Zeitschriftenaufsatz
| Veröffentlicht | Englisch
Download
![Open access file OA](https://pub.uni-bielefeld.de/images/access_open.png)
Autor*in
Kloster, Michael;
Burfeid-Castellanos, Andrea M.;
Langenkämper, DanielUniBi
;
Nattkemper, Tim WilhelmUniBi
;
Beszteri, Bánk
![](https://pub.uni-bielefeld.de/images/icon_orcid.png)
![](https://pub.uni-bielefeld.de/images/icon_orcid.png)
Einrichtung
Abstract / Bemerkung
Diatoms represent one of the morphologically and taxonomically most diverse groups of microscopic eukaryotes. Light microscopy-based taxonomic identification and enumeration of frustules, the silica shells of these microalgae, is broadly used in aquatic ecology and biomonitoring. One key step in emerging digital variants of such investigations is segmentation, a task that has been addressed before, but usually in manually captured megapixel-sized images of individual diatom cells with a mostly clean background. In this paper, we applied deep learning-based segmentation methods to gigapixel-sized, high-resolution scans of diatom slides with a realistically cluttered background. This setup requires large slide scans to be subdivided into small images (tiles) to apply a segmentation model to them. This subdivision (tiling), when done using a sliding window approach, often leads to cropping relevant objects at the boundaries of individual tiles. We hypothesized that in the case of diatom analysis, reducing the amount of such cropped objects in the training data can improve segmentation performance by allowing for a better discrimination of relevant, intact frustules or valves from small diatom fragments, which are considered irrelevant when counting diatoms. We tested this hypothesis by comparing a standard sliding window / fixed-stride tiling approach with two new approaches we term object-based tile positioning with and without object integrity constraint. With all three tiling approaches, we trained Mask-R-CNN and U-Net models with different amounts of training data and compared their performance. Object-based tiling with object integrity constraint led to an improvement in pixel-based precision by 12–17 percentage points without substantially impairing recall when compared with standard sliding window tiling. We thus propose that training segmentation models with object-based tiling schemes can improve diatom segmentation from large gigapixel-sized images but could potentially also be relevant for other image domains.
Erscheinungsjahr
2023
Zeitschriftentitel
PLOS ONE
Band
18
Ausgabe
2
Art.-Nr.
e0272103
Urheberrecht / Lizenzen
eISSN
1932-6203
Page URI
https://pub.uni-bielefeld.de/record/2969165
Zitieren
Kloster M, Burfeid-Castellanos AM, Langenkämper D, Nattkemper TW, Beszteri B. Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint. PLOS ONE. 2023;18(2): e0272103.
Kloster, M., Burfeid-Castellanos, A. M., Langenkämper, D., Nattkemper, T. W., & Beszteri, B. (2023). Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint. PLOS ONE, 18(2), e0272103. https://doi.org/10.1371/journal.pone.0272103
Kloster, Michael, Burfeid-Castellanos, Andrea M., Langenkämper, Daniel, Nattkemper, Tim Wilhelm, and Beszteri, Bánk. 2023. “Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint”. PLOS ONE 18 (2): e0272103.
Kloster, M., Burfeid-Castellanos, A. M., Langenkämper, D., Nattkemper, T. W., and Beszteri, B. (2023). Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint. PLOS ONE 18:e0272103.
Kloster, M., et al., 2023. Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint. PLOS ONE, 18(2): e0272103.
M. Kloster, et al., “Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint”, PLOS ONE, vol. 18, 2023, : e0272103.
Kloster, M., Burfeid-Castellanos, A.M., Langenkämper, D., Nattkemper, T.W., Beszteri, B.: Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint. PLOS ONE. 18, : e0272103 (2023).
Kloster, Michael, Burfeid-Castellanos, Andrea M., Langenkämper, Daniel, Nattkemper, Tim Wilhelm, and Beszteri, Bánk. “Improving deep learning-based segmentation of diatoms in gigapixel-sized virtual slides by object-based tile positioning and object integrity constraint”. PLOS ONE 18.2 (2023): e0272103.
Alle Dateien verfügbar unter der/den folgenden Lizenz(en):
Creative Commons Namensnennung 4.0 International Public License (CC-BY 4.0):
Volltext(e)
Name
journal.pone.0272103.pdf
3.90 MB
Access Level
![Open access file OA](https://pub.uni-bielefeld.de/images/access_open.png)
Zuletzt Hochgeladen
2023-02-27T08:43:40Z
MD5 Prüfsumme
a2af02c61df7812517caf8ded2246a38
Link(s) zu Volltext(en)
Access Level
![Open access file OA](https://pub.uni-bielefeld.de/images/access_open.png)
Daten bereitgestellt von European Bioinformatics Institute (EBI)
Zitationen in Europe PMC
Daten bereitgestellt von Europe PubMed Central.
References
Daten bereitgestellt von Europe PubMed Central.
Export
Markieren/ Markierung löschen
Markierte Publikationen
Web of Science
Dieser Datensatz im Web of Science®Quellen
PMID: 36827378
PubMed | Europe PMC
Suchen in