kth.sePublications
Change search
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf
Machine learning-assisted image analysis and metabarcoding for monitoring of plankton in the seas surrounding Sweden
KTH, School of Engineering Sciences in Chemistry, Biotechnology and Health (CBH), Industrial Biotechnology.
2023 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesisAlternative title
Maskininlärningsbaserad bildanalys och DNA-streckkodning för övervakning av plankton i svenska havsområden (Swedish)
Abstract [sv]

I miljöövervakningen av haven runt Sverige har manuell mikroskopi av plankton länge varit den huvudsakliga tekniken för att övervaka växtplanktonbestånden och algblomningar. Nya tekniker utvärderas, men det är inte känt hur resultaten från de nyare teknikerna relaterar till varandra. Två tekniker som utvärderas av SMHI, flödesmikroskopi och DNA-streckkodning, har inte tidigare jämförts i litteraturen. Båda teknikerna har dock jämförts med traditionell mikroskopi. I det här projektet har provserier för DNA-streckkodning och automatiserad mikroskopi med Imaging FlowCytobot (IFCB) samlats in parallellt under en expedition i Egentliga Östersjön, Öresund, Kattegatt och Skagerrak. En bildklassificerare konstruerades med ett konvolutionellt neuronnät, som användes för att klassificera bilderna som tagits med IFCB:n. Resultaten från IFCB:n jämfördes med dem från DNA-streckkodning av 18S rRNA-genen. Jämförelsen visade stark korrelation mellan klassificeringen av bilder och DNA-streckkodning för vissa kiselalger (R>0.8), men teknikernas resultat skilde sig också åt i många fall. Skillnaderna kan studeras för att hitta svagheter i de båda teknikerna och utveckla dem vidare.

Abstract [en]

In environmental monitoring of the seas around Sweden, manual counting with microscopy is used to monitor the plankton communities and algal blooms. New techniques are currently being evaluated, including imaging flow cytometry and DNA metabarcoding, but it is not known how results from the different techniques relate to one another. Previous work has not compared imaging flow cytometry with metabarcoding, although both methods have been compared to traditional microscopy. In this project, samples for DNA metabarcoding and imaging flow cytometry with the Imaging FlowCytobot (IFCB) have been collected in parallel in the Baltic Proper, Öresund, Kattegat, and Skagerrak. To be able to process the large number of images from cytometry, an image classification algorithm based on convolutional neural networks and transfer learning was developed, which was used to classify the images collected. The results were compared to those obtained with 18S rRNA metabarcoding of the protist community. This new approach of comparing imaging flow cytometry with metabarcoding resulted in a strong (R>0.8) correlation for some diatom taxa, but discrepancies between the technologies were also observed. The discrepancies can be further studied to identify weaknesses in both techniques and refine them further.

Place, publisher, year, edition, pages
2023.
Series
TRITA-CBH-GRU ; 2023:198
Keywords [en]
Imaging FlowCytobot, imaging-in-flow cytometry, plankton imaging, convolutional neural network, metabarcoding
Keywords [sv]
Imaging FlowCytobot, flödesmikroskopi, övervakning av plankton, konvolutionella neuronnät, DNA-streckkodning
National Category
Other Environmental Biotechnology
Identifiers
URN: urn:nbn:se:kth:diva-329969OAI: oai:DiVA.org:kth-329969DiVA, id: diva2:1774747
Subject / course
Biotechnology
Educational program
Master of Science - Industrial and Environmental Biotechnology
Supervisors
Examiners
Available from: 2023-06-26 Created: 2023-06-26

Open Access in DiVA

fulltext(4953 kB)323 downloads
File information
File name FULLTEXT01.pdfFile size 4953 kBChecksum SHA-512
47fd900174f61b86cd38b4feb72bcb943b5f3f6f49a098a3df585f8d09f4839e62f7a1614ddba34f115a0117fef24118e8d25e3d4b04a844df408369f29b7435
Type fulltextMimetype application/pdf

By organisation
Industrial Biotechnology
Other Environmental Biotechnology

Search outside of DiVA

GoogleGoogle Scholar
Total: 324 downloads
The number of downloads is the sum of all downloads of full texts. It may include eg previous versions that are now no longer available

urn-nbn

Altmetric score

urn-nbn
Total: 843 hits
CiteExportLink to record
Permanent link

Direct link
Cite
Citation style
  • apa
  • ieee
  • modern-language-association-8th-edition
  • vancouver
  • Other style
More styles
Language
  • de-DE
  • en-GB
  • en-US
  • fi-FI
  • nn-NO
  • nn-NB
  • sv-SE
  • Other locale
More languages
Output format
  • html
  • text
  • asciidoc
  • rtf