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
Processing of Optical Coherence Tomography Images: Filtering and Segmentation of Pathological Thyroid Tissue
Linköping University, Department of Biomedical Engineering. Linköping University, Faculty of Medicine and Health Sciences.
2016 (English)Independent thesis Advanced level (degree of Master (Two Years)), 20 credits / 30 HE creditsStudent thesis
Abstract [en]

In the human body, the main function of the healthy thyroid gland is the regulation of the metabolism and hormone production. Included in the thyroid are organized structured and uniformly shaped follicles ranging from 50-500 μm in diameter. Pathologies lead to morphological changes of these follicles, affecting the density and size, but can also lead to an absence. In this study optical coherence tomography (OCT) was used to examine pathological thyroid tissue by extracting structural information of the follicles from image segmentation. However, OCT images usually include a high amount of speckle noise which affects the segmentation outcome. Due to that, the OCT images need to be improved. The aim of this thesis was to investigate the appropriate filtering methods to enhance the images and thus improve the segmentation outcome.

The images of pathological thyroid tissues with a size of 0:5-1 cm where scanned by a spectral domain OCT system (Telesto II, Thorlabs GmbH, Germany) using a center wavelength of 1300nm. The obtained 2D and 3D images were saved as .oct file as well as implemented and visualized in a MATLAB graphical user interface (GUI) for further processing. For image improvement, four filtering enhancement methods were applied to the 2D images such as the enhanced resolution imaging (ERI), adaptive Wiener filter, discrete wavelet transform (DWT) and multi-frame wavelet transform (WT). The processed images were further converted to grayscale and binary images for intensity-based segmentation. The output of all methods were compared and evaluated using signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), enhanced number of looks (ENL), edge profiles and outcome of the segmented images.

It was demonstrated that the complex DWT (cDWT) with a higher threshold and the multi-frame WT using the haar wavelet showed enhanced results over the other filtering methods. The computed SNR could be increased up to 52% and the ENL value up to 4802%, applying the multi-frame WT, while the CNR could be increased up to 106% for cDWT. The lowest obtained gradient was equal to an intensity decrease of -61% and -68% for multi-frame WT and cDWT, respectively. The filtering method could increase the smoothness of the image while the edge sharpness could be kept. The segmentation could detect both small and large follicles. ERI did not show any improvement in the segmentation but could enhance the structural detail of the image. Larger neighbourhoods of the adaptive Wiener filter showed a highly blurred image and led to merged follicles in the image segmentation.

The wavelet filters DWT and multi-frame WT gave most satisfying results since high and low frequencies were divided into subbands, where individual information on vertical, horizontal and diagonal edges was stored. Applied cDWT had an even higher amount of subbands, so that more information on signal and speckle noise could be specified. Due to this fact, it was possible to achieve a decreased noise level while edge sharpness where maintained. Using a multi-frame image an increased SNR was obtained, as the intensity information stayed constant over the individual frames while the noise information changed.

Wavelet based filtering showed higher improved results in comparison to the adaptive Wiener filter or the ERI in the 2D domain. By applying filtering methods in higher dimensions such as 3D or even 4D, better results in noise reduction are expected. Improved settings for the individual filtering methods as well as enhancement in segmentation are part of the future work.

Place, publisher, year, edition, pages
2016. , p. 68
National Category
Medical Biotechnology
Identifiers
URN: urn:nbn:se:liu:diva-161988ISRN: LiTH-IMT/ERASMUS-R-13/43-SEOAI: oai:DiVA.org:liu-161988DiVA, id: diva2:1370435
Subject / course
Erasmus (International Exchange Student Program)
Supervisors
Examiners
Available from: 2019-11-15 Created: 2019-11-15 Last updated: 2019-11-15Bibliographically approved

Open Access in DiVA

Processing of Optical Coherence Tomography Images: Filtering and Segmentation of Pathological Thyroid Tissue(5499 kB)9 downloads
File information
File name FULLTEXT01.pdfFile size 5499 kBChecksum SHA-512
f1319935f44dfd3a4af989a72b7cae6492328120af19fab8c780797045f6e969f3f07d423cfa9947547043ce021276e9a7d96d8de8844991329143066ec76f9a
Type fulltextMimetype application/pdf

By organisation
Department of Biomedical EngineeringFaculty of Medicine and Health Sciences
Medical Biotechnology

Search outside of DiVA

GoogleGoogle Scholar
Total: 9 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: 18 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