Multifocal and all-in-focus imaging in optical projection tomography microscopy
Cederlöf, Antti (2021)
2021
Bioteknologian ja biolääketieteen tekniikan kandidaattiohjelma - Bachelor's Programme in Biotechnology and Biomedical Engineering
Lääketieteen ja terveysteknologian tiedekunta - Faculty of Medicine and Health Technology
This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Hyväksymispäivämäärä
2021-07-28
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202108176596
https://urn.fi/URN:NBN:fi:tuni-202108176596
Tiivistelmä
Optical microscopy produces images with visible light as a source. The properties of light give rise to a limited depth-of-feld in optical imaging. To detect objects in different distances from the imaging system, the in-focus area called focal plane of the system should be aligned according to the imaged object. In medical imaging, however, the objects are usually three-dimensional and a single image slice with a thin focal plane does not give enough information.
Optical projection tomography (OPT) is a tomographic imaging technique used for 3D optical imaging of mesoscopic samples. However, limited depth-of-feld is also an issue in OPT as the projection images have limited depth information. In order to extend the depth-of-feld in optical imaging, multifocal imaging algorithms have been developed. They function by combining the in-focus information of multiple optical images with different focal planes to form an all-in-focus image. Multifocal images have been used independently to acquire three-dimensional information of a specimen to a 2D image, and as projection images in OPT to achieve extended depth-of-feld.
There are several all-in-focus algorithms that have different means of detecting the in-focus pixels in the image slices. The algorithms that have been used in OPT are from a simpler end. The more advanced ones have been evaluated in 2D optical imaging with multicellular samples. In this thesis, the more advanced wavelet transform and model based multifocal algorithms are evaluated with single cell-sized bead samples. The resulting all-in-focus images are evaluated while considering whether they would make a good projection image for OPT reconstruction.
Visually evaluating, the all-in-focus images seem to be ft for projections in OPT. It seems that the model based algorithm would probably give the best projection images, while the other ones also look like viable options. In order to fnd out how they would perform, the algorithms should be evaluated in OPT.
Optical projection tomography (OPT) is a tomographic imaging technique used for 3D optical imaging of mesoscopic samples. However, limited depth-of-feld is also an issue in OPT as the projection images have limited depth information. In order to extend the depth-of-feld in optical imaging, multifocal imaging algorithms have been developed. They function by combining the in-focus information of multiple optical images with different focal planes to form an all-in-focus image. Multifocal images have been used independently to acquire three-dimensional information of a specimen to a 2D image, and as projection images in OPT to achieve extended depth-of-feld.
There are several all-in-focus algorithms that have different means of detecting the in-focus pixels in the image slices. The algorithms that have been used in OPT are from a simpler end. The more advanced ones have been evaluated in 2D optical imaging with multicellular samples. In this thesis, the more advanced wavelet transform and model based multifocal algorithms are evaluated with single cell-sized bead samples. The resulting all-in-focus images are evaluated while considering whether they would make a good projection image for OPT reconstruction.
Visually evaluating, the all-in-focus images seem to be ft for projections in OPT. It seems that the model based algorithm would probably give the best projection images, while the other ones also look like viable options. In order to fnd out how they would perform, the algorithms should be evaluated in OPT.
Kokoelmat
- Kandidaatintutkielmat [8683]