Automatic segmentation of carotid arteries using Independent Component Analysis for quantitative brain PET

Vasbinder, Laura (2023) Automatic segmentation of carotid arteries using Independent Component Analysis for quantitative brain PET. Bachelor's Thesis, Biomedical Engineering.

Preview	Text bBME_2023_VasbinderLAW.pdf Download (3MB) | Preview
	Text toestemming.pdf Restricted to Registered users only Download (505kB)

Abstract

Accurate estimation of a plasma input function is crucial for quantitative brain PET. The gold standard to obtain the input function is arterial blood sampling, but this procedure is complex and invasive. Therefore it is of clinical interest to develop a noninvasive alternative. In this study, we investigated the feasibility of obtaining accurate image-derived input functions (IDIF) through automatic segmentation of the carotids using independent component analysis (ICA) and application of partial volume correction, as an alternative input function for quantitative brain FDG PET. Methods: Dynamic FDG PET data of 9 patients was used. The carotids were segmented using ICA, and the resulting time activity curves (TAC) were corrected for the partial volume effect (PVE). The corrected TACs were then used as input functions for kinetic modeling. The obtained model parameters were compared to reference values obtained with an IDIF extracted from the ascending aorta. Results: The shape and temporal characteristics of the corrected TACs resembled those of the reference. The results obtained by kinetic modeling showed a good agreement between the estimated Ki values obtained by using the corrected TAC and the reference TAC as input functions. Conclusion: The obtained results suggest that an IDIF obtained from automatic segmentation of the carotids by ICA and corrected for PVE can be used as an accurate alternative input function for Patlak analysis, but not for compartment modelling.

Item Type:	Thesis (Bachelor's Thesis)
Supervisor name:	Tsoumpas, C.
Degree programme:	Biomedical Engineering
Thesis type:	Bachelor's Thesis
Language:	English
Date Deposited:	05 Jul 2023 13:29
Last Modified:	05 Jul 2023 13:29
URI:	https://fse.studenttheses.ub.rug.nl/id/eprint/30161

Actions (login required)

View Item