Automated Method for Generation of Input Function in PET Studies using MVW-PC Images

Modeling is a method for extracting quantitative values from PET. The signal from the reference region or from liquid blood samples is required as reference. Since blood sampling is unsafe, this report presents a mechanical method dependant on MVW-PCA for utilizing blood data from the images. The research was performed on clinical PET data from many human brains utilizing the tracer PIB. Two veins put together in a MVW-PC and an average of the TACs in the relevant locations was formed. Finally, a correcting function was calculated.The curves generated from the image data were very much like the curves generated from liquid blood samples, with all the largest errors in the early stages of the scan.The used method shows prospects for generating really good results if worked onmore. Among the strengths of the approach this is not restricted to a unique tracer or time protocol, ever since the MVW-PC will be chosen based on the weights for the first 60 seconds.

Contents: Automated Method for Generation of Input Function in PET Studies using MVW-PC Images

1 Introduction
1.1 Setting
1.2 Structure of the thesis
1.3 Aim
2 Background
2.1 Imaging modalities
2.1.1 Anatomical information
2.1.2 Physiological information
2.1.3 Integrated imaging modalities
2.1.4 Differences and applications
2.2 PET
2.2.1 Noise
2.2.2 Reconstruction algorithms
2.3 PCA
2.3.1 Algorithm
2.3.2 Pre-normalizations
2.4 Quantification approaches in PET
2.4.1 Summing or averaging
2.4.2 Kinetic modelling
2.5 Factors that affect quantification
3 Methods
3.1 MVW-PCA
3.2 Synthetic images
3.3 Generation of input function from clinical data
3.3.1 VOI selection
3.3.2 Averaging
3.3.3 Scaling to input function
4 Results
4.1 Synthetic images
4.2 Generation of input function from clinical data
4.2.1 VOI selection
4.2.2 Averaging
4.2.3 Scaling to input function
5 Closing remarks
5.1 Discussion
5.1.1 PCA on PET images
5.1.2 Synthetic images
5.1.3 Generation of input function from clinical data…