Analysis of Dynamic PET Data

In dynamic positron emission tomography (PET) an artery sampling is needed for assessment and validation of parameters in kinetic models. The sampling can be uncomfortable and painful for the patient and technically demanding for the personnel performing the sampling. Noninvasive estimation of the artery time activity curve (TAC) is thus very useful, since the sampling then can be avoided. Methods are tested on simulated data which is an approximation to PET data. The results from the simulated data are used to quantify how will the different methods perform and to illustrate their limitations. The methods are then used on real PET data, and the estimated TACs are compared to the sampled artery TAC. Non-negative matrix factorization (NMF) and independent component analysis (ICA) show the best results with correlations around 0.95 with the artery sampled TAC. However, the scaling of the resulting components is lost in the deconvolution so a rescaling scheme is used to get the correct scale in the results. A factor is calculated to solve this scaling problem. The peaks found in the NMF and ICA components are higher than the ones found by the other methods. The ICA and the NMF results are very similar when applied to real PET data. Therefore, the NMF is chosen as the most appropriate method as it is more stable and not as complicated as the ICA.

[1]  Kaare Brandt Petersen,et al.  Flexible and efficient implementations of Bayesian independent component analysis , 2007, Neurocomputing.

[2]  L. K. Hansen,et al.  Defining a local arterial input function for perfusion MRI using independent component analysis , 2004, Magnetic resonance in medicine.

[3]  James C. Bezdek,et al.  Pattern Recognition with Fuzzy Objective Function Algorithms , 1981, Advanced Applications in Pattern Recognition.

[4]  H. Sebastian Seung,et al.  Algorithms for Non-negative Matrix Factorization , 2000, NIPS.

[5]  Kunihiro Chihara,et al.  Omission of serial arterial blood sampling in neuroreceptor imaging with independent component analysis , 2005, NeuroImage.

[6]  Mean Field ICA , 2006 .

[7]  E. Oja,et al.  Independent Component Analysis , 2013 .

[8]  G. Schulthess,et al.  Clinical positron emission tomography (PET) : correlation with morphological cross-sectional imaging , 2000 .

[9]  Y Yonekura,et al.  An introduction to PET and SPECT neuroreceptor quantification models. , 2001, Journal of nuclear medicine : official publication, Society of Nuclear Medicine.

[10]  J. C. Dunn,et al.  A Fuzzy Relative of the ISODATA Process and Its Use in Detecting Compact Well-Separated Clusters , 1973 .

[11]  Lars Kai Hansen,et al.  Experiences with Matlab and VRML in Functional Neuroimaging Visualizations , 1999 .

[12]  Ole Winther,et al.  TAP Gibbs Free Energy, Belief Propagation and Sparsity , 2001, NIPS.

[13]  Daniel D. Lee,et al.  APPLICATION OF NON-NEGATIVE MATRIX FACTORIZATION TO DYNAMIC POSITRON EMISSION TOMOGRAPHY , 2001 .

[14]  A A Lammertsma.  Noninvasive estimation of cerebral blood flow. , 1994, Journal of nuclear medicine : official publication, Society of Nuclear Medicine.

[15]  Claus Svarer,et al.  Cluster analysis in kinetic modelling of the brain: a noninvasive alternative to arterial sampling , 2004, NeuroImage.

[16]  M. Opper,et al.  Adaptive and self-averaging Thouless-Anderson-Palmer mean-field theory for probabilistic modeling. , 2001, Physical review. E, Statistical, nonlinear, and soft matter physics.

[17]  H. Sebastian Seung,et al.  Learning the parts of objects by non-negative matrix factorization , 1999, Nature.

[18]  J. MacQueen.  Some methods for classification and analysis of multivariate observations , 1967 .

[19]  Ayumu Matani,et al.  Extraction of a plasma time-activity curve from dynamic brain PET images based on independent component analysis , 2005, IEEE Transactions on Biomedical Engineering.

[20]  Hiroshi Watabe,et al.  Noninvasive estimation of cerebral blood flow using image-derived carotid input function in H/sub 2//sup 15/O dynamic PET , 2001, 2001 IEEE Nuclear Science Symposium Conference Record (Cat. No.01CH37310).

[21]  Vincent J. Cunningham,et al.  Parametric Imaging of Ligand-Receptor Binding in PET Using a Simplified Reference Region Model , 1997, NeuroImage.

[22]  M. Opper,et al.  Tractable approximations for probabilistic models: the adaptive Thouless-Anderson-Palmer mean field approach. , 2001, Physical review letters.

[23]  M. Alexander,et al.  Principles of Neural Science , 1981 .