Alignment of 3-dimensional cardiac structures in O-15-labeled water PET emission images with mutual information.

BACKGROUND: The aim of this study was to develop a method to correct the heart position between two oxygen 15-labeled water cardiac positron emission tomography (PET) image sets to be able to use the equivalent regions of interest for the quantification of the perfusion values in the same myocardial segments. METHODS AND RESULTS: Independent component analysis was applied to the dynamic image sets (simulated phantom and 6 rest-pharmacologic stress and 10 rest-rest image sets of healthy female volunteers) acquired at different time points to separate the cardiac structures (ventricles and myocardium). The separated component images from independent component analysis from the 2 studies of the same individual were aligned with a normalized mutual information-based registration method. The alignment parameters were applied to position the regions of interest in the floating image sets for calculation of the myocardial blood flow values. In the rest case the mean myocardial blood flow value was 0.76 +/- 0.12 mL x g(-1) x min(-1) for the manual method and 0.79 +/- 0.10 mL x g(-1) x min(-1) for the proposed method (by use of the right ventricle component in the alignment), and in the stress case these values were 3.39 +/- 0.70 mL x g(-1) x min(-1) and 4.01 +/- 0.71 mL x g(-1) x min(-1), respectively. No statistically significant difference was found between the methods. CONCLUSION: In the tests with the phantom and patient images the alignment of cardiac structures was shown to be successful. The alignment could be done without the use of information from the myocardial compartment.

ICA based automatic segmentation of dynamic H(2)(15)O cardiac PET images.

In this study, we applied an iterative independent component analysis (ICA) method for the separation of cardiac tissue components (myocardium, right, and left ventricle) from dynamic positron emission tomography (PET) images. Previous phantom and animal studies have shown that ICA separation extracts the cardiac structures accurately. Our goal in this study was to investigate the methodology with human studies. The ICA separated cardiac structures were used to calculate the myocardial perfusion in two different cases: 1) the regions of interest were drawn manually on the ICA separated component images and 2) the volumes of interest (VOI) were automatically segmented from the component images. For the whole myocardium, the perfusion values of 25 rest and six drug-induced stress studies obtained with these methods were compared to the values from the manually drawn regions of interest on differential images. The separation of the rest and stress studies using ICA-based methods was successful in all cases. The visualization of the cardiac structures from H (2) (15) O PET studies was improved with the ICA separation. Also, the automatic segmentation of the VOI seemed to be feasible.

Unsupervised segmentation of cardiac PET transmission images for automatic heart volume extraction.

In this study, we propose an automatic method to extract the heart volume from the cardiac positron emission tomography (PET) transmission images. The method combines the automatic 3D segmentation of the transmission image using Markov random fields (MRFs) to surface extraction using deformable models. Deformable models were automatically initialized using the MRFs segmentation result. The extraction of the heart region is needed e.g. in independent component analysis (ICA). The volume of the heart can be used to mask the emission image corresponding to the transmission image, so that only the cardiac region is used for the analysis. The masking restricts the number of independent components and reduces the computation time. In addition, the MRF segmentation result could be used for attenuation correction. The method was tested with 25 patient images. The MRF segmentation results were of good quality in all cases and we were able to extract the heart volume from all the images.