Correction for patient and organ movement in SPECT: application to exercise thallium-201 cardiac imaging.

We describe a technique for correction of artifacts in exercise 201Tl single photon emission computed tomography (SPECT) images arising from abrupt or gradual translational movement of the heart during acquisition. The procedure involves the tracking of the "center of the heart" in serial projection images using an algorithm which we call "diverging squares". Each projection image is then realigned in the x-y plane so that the heart center conforms to the projected position of a fixed point in space. The shifted projections are reconstructed using the normal filtered backprojection algorithm. In validation studies, the motion correction procedure successfully eliminated movement artifacts in a heart phantom. Image quality was also improved in over one-half of 36 exercise thallium patient studies. The corrected images had smoother and more continuous left ventricular walls, greater clarity of the left ventricular cavity, and reduced streak artifacts. Rest injected or redistribution images, however, were often made worse, due to reduced heart to liver activity ratios and poor tracking of the heart center. Analysis of curves of heart position versus projection angle suggests that translation of the heart is common during imaging after exercise, and results from both abrupt patient movements, and a gradual upward shift of the heart. Our motion correction technique appears to represent a promising new approach for elimination of movement artifacts and enhancement of resolution in exercise 201Tl cardiac SPECT images.

Computer-aided measurement of transverse axon sections for morphology studies.

A study has been made of a general-purpose minicomputer image analysis system to provide medical researchers with data needed in morphology studies of myelinated axon sections. The system was designed to provide data equivalent to that produced by manually outlining and quantifying the axon and myelin boundaries. The system hardware and algorithms are described and experimental results presented comparing manual and automated data. The results show that a system with limited automation is an order of magnitude faster than manual methods in providing data with comparable accuracy and improved repeatability.

Noncompartmental and compartmental modeling of the kinetics of carbon-11 labeled pyrilamine in the human brain.

The kinetic pattern of a 11C-labeled histamine H1 receptor antagonist, [11C]pyrilamine, was investigated in the human brain by factor analysis of dynamic PET studies. Tissue time activity curves were also processed by compartment model curve fitting preceded by deconvolution analysis. Factor analysis revealed two statistically significant and physiologically meaningful kinetic patterns: one for specific and another for nonspecific binding of the radioligand. From these two factors a compartment model containing two tissue compartments (one for specific binding and another for nonspecific binding and free ligand) was constructed. The two-compartment model was also supported by the impulse response function, which was obtained by deconvolution and showed two components. The factor image constructed from factor two demonstrated a distribution pattern characteristic for brain regions rich (frontal, parietal, and temporal lobes) or poor (occipital lobe and cerebellum) in H1 receptors. Blockade of H1 receptors with unlabeled pyrilamine, diphenhydramine, or hydroxyzine caused a significant reduction of this factor. Blockade produced no significant changes in factor one representing nonspecific binding. We conclude that the kinetics of [11C]pyrilamine in the brain can be described by two tissue compartments, one related to the distribution of the H1 receptors. Factor analysis of dynamic studies can be used to locally separate these two compartments, for identification of regions rich and poor in H1 receptors and for noninvasive quantitative investigation of the effects of H1 receptor blockers such as pyrilamine, diphenhydramine, or hydroxyzine.