Axel Dipl Ing Wirth

Method for providing a magnetic resonance image from respiration-gated image data

A method and a device for the composition of an MR image with respiration-controlled recording of the spin echoes. The phase coding gradient of an MR equipment is stepped as a function of a motion phase of the respiratory motion once per respiration cycle. With a constant phase coding gradient several scans, possibly EKG-triggered, are then recorded within the respiration cycle. The spin echo signals recorded for each scan are stored in a first memory. In a second memory, the amplitude of the respiratory motion at the respective scan moment is stored. Then a comparison of all stored amplitude of the respiratory motion with a reference value R is made. For the composition of the MR image only that scan per respiration cycle is used for which the respective amplitude of the respiratory motion fulfills a given criterion with respect to the reference value R.

Magnetic resonance equipment for the production of images of a test object

The disclosure relates to nuclear spin tomography equipment in which means (22, 24) are provided for the control of the acquisition of the Fourier rows of the measurement matrix as a function of a respiration signal. The equipment includes a Fourier row counter (25) which influences a circuit (26) that controls the measurement signal acquisition. In a first embodiment of the invention the measurement signal acquisition is always released for all rows outside a predetermined region, but is a function of the respiration phase signal inside this region. The predetermined row region may comprise approximately 10% of all rows to be measured and should lie symmetrically about the center of the matrix. A further reduction of the measuring time is realized, according to another embodiment of the invention, by dispensing altogether with measurement in the row edge region of the matrix.

SELECTION OF CHARACTERISTIC PATTERNS IN MAGNETIC AND ELECTRIC ENCEPHALOGRAMS USING A NEURAL NETWORK

Selection of simple patterns on the maps visualizing distributions of magnetic and electric fields of human encephalogram has been performed using active classification approach, which modifies Hopfield network with Hebbian learning rule. For magnetic data continuous sequence of classes has been established. Correlated combinations of electric and magnetic maps were observed. The method may have use for detection of features on-line with appropriate parallel hardware.