Image data registration for dynamic perfusion CT
AssigneeKoninklijke Philips N.V.
InventorsMichael Grass, Alfonso Agatino Isola
Life of this patent
Estimating the term…
Abstract
A method of registering a 4D contrast enhanced image data set, wherein the 4D contrast enhanced image data set includes image data of the same volume of interest acquired at different timeframes with changing contrast enhancement, the volume of interest includes moving structure, and the different timeframes correspond to a predetermined motion phase of interest in different motion cycles of the moving structure, the method, comprising: registering image data corresponding to a plurality of the different timeframes with reference image from one of the timeframes.
Claims19
A method of registering a 4D contrast enhanced image data set of a perfusion scan with enhancement by a contrast agent, wherein the 4D contrast enhanced image data set includes image data of the same volume of interest acquired at different timeframes with changing contrast enhancement, the volume of interest includes moving structure, and the different timeframes correspond to a predetermined motion phase of interest in different motion cycles of the moving structure, the method, comprising:
- registering image data corresponding to a plurality of the different timeframes with reference image from one of the timeframes, wherein the image data is registered using an iterative gradient based registration algorithm and a similarity measure, wherein the registration algorithm is based on stochastic gradient descent optimization algorithm, and wherein the similarity measure includes a zero mean normalized cross-correlation similarity measure.
A system, comprising:
- a warping component that warps image data of a 4D image data set to register the image data with reference image data from the 4D image data set based on a stochastic gradient descent optimization algorithm with an adaptive step size prediction in combination with a zero mean normalized cross-correlation similarity measure, wherein the image data and the reference image data correspond to a same volume of interest and different time frames
- a similarity determiner that determines a value indicative of a similarity between the warped image data and the reference image data.
A method of registering a 4D image data set of a perfusion scan, wherein the 4D image data set includes image data of the same volume of interest acquired at different timeframes with changing contrast enhancement by a contrast agent, the volume of interest includes a moving structure, and the different timeframes correspond to a predetermined motion phase of interest in different motion cycles of the moving structure, the method, comprising:
- warping image data of a selected timeframe to fit reference image data;
- determining a measure of similarity between the warped image data and the reference image data and the measure of similarity includes a zero mean normalized cross-correlation;
- and
- registering the image data to the reference image data using an iterative gradient based registration algorithm and the similarity measure when the measure of similarity satisfies a predetermined criteria and warping the warped image data when the measure of similarity does not satisfy the predetermined criteria.
Dependencies are inferred from the claim text.