Browse Prior Art Database

UNSUPERVISED NOISE ESTIMATION BASED ON DIFFERENCE IMAGES IN COMPUTER TOMOGRAPHY

IP.com Disclosure Number: IPCOM000241190D
Publication Date: 2015-Apr-02
Document File: 7 page(s) / 1M

Publishing Venue

The IP.com Prior Art Database

Abstract

The present invention proposes a technique to estimate noise from images in a computer tomography (CT). The technique includes estimation of range parameter(s) required for advanced image filtering in the CT. The technique segments the pixels of a certain tissue type and calculates estimated noise based on a representative difference image. Examples of difference images, which apply to this framework, include multiple images with different temporal characteristic and/or multiple images with different spectral characteristics. As the technique is based on difference image, anatomy is expected to be similar but noise pattern is expected to change.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 38% of the total text.

UNSUPERVISED NOISE ESTIMATION BASED ON DIFFERENCE IMAGES IN COMPUTER TOMOGRAPHY

FIELD OF INVENTION

The invention generally relates to computer tomography and more particularly to noise estimation based on difference images in computer tomography.

BACKGROUND OF THE INVENTION

Generally, there are variety of image space filtering or regularization procedures, which include both an image domain parameter and a range domain parameter. Both the domain parameters include bilateral filtering, non-local mean filtering and block matched filters. In medical imaging noise in images vary significantly based on patient scan and scan parameters utilized during acquisition. In computer tomography (CT) scan parameters include patient habitus, X-ray spectrum, X-ray filtration, detector response, reconstruction kernel, and exposure time, among others. These techniques differ from traditional linear filtering in which a range-based component is part of the filtering in addition to spatial component.

A conventional technique includes noise mask generation and noise estimation from image data. Additionally the technique includes back projection of the projection noise estimation into image space. However, the conventional technique is computationally intensive due to additional back projection of data.

 It would be desirable to have an efficient technique for measurement of image noise, which is more robust to background structure in the CT.

BRIEF DESCRIPTION OF THE INVENTION

The present invention proposes a technique to estimate noise from images in computer tomography (CT). The technique includes estimation of range parameter(s) required for advanced image filtering in the CT. The technique segments pixels of a certain tissue type and calculates estimated noise based on a representative difference image.

DETAILED DESCRIPTION OF THE INVENTION

The present invention proposes a technique to estimate noise from images in a computer tomography (CT). The technique includes estimation of a range parameter(s) required for advanced image filtering in the CT. The technique segments the pixels of a certain tissue type and calculates estimated noise based on a representative difference image. Examples of difference images which apply to such framework include multiple images with different temporal characteristics and/or multiple images with different spectral characteristics. As the technique is based on difference images, anatomy is expected to be similar. However, noise pattern is expected to change. In computer tomography system, there are several similar scenarios occurring which include multiple images with different temporal responses and multiple images with different spectral responses. For instance when a full-scan CT is acquired, there are multiple techniques to generate images with different temporal responses, such as, full-scan and a half-scan image or two half-scan images reconstructed with different center view angles. In case of dual energy CT, this is differ...