Method of reconstructing tomographic images

• US 6,788,758 B2
• Filed: 07/16/2003
• Issued: 09/07/2004
• Est. Priority Date: 01/17/2001
• Status: Expired due to Fees

• Current Assignee: African Medical Imaging Limited
• Original Assignee: African Medical Imaging Limited
• Inventors: De Villiers, Mattieu Stefan
• Primary Examiner(s): BRUCE, DAVID VERNON
• Application Number:
  US10/466,496
• Publication Number:
  US 20040066910A1
• Time in Patent Office: 419 Days
• Field of Search: 378/4, 378/8, 378/15, 378/19, 378/901
• US Class Current: 378/4
• CPC Class Codes: G06T 11/005 Specific pre-processing for tomographic reconstruction, e.g. calibration, source positioning, rebinning, scatter correction, retrospective gating
• CPC Class Codes: G06T 2211/424 Iterative
• CPC Class Codes: G06T 2211/436 Limited angle
• CPC Class Codes: Y10S 378/901 Computer tomography program or processor

Abstract

A method of reconstructing a tomographic image of an object from incomplete projection data using a limited angle tomography technique. The method includes using the protection data to obtain a first reconstruction of the image. Thereafter, regions in the first reconstruction that can be predicted with an acceptable degree of certainty in a final reconstruction, are identified. Prior knowledge in the form of possible density levels and piece-wise smoothness that regions in an image can assume, are applied to those regions of acceptable certainty in the first reconstruction so as to obtain a second reconstruction. The projection data is then applied to projections of the second reconstruction using a constraining method to obtain a third reconstruction. In the same way regions of acceptable certainty are identified and the prior knowledge and the projection data are applied to the third reconstruction and subsequent reconstructions until a final reconstruction is achieved.

Claims

1. A method of reconstructing a tomographic image of an object from incomplete projection data using a limited angle tomography technique, the method including the steps of:

using the projection data to obtain a first reconstruction of the image;

creating an intermediate image in which each pixel represents the probability that a corresponding pixel of the first reconstruction of the image resembles the corresponding pixel of a final reconstruction of the image;

based on the intermediate image, identifying regions in the first reconstruction of the image that resemble their final reconstruction;

applying prior knowledge only to said regions in the first reconstruction of the image that resemble their final reconstruction, to obtain a second reconstruction of the image which is more accurate than the first reconstruction;

applying the projection data to projections of the second reconstruction of the image by constraining the projections of the second reconstruction of the image to the projection data, to obtain a third reconstruction of the image which is more accurate than the second reconstruction; and

iteratively creating intermediate images, identifying regions in the third reconstruction of the image and subsequent reconstructions of the image, that resemble their final reconstruction and applying the prior knowledge and the projection data in the same way to the third reconstruction of the image and subsequent reconstructions thereof until a final reconstruction is achieved.

2. A method as claimed in claim 1, which includes iteratively applying the projection data to projections of reconstructions of the image by constraining the projections to the projection data using an iterative update equation method wherein the resultant error obtained between calculated projections of each reconstruction and the projection data is repeatedly back-projected onto the reconstructions so that the error decreases in subsequent reconstructions of the image.

3. A method as claimed in claim 1 or claim 2, wherein the prior knowledge is applied to reconstructions of the image by thresholding preselected pixels to the closest intensity value.

4. A computer program product for reconstructing a tomographic image of an object from incomplete projection data using a limited angle tomography technique, the computer program product including program instructions for:

using the projection data to obtain a first reconstruction of the image;

creating an intermediate image in which each pixel represents the probability that a corresponding pixel of the first reconstruction of the image resembles the corresponding pixel of a final reconstruction of the image;

based on the intermediate image, identifying regions in the first reconstruction of the image that resemble their final reconstruction;

applying prior knowledge only to said regions in the first reconstruction of the image that resemble their final reconstruction, to obtain a second reconstruction of the image which is more accurate than the first reconstruction;

applying the projection data to projections of the second reconstruction of the image by constraining the projections of the second reconstruction of the image to the projection data, to obtain a third reconstruction of the image which is more accurate than the second reconstruction; and

iteratively creating intermediate images, identifying regions in the third reconstruction of the image and subsequent reconstructions of the image, that resemble their final reconstruction and applying the prior knowledge and the projection data in the same way to the third reconstruction of the image and subsequent reconstructions thereof until a final reconstruction is achieved.

5. A computer program product as claimed in claim 4, which includes program instructions for iteratively applying the projection data to projections of reconstructions of the image by constraining the projections to the projection data using an iterative update equation method wherein the resultant error obtained between calculated projections of each reconstruction and the projection data is repeatedly back-projected onto the reconstructions so that the error decreases in subsequent reconstructions of the image.

6. A computer program product as claimed in claim 4 or claim 5, which includes program instructions for applying the prior knowledge to reconstructions of the image, thresholding preselected pixels to the closest intensity value.

7. Processing means having software executable thereon for reconstructing a tomographic image of an object from incomplete projection data using a limited angle tomography technique, the software being configured to:

use the projection data to obtain a first reconstruction of the image;

create an intermediate image in which each pixel represents the probability that a corresponding pixel of the first reconstruction of the image resembles the corresponding pixel of a final reconstruction of the image;

based on the intermediate image, identify regions in the first reconstruction of the image that resemble their final reconstruction;

apply prior knowledge only to said regions in the first reconstruction of the image that resemble their final reconstruction, to obtain a second reconstruction of the image which is more accurate than the first reconstruction;

apply the projection data to projections of the second reconstruction of the image by constraining the projections of the second reconstruction of the image to the projection data, to obtain a third reconstruction of the image which is more accurate than the second reconstruction; and

iteratively create intermediate images, identify regions in the third reconstruction of the image and subsequent reconstructions of the image, that resemble their final reconstruction and apply the prior knowledge and the projection data in the same way to the third reconstruction of the image and subsequent reconstructions thereof until a final reconstruction is achieved.

8. Processing means as claimed in claim 7, wherein the software executable thereon is configured to iteratively apply the projection data to projections of reconstructions of the image by constraining the projections to the projection data using an iterative update equation method reconstruction and the projection data is repeatedly back-projected onto the reconstructions so that the error decreases in subsequent reconstructions of the image.

9. Processing means as claimed in claim 7 or claim 8, wherein the software is configured to apply the prior knowledge to reconstructions of the image by thresholding preselected pixels to the closest intensity value.