Multi-dimensional wavelet tomography

• US 5,841,890 A
• Filed: 05/06/1996
• Issued: 11/24/1998
• Est. Priority Date: 05/06/1996
• Status: Expired due to Term

• Current Assignee: MedFlex LLC
• Original Assignee: Northrop Grumman Corporation
• Inventors: Kraske, Wolfgang F.
• Primary Examiner(s): COUSO, YON JUNG
• Application Number:
  US08/643,584
• Time in Patent Office: 932 Days
• Field of Search: 382/131, 382/276, 382/277, 382/260, 364/413.14, 364/413.16, 364/413.19, 364/413.21, 378/21, 378/4
• US Class Current: 382/131
• CPC Class Codes: G06T 11/008 Specific post-processing after tomographic reconstruction, e.g. voxelisation, metal artifact correction
• CPC Class Codes: G06T 2211/424 Iterative

Abstract

Multi-dimensional wavelet tomography is performed by iteratively projecting a multi-dimensional image across a range of orientations so as to produce a plurality of one-dimensional projections which collectively form a projection image thereof; and then performing wavelet decomposition upon the projection image thereof so as to form a plurality of projection subimages each containing particular boundary and angular characteristics of the original image. Such wavelet decomposition facilitates segmentation and reconstruction of selected characteristics of the original image.

Claims

1. A method for performing multi-dimensional wavelet tomography, said method comprising the steps of:

a) iteratively projecting a multi-dimensional original image so as to produce a plurality of one-dimensional projections thereof;

b) performing wavelet decomposition upon the one-dimensional projections to form a plurality of subimages thereof, the step of performing wavelet decomposition comprising;

i) convolving the one-dimensional projections to form a scale filtered subimage and a wavelet filtered image; and

ii) downsampling the scale filtered subimage and the wavelet filtered subimage;

iii) wherein steps (i) and (ii) are iterated to obtain a desired resolution;

c) reconstructing a desired portion of the original image from at least one of said subimages, the step of reconstructing a desired portion of the original image comprising back projection filtering the wavelet decomposed one-dimensional projections, back projection filtering comprising;

i) upsampling a scale filtered subimage and a wavelet filtered image;

ii) convolving the upsampled scale filtered image and wavelet filtered image; and

iii) combining the convolved upsampled scale filtered image and wavelet filtered image.

2. A method for performing multi-dimensional wavelet tomography, said method comprising the steps of:

a) iteratively projecting a multi-dimensional original image so as to produce a plurality of one-dimensional projections thereof;

b) performing wavelet decomposition upon the one-dimensional projections to form a plurality of subimages thereof, the step of performing wavelet decomposition comprising;

i) performing wavelet decomposition upon a two-dimensional array of one-dimensional projections in a first direction to form first scale and wavelet images;

ii) transposing the first scale and wavelet images; and

iii) performing wavelet decomposition upon the transposed first scale and wavelet images to form a horizontal scale vertical scale subimage, a horizontal scale vertical wavelet subimage, a horizontal wavelet vertical scale subimage, and a horizontal wavelet vertical wavelet subimage;

c) wherein such wavelet decomposition facilitates segmentation and reconstruction of a selected portion of the original image.

3. A method for performing multi-dimensional wavelet tomography, said method comprising the steps of:

a) iteratively projecting a multi-dimensional original image so as to produce a plurality of one-dimensional projections thereof;

b) performing wavelet decomposition upon the one-dimensional projections to form a plurality of subimages thereof;

c) reconstructing a desired portion of the original image from at least one of said subimages, the step of reconstructing a desired portion of the original image comprising;

i) transposing a horizontal scale vertical scale subimage, horizontal scale vertical wavelet subimage, horizontal wavelet vertical scale subimage, and a horizontal wavelet vertical wavelet subimage;

ii) performing one-dimensional recomposition upon the transposed horizontal scale vertical scale subimage and the transposed horizontal scale vertical scale subimage and the transposed horizontal scale vertical wavelet subimage to form a first recomposed image and performing one-dimensional recomposition upon the transposed horizontal wavelet vertical scale subimage and the transposed horizontal wavelet vertical wavelet subimage to form a second recomposed image;

iii) transposing the first and second recomposed images, to form a horizontal scale image and a horizontal wavelet image, respectively; and

iv) performing one-dimensional recomposition upon the horizontal scale image and the horizontal wavelet image to form a reconstructed image.