Automatic image segmentation methods and analysis

Patent metadata

Field	Value
Publication number	US-2025117949-A1
Application number	US-202418984863-A
Country	US
Kind code	A1
Filing date	Dec 17, 2024
Priority date	Nov 23, 2007
Publication date	Apr 10, 2025
Grant date	—

Abstract

The invention provides methods and apparatus for image processing that perform image segmentation on data sets in two-and/or three-dimensions so as to resolve structures that have the same or similar grey values (and that would otherwise render with the same or similar intensity values) and that, thereby, facilitate visualization and processing of those data sets.

First claim

1 . A method comprising:
(a) selecting a 2D (two dimensional) image slice from a plurality of first 2D image slices spanning a 3D (three dimensional) volume, using a digital data processing system configured for image processing; (b) computing a histogram of the 2D image slice using the digital data processing system; (c) assigning using the digital data processing system the 2D image slice to a body region of the 3D volume based at least in part on a comparison of the histogram with an anatomic atlas of average histograms, where the body region is selected from the group consisting of a lung, a hip, an abdomen, a lower abdomen, a pelvis, an upper leg, a knee, a lower leg, and a foot; (d) identifying at least one geometric characteristic based on shape and/or size using the digital data processing system, where the at least one geometric characteristic is consistent with the body region assigned in step (c); (e) performing a threshold segmentation of the 2D image slice using the digital data processing system; (f) identifying one or both a first structure of interest and a first connected component based on the at least one geometric characteristic using the digital data processing system; (g) labeling using the digital data processing system a first pixel as a first seed point belonging to the first structure of interest and/or the first connected component based on one or more of the first structure of interest, the first connected component and the at least one geometric characteristic; (h) flood-filling a plurality of pixels using the digital data processing system from the first seed point into the first structure of interest and/or the first connected component based on one or more of the threshold segmentation, the first structure of interest and the first connected component; (i) comparing using the digital data processing system the first structure of interest and/or the first connected component with the at least one geometric characteristic; (j) reassigning at least one pixel in the first structure of interest and/or the first connected component using the digital data processing system to a second structure of interest and/or a second connected component based on the comparison in step (i); and (k) displaying using the digital data processing system the 2D image slice with one or both the first structure of interest and/or the first connected component and the second structure of interest and/or the second connected component assigned.

2 . The method of claim 1 , where in step (h) flood-filling is used to grow the first structure of interest and/or the first connected component by gradually increasing an intensity of the first seed point.

3 . The method of claim 1 , where in step (h) flood-filling is used to grow the first structure of interest and/or the first connected component provided a pixel being added to the first structure of interest and/or the first connected component has an intensity value above an intensity of the first seed point.

4 . The method of claim 1 , where the first structure of interest and/or the first connected component is resolved from the second structure of interest and/or the second connected component by reassigning the at least one pixel.

5 . The method of claim 1 , where step (c) further comprises computing a cross correlation between the histogram and the anatomic atlas of average histograms.

6 . The method of claim 1 , further comprising reassigning a pixel overruns a seed point placed in a voxel representing the second structure of interest and/or the second connected component.

7 . The method of claim 1 , where the at least one geometric characteristic is selected from the group consisting of a bone structure and a vessel structure.

8 . The method of claim 1 , where the first structure of interest and/or the first connected component is removed from the 2D image slice displayed.

9 . The method of claim 8 , where the second structure of interest and/or the second connected component is displayed in the 2D image slice.

10 . The method of claim 1 , further comprising generating a first plurality of 2D image slices measured without a contrasting agent and generating a second plurality of 2D image slices measured with the contrasting agent and comparing a first 2D image slice from the first plurality of 2D image slices with a second 2D image slice from the second plurality of 2D image slices.

11 . The method of claim 1 , further comprising computing a second histogram of the first structure of interest and/or the first connected component.

12 . A method comprising:
(a) selecting a 2D (two dimensional) image slice from a plurality of first 2D image slices spanning a 3D (three dimensional) volume, using a digital data processing system configured for image processing; (b) computing a histogram of the 2D image slice using the digital data processing system; (c) assigning using the digital data processing system the 2D image slice to a body region of the 3D volume based at least in part on a comparison of the histogram with an anatomic atlas of average histograms, where the body region is selected from the group consisting of a lung, a hip, an abdomen, a lower abdomen, a pelvis, an upper leg, a knee, a lower leg, and a foot; (d) identifying at least one geometric characteristic based on shape and/or size using the digital data processing system, where the at least one geometric characteristic is consistent with the body region assigned in step (c); (e) performing a threshold segmentation of the 2D image slice using the digital data processing system; (f) identifying using the digital data processing system one or both a first structure of interest and a first connected component based on the at least one geometric characteristic; (g) labeling using the digital data processing system a first pixel as a first seed point belonging to the first structure of interest and/or the first connected component based on one or more of the first structure of interest, the first connected component and the at least one geometric characteristic; (h) flood-filling a plurality of pixels using the digital data processing system from the first seed point into the first structure of interest and/or the first connected component based on one or more of the threshold segmentation, the first structure of interest and the first connected component; (i) comparing using the digital data processing system the first structure of interest and/or the first connected component with the at least one geometric characteristic; (j) reassigning at least one pixel in the first structure of interest and/or the first connected component using the digital data processing system to a second structure of interest and/or a second connected component based on the comparison in step (i); (k) removing using the digital data processing system the first structure of interest and/or the first connected component using the digital data processing system; and ( 1 ) displaying using the digital data processing system the 2D image slice with the second structure of interest and/or the second connected component assigned.

13 . The method of claim 12 , where in step (h) flood-filling is used to grow the first structure of interest and/or the first connected component by gradually increasing an intensity of the first seed point.

14 . The method of claim 12 , where in step (h) flood-filling is used to grow the first structure of interest and/or the first connected component provided a pixel being added to the first structure of interest and/or the first connected component has an intensity value above an intensity of the firs

Assignees

• PME IP Pty Ltd

Inventors

• Westerhoff Malte
• Stalling Detlev
• Seebass Martin

Classifications

• G06T2207/20112

  Image segmentation details · CPC title

• G06T2207/10081

  Computed x-ray tomography [CT] · CPC title

• G06T2200/04

  involving 3D image data · CPC title

• G06T7/0012

  Biomedical image inspection · CPC title

• G06V10/50

  by performing operations within image blocks; by using histograms, e.g. histogram of oriented gradients [HoG]; by summing image-intensity values; Projection analysis · CPC title