# US 7,004,904 B2 · Image enhancement and segmentation of structures in 3D ultrasound images for volume measurements

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 7,004,904 B2
- Document ID: US7004904B2
- Status: Granted
- Kind: B2
- Application number: 10/633,186
- Filed: 2003-07-31
- Published: 2004-02-05
- Granted: 2006-02-28
- Earliest claimed: 2002-08-02
- Assignees: Diagnostic Ultrasound Corporation
- Inventors: Vikram Chalana; Stephen Dudycha; Gerald McMorrow
- Web page: https://patentgrep.com/patent/US7004904B2

## CPC

- A61B 8/00: Diagnosis using ultrasonic, sonic or infrasonic waves
- A61B 5/204: {Determining bladder volume}
- A61B 8/0858: {involving measuring tissue layers, e.g. skin, interfaces}
- A61B 8/0866: {involving foetal diagnosis; pre-natal or peri-natal diagnosis of the baby}
- A61B 8/462: {characterised by constructional features of the display}
- A61B 8/483: {involving the acquisition of a 3D volume of data}
- G06T 7/0012: {Biomedical image inspection}
- G06T 7/11: Region-based segmentation
- G06T 7/12: Edge-based segmentation
- G06T 7/149: involving deformable models, e.g. active contour models
- G06T 7/62: of area, perimeter, diameter or volume
- A61B 2560/0456: Apparatus provided with a docking unit
- G06T 2207/10132: Ultrasound image
- G06T 2207/10136: 3D ultrasound image
- G06T 2207/30004: Biomedical image processing
- G06T 2207/30016: Brain
- G06T 2207/30044: Fetus; Embryo
- G06T 2207/30048: Heart; Cardiac
- A61B: DIAGNOSIS; SURGERY; IDENTIFICATION (usefulness limited to only animals A61D)
- G06T: IMAGE DATA PROCESSING OR GENERATION, IN GENERAL

## Abstract

A segmentation algorithm is optimized to robustly locate and measure the volume of fluid filled or non-fluid filled structures or organs from imaging systems derived from ultrasound, computer assisted tomography, magnetic resonance, and position emission tomography. A clinical specimen is measured with a plurality of 2D scan planes processed by the segmentation algorithm to estimate the 2D based area the 3D based volumes of fluid-filled and non-fluid filled organs or structures.

## Claims

### Claim 1

A method for determining areas of structures from an image, the method comprising:
acquiring a digital image;
enhancing the digital image using non-linear filters;
segmenting the digital image using a spatial gradient algorithm, a initial point positioning algorithm, and an optimal path algorithm that combines the spatial gradient and the initial point positioning algorithms to produce homogeneous regions; and
determining the areas of the homogenous regions.

### Claim 2 (depends on 1)

The method of claim 1, wherein the non-linear filters include a heat filter and a shock filter, the heat filter applied to the digital image followed by application of the shock filter to the digital image.

### Claim 3 (depends on 2)

The method of claim 2, wherein the heat filter is defined by a partial differential equation of an inputted image pixel intensity u expressed in an equation defined as
∂
u
∂
t
=
∂
2
⁢
u
∂
x
2
+
∂
2
⁢
u
∂
y
2
,
where u is the image being processed,
∂
2
⁢
u
∂
x
2
is the second partial derivative of u along the x-axis, and
∂
2
⁢
u
∂
y
2
is the second partial derivative of u along the y-axis.

### Claim 4 (depends on 2)

The method of claim 2, wherein the shock filter is a partial differential equation of an imputed image pixel intensity u expressed in an equation defined as
∂
u
∂
t
=
-
sign
(
∂
2
⁢
u
d
⁢
⁢
x
2
)
⁢
⁢

∂
u
d
⁢
⁢
x

,
where u is the image being processed,
∂
2
⁢
u
∂
x
2
is the second partial derivative of u,

∂
u
d
⁢
⁢
x

is the absolute value of the first derivative of u along the x-axis.

### Claim 5 (depends on 1)

The method of claim 1, wherein the optimum path algorithm includes a total cost function C(p) weighted between an edge distance cost function, a path direction cost function, and a previous contour distance cost function from the equation
C(p)=αCe(p)+βCd(p)+γCc(p),
where Ce(p) is the edge distance cost function, Cd(p) is the path direction cost function, Cc(p) is the previous contour distance cost function, and α, β, and γ are numerical values.

### Claim 6 (depends on 5)

The method of claim 5, wherein the edge distance function Ce(p) is defined from the equation
C
e
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
(
1
μ
+

⁢
∇
I
i

)
,
where Ii is the pixel intensity, ∥∇Ii∥ is the gradient magnitude the pixel at location i, and μ is a constant value.

### Claim 7 (depends on 5)

The method of claim 5, wherein the path direction cost function Cd(p) is defined from the equation
C
d
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
(
I
i
i
⁢
⁢
n
-
I
i
out
)
,
where Iiin is the image intensity of the inside and dark pixel and Iiout is the image intensities the outside and bright pixel along a path bisecting the Iiin and Iiout pixels and connecting pixels points i and i−1 adjacent to each side of the Iiin pixel.

### Claim 8 (depends on 5)

The method of claim 5, wherein the previous contour function Cc(p) is defined by the equation
C
c
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
D
i
P
,
where P is the previous contour and DiP is the distance of point i from the closest point on the previous contour.

### Claim 9

A method for determining volumes of structures from a set of images, the method comprising:
acquiring at least two digital images;
enhancing each digital image using non-linear filters;
segmenting the digital image using a spatial gradient algorithm, a initial point positioning algorithm, and an optimal path algorithm that combines the spatial gradient and the initial point positioning algorithmto produce homogeneous regions;
assembling the digital images into an array;
determining the areas of and the volumes between the homogeneous regions in the array.

### Claim 10 (depends on 9)

The method of claim 9, wherein the array includes a rotational assembly, a wedge assembly, and a translational assembly.

### Claim 11 (depends on 9)

The method of claim 9, wherein the non-linear filters include a heat filter and a shock filter, the heat filter applied to the digital image followed by application of the shock filter to the digital image.

### Claim 12 (depends on 11)

The method of claim 11, wherein the heat filter is defined by a partial differential equation of an inputted image pixel intensity u expressed in an equation defined as:
∂
u
∂
t
=
∂
2
⁢
u
∂
x
2
+
∂
2
⁢
u
∂
y
2
,
where u is the image being processed,
∂
2
⁢
u
∂
x
2
is the second partial derivative of u along the x-axis, and
∂
2
⁢
u
∂
y
2
is the second partial derivative of u along the y-axis.

### Claim 13 (depends on 11)

The method of claim 11, wherein the shock filter is a partial differential equation of an imputed image pixel intensity u expressed in an equation defined as
∂
u
∂
t
=
-
sign
⁡
(
∂
2
⁢
u
dx
2
)
⁢

∂
u
dx

,
where u is the image being processed,
∂
2
⁢
u
∂
x
2
is the second partial derivative of u,

∂
u
dx

is the absolute value of the first derivative of u along the x-axis.

### Claim 14 (depends on 9)

The method of claim 9, wherein the optimum path algorithm includes a total cost function C(p) weighted between an edge distance cost function, a path direction cost function, and a previous contour distance cost function from the equation
C(p)=αCe(p)+βCd(p)+γCc(p),
where Ce(p) is the edge distance cost function, Cd(p) is the path direction cost function, Cc(p) is the previous contour distance cost function, and α, β, and γ are numerical values.

### Claim 15 (depends on 14)

The method of claim 14, wherein the edge distance function Ce(p) is defined from the equation
C
e
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
(
1
μ
+

∇
I
i

)
,
where Ii is the pixel intensity, ∥∇Ii∥ is the gradient magnitude the pixel at location i, and μ is a constant value.

### Claim 16 (depends on 14)

The method of claim 14, wherein the path direction cost function Cd(p) is defined from the equation
C
d
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
(
I
i
in
-
I
i
out
)
,
where Iiin is the image intensity of the inside and dark pixel and Iiout is the image intensities the outside and bright pixel along a path bisecting the Iiin and Iiout pixels and connecting pixels points i and i−1 adjacent to each side of the Iiin pixel.

### Claim 17 (depends on 14)

The method of claim 14, wherein the previous contour function Cc(p) is defined by the equation
C
c
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
D
i
P
,
where P is the previous contour and DiP is the distance of point i from the closest point on the previous contour.

### Claim 18

A method to determine volume of a structure in digital images acquired from electromagnetic and non-electromagnetic sources, the method comprising:
positioning a transceiver exterior to a patient such that at least a portion of the structure is within a field of view of the transceiver, the transceiver, configured to send electromagnetic radiation and to receive echoes of the electromagnetic radiation;
sending the radiation from the transceiver to the structure;
receiving echoes of the radiation reflected from the structure to the transceiver;
associating the received echoes to form a plurality of 2D scanplanes so that they form an array;
enhancing the images of the structure in each plane of the array using non-linear filters; and
determining the structure volume spanning between and through each plane in the array.

### Claim 19 (depends on 18)

The method of claim 18, wherein plurality of 2D scanplanes are assembled into a plurality of arrays including a rotational array, a translational array, or a wedge array.

### Claim 20 (depends on 18)

The method of claim 18, wherein the non-linear filters include a heat filter and a shock filter, the heat filter applied to the digital images followed by application of the shock filter to the digital images.

### Claim 21 (depends on 20)

The method of claim 20, wherein the heat filter is defined by a partial differential equation of an inputted image pixel intensity u expressed in an equation defined as
∂
u
∂
t
=
∂
2
⁢
u
∂
x
2
+
∂
2
⁢
u
∂
y
2
,
where u is the image being processed,
∂
2
⁢
u
∂
x
2
is the second partial derivative of u along the x-axis, and
∂
2
⁢
u
∂
y
2
is the second partial derivative of u along the y-axis.

### Claim 22 (depends on 20)

The method of claim 20, wherein the shock filter is a partial differential equation of an imputed image pixel intensity u expressed in an equation defined as
∂
u
∂
t
=
-
sign
⁡
(
∂
2
⁢
u
dx
2
)
⁢

∂
u
dx

,
where u is the image being processed,
∂
2
⁢
u
∂
x
2
is the second partial derivative of u,

∂
u
dx

is the absolute value of the first derivative of u along the x-axis.

### Claim 23 (depends on 18)

The method of claim 18, wherein the plurality algorithms includes a spatial gradient algorithm, a previous contour algorithm, and. an optimal path algorithm that combines the spatial gradient and the previous contour algorithms to produce the homogeneous regions.

### Claim 24 (depends on 23)

The method of claim 23, wherein the optimum path algorithm includes a total cost function C(p) weighted between an edge distance cost function, a path direction cost function, and a previous contour distance cost function from the equation
C(p)=αCe(p)+βCd(p)+γCc(p),
where Ce(p) is the edge distance cost function, Cd(p) is the path direction cost function, Cc(p) is the previous contour distance cost function, and α, β, and γ are numerical values.

### Claim 25 (depends on 24)

The method of claim 24, wherein the edge distance function Ce(p) is defined from the equation
C
e
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
⁢
(
1
μ
+

∇
I
i

)
,
where Ii is the pixel intensity, ∥∇Ii∥ is the gradient magnitude the pixel at location i, and μ isconstant value.

### Claim 26 (depends on 24)

The method of claim 24, wherein the path direction cost function Cd(p) is defined from the equation
C
d
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
⁢
(
I
i
in
-
I
i
out
)
,
where Iiin is the image intensity of the inside and dark pixel and Iiout is the image intensities the outside and bright pixel along a path bisecting the Iiin and Iiout pixels and connecting pixels points i and i−1 adjacent to each side of the Iiin pixel.

### Claim 27 (depends on 24)

The method of claim 24 wherein the previous contour function Cc(p) is defined by the equation.
C
c
⁡
(
p
)
=
∑
i
=
p1
p2
⁢
D
i
P
,
where P is the previous contour and DiP is the distance of point i from the closest point on the previous contour.