Pipeline process for automatically measuring object boundary from ultrasound image samples

• US 5,795,296 A
• Filed: 02/20/1997
• Issued: 08/18/1998
• Est. Priority Date: 03/29/1996
• Status: Expired due to Fees

• Current Assignee: University of Washington
• Original Assignee: University of Washington
• Inventors: Chalana, Vikram, Pathak, Sayan, Kim, Yongmin
• Primary Examiner(s): JAWORSKI, FRANCIS J
• Application Number:
  US08/802,979
• Time in Patent Office: 544 Days
• Field of Search: 128/660.01, 128/660.07, 128/660.09, 128/661.03, 128/661.09, 128/661.1, 128/382
• US Class Current: 600/443
• CPC Class Codes: A61B 5/1075 for measuring dimensions by non-invasive methods, e.g. for determining thickness of tissue layer A61B8/0858 takes precedence
• CPC Class Codes: A61B 8/0816 using echo-encephalography
• CPC Class Codes: A61B 8/0866 involving foetal diagnosis; pre-natal or peri-natal diagnosis of the baby
• CPC Class Codes: G01S 7/52036 using analysis of echo signal for target characterisation
• CPC Class Codes: G01S 7/52061 Plan position indication (PPI display); C-scan display
• CPC Class Codes: G01S 7/52073 Production of cursor lines, markers or indicia by electronic means

Abstract

An ultrasound system automatically measures fetal head size from ultrasound images. An ultrasound image of the fetal head is detected. A radial maxima point is identified on each of a plurality of radii extending from a substantially common vertex point within the fetal head image. Each radial maxima point corresponds to an ultrasound sample along its corresponding radius, and has a maximum ultrasound echo strength. Outlier points are removed and the curve filtered to derive an initial fetal head boundary. An inner fetal head boundary and outer fetal head boundary are derived from the initial fetal head boundary and a predetermined fetal skull thickness, and fetal head size is computed from the inner fetal head boundary and the outer fetal head boundary. Processing is allocated among multiprocessors and performed in pipeline fashion to enable real-time interactive imaging and measuring.

Claims

1. A method for automatically measuring fetal head size based upon ultrasound image data of the fetal head, the method comprising the steps of:

detecting an ultrasound image of the fetal head;

identifying a radial maxima point on each of a plurality of radii extending from a substantially common vertex point within the fetal head image, each radial maxima point corresponding to an ultrasound sample along its corresponding radius and having a maximum ultrasound echo strength;

defining a first curve from the radial maxima points;

modifying the first curve to define an enclosed initial fetal head boundary;

deriving an inner fetal head boundary and outer fetal head boundary from the initial fetal head boundary and a predetermined fetal skull thickness; and

computing fetal head size from the inner fetal head boundary and outer fetal head boundary.

2. The method of claim 1, in which the steps of detecting, identifying, defining, modifying, deriving and computing are repeated for multiple frames of ultrasound image data;

and further comprising the step of;

selecting computed fetal head size from one of the multiple frames as a final fetal head size.

3. An ultrasound system for automatically measuring fetal head size, the ultrasound system comprising:

a transducer array for transmitting ultrasound energy into a patient in the vicinity of an internal fetus and for receiving echoes of the ultrasound energy, the received echoes being transformed into electronic echo signals;

an image processor for generating ultrasound image data from the received echo signals;

a display device for displaying the ultrasound image data generated by the image processor;

means for identifying a radial maxima point on each of a plurality of radii extending from a substantially common vertex point within a fetal head image, each radial maxima point corresponding to an ultrasound sample along its corresponding radius and having a maximum ultrasound echo strength;

means for defining a first curve from the radial maxima points;

means for modifying the first curve to define an initial fetal head boundary;

means for deriving an inner fetal head boundary and outer fetal head boundary from the initial fetal head boundary and a predetermined fetal skull thickness; and

means for computing fetal head size from the inner fetal head boundary and outer fetal head boundary.

4. The system of claim 3, in which the identifying means, defining means, modifying means, deriving means and computing means perform on multiple frames of ultrasound image data;

and wherein the computing means selects a computed fetal head size for one of the multiple frames as a final fetal head size.

5. The system of claim 4, wherein the computing means performs in real time and wherein the display device displays computed fetal head size.

6. A method for automatically detecting a curved boundary within an ultrasound image of an object, the image data being formed by scanning the object with an ultrasound transducer to achieve image data, the detected boundary substantially corresponding to a physical boundary of the object, the method comprising the steps of:

testing the image data to identify a plurality of radial maxima points relative to a substantially common vertex point of the object, each of the plurality of radial maxima points being a point along a radial line extending from the vertex point and corresponding to a maximum ultrasound echo strength;

filtering out radial maxima points farther than a threshold distance from the vertex point, the threshold distance being based upon an average distance between the vertex point and the plurality of radial maxima points, wherein the step of filtering points farther than a radial distance results in an initial curve, corresponding to the physical boundary of the object; and

modifying the initial curve to derive a first boundary, wherein the step of modifying comprises the step of performing an iterative estimation of the initial curve using an active contour model.

7. A method for automatically detecting a curved boundary within an ultrasound image of an object, the image data being formed by scanning the object with an ultrasound transducer to achieve image data, the detected boundary substantially corresponding to a physical boundary of the object, the method comprising the steps of:

at a first multiprocessor, inputting a frame of the image data and data indicative of a point within the curved boundary of the object;

at the first multiprocessor, generating an image force value for the frame of image data for a first direction and a second direction, where the second direction is orthogonal to the first direction;

at a second multiprocessor, testing image data for a prior frame to identify a plurality of radial maxima points relative to a substantially common vertex point of the object, each of the plurality of radial maxima points being a point along a radial line extending from the vertex point and corresponding to a maximum ultrasound echo strength;

at the second multiprocessor, filtering out radial maxima points farther than a threshold distance from the vertex point, the threshold distance being based upon an average distance between the vertex point and the plurality of radial maxima points, wherein the step of filtering points farther than a radial distance results in an initial curve corresponding to the physical boundary of the object; and

at the second multiprocessor, modifying the initial curve to derive a first boundary, wherein the step of modifying the initial curve comprises the step of performing an iterative estimation of the initial curve using an active contour model and image force values for the prior frame of image data derived by the first multiprocessor; and

wherein the steps of inputting and generating at the first multiprocessor are performed by the first multiprocessor during an n-th iteration, and wherein the steps of testing, filtering farther, filtering asymmetric, and modifying are performed by the second multiprocessor during the n-th iteration.

8. The method of claim 7, further comprising the step of:

at the second multiprocessor, either one of expanding or contracting the first boundary and modifying the expanded or contracted first boundary to derive a second boundary, wherein the step of modifying the expanded or contracted first boundary, comprises the step of performing an iterative estimation of the expanded or contracted first boundary using the active contour model and image force values for the prior frame of image data derived by the first multiprocessor; and

wherein the step of modifying the expanded or contracted first boundary is performed by the second multiprocessor during the n-th iteration.

9. The method of claim 8, wherein the object is a fetal head, and further comprising the step of:

during the n-th iteration at the second multiprocessor, computing fetal head size from the first and second boundaries derived from the image data of the (n-1)-th frame.

10. An ultrasound system for automatically detecting a curved boundary within an ultrasound image of an object, the image data being formed by scanning the object with an ultrasound transducer to achieve image data, the detected boundary substantially corresponding to a physical boundary of the object, comprising

a transducer array for transmitting ultrasound energy into a patient in the vicinity of an internal fetus and for receiving echoes of the ultrasound energy, the received echoes being transformed into electronic echo signals;

a sequence of frames of ultrasound image data derived from the received echo signals;

a processing apparatus comprising a first multiprocessor and a second multiprocessor for processing frames of ultrasound image data in pipeline fashion to derive a boundary of the object in the respective frames; and

a display device for displaying an ultrasound image of the object and the derived boundary for frames of the ultrasound image data;

wherein for an n-th frame of image data among the sequence of frames, the n-th frame of image data and data indicative of a point within the curved boundary of the object are input to the first multiprocessor, the first multiprocessor generating an image force value for the n-th frame of image data for each a first direction and a second direction, where the second direction is orthogonal to the first direction,

wherein while the first multiprocessor inputs the n-th frame of image data and generates the image force values, the second multiprocessor performs a sequence of operations, including (i) testing the image data for an (n-1)-th frame to identify a plurality of radial maxima points relative to a substantially common vertex point of the object, each of the plurality of radial maxima points being a point along a radial line extending from the vertex point and corresponding to a maximum ultrasound echo strength;

(ii) filtering out radial maxima points farther than a threshold distance from the vertex point, the threshold distance being based upon an average distance between the vertex point and the plurality of radial maxima points, the filtering resulting in an initial curve corresponding to the physical boundary of the object; and

(iii) modifying the initial curve to derive the boundary, wherein modifying the initial curve includes performing an iterative estimation of the initial curve using an active contour model and image force values for (n-1)-th frame of image data derived by the first multiprocessor.

11. The system of claim 10, wherein while the first multiprocessor inputs the n-th frame of image data and generates the image force values, the sequence of operations performed by the second multiprocessor further includes outputting the (n-1)-th frame of image data and the boundary for the (n-1)-th frame.

12. The system of claim 10, wherein while the first multiprocessor inputs the n-th frame of image data and generates the image force values, the sequence of operations performed by the second multiprocessor further includes (iv) either one of expanding or contracting the first boundary and (v) modifying the expanded or contracted first boundary to derive a second boundary, wherein modifying the expanded or contracted first boundary includes performing an iterative estimation of the expanded or contracted first boundary using the active contour model and image force values for the (n-1)-th frame of image data derived by the first multiprocessor.

13. The system of claim 12, wherein the object is a fetal head, and wherein while the first multiprocessor inputs the n-th frame of image data and generates the image force values, the sequence of operations performed by the second multiprocessor further includes computing fetal head size from the first and second boundaries for the image data of the (n-1)-th frame.

14. An ultrasound system for automatically detecting a curved boundary within an ultrasound image of an object, the system comprising:

an ultrasound transducer which scans the object to achieve image data of the object, the detected boundary substantially corresponding to a physical boundary of the object,

a processor for processing the image data; and

a display for displaying an image of the object and the detected boundary;

the processing by the processor comprising;

(i) testing the image data to identify a plurality of radial maxima points relative to a substantially common vertex point of the object, each of the plurality of radial maxima points being a point along a radial line extending from the vertex point and corresponding to a maximum ultrasound echo strength;

(ii) filtering out radial maxima points farther than a threshold distance from the vertex point, the threshold distance being based upon an average distance between the vertex point and the plurality of radial maxima points, wherein the filtering of filtering points farther than a radial distance results in an initial curve, corresponding to the physical boundary of the object; and

(iii) modifying the initial curve to derive a first boundary, wherein the modifying comprises performing an iterative estimation of the initial curve using an active contour model, wherein the first boundary is the displayed detected boundary.