Automated Fetal Measurement From Three-Dimensional Ultrasound Data

• US 20090093717A1
• Filed: 09/29/2008
• Published: 04/09/2009
• Est. Priority Date: 10/04/2007
• Status: Active Grant

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Corporate Research Incorporated (Siemens AG), Siemens Medical Solutions USA Incorporated (Siemens AG)
• Inventors: Carneiro, Gustavo Henrique Monteiro de Barros, Amat, Fernando, Georgescu, Bogdan, Comaniciu, Dorin, Good, Sara
• Granted Patent: US 8,556,814 B2
• Time in Patent Office: Days
• 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 5/7267 involving training the classification device
• 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: A61B 8/13 Tomography A61B8/10, A61B8/12 take precedence; tomography for radiation diagnosis A61B6/02
• CPC Class Codes: A61B 8/466 adapted to display 3D data 3D imaging and stereoscopic displays in devices for short-range imaging with acoustic waves G01S15/8993 and G01S7/52068, respectively; 3D image rendering G06T15/00; manipulating 3D models for computer graphics G06T19/00
• CPC Class Codes: G06T 2207/10136 3D ultrasound image
• CPC Class Codes: G06T 2207/30044 Fetus; Embryo
• CPC Class Codes: G06T 7/11 Region-based segmentation
• CPC Class Codes: G06V 10/768 using context analysis, e.g. recognition aided by known co-occurring patterns
• CPC Class Codes: G06V 2201/03 Recognition of patterns in medical or anatomical images
• CPC Class Codes: G16H 30/40 for processing medical images, e.g. editing
• CPC Class Codes: G16H 50/20 for computer-aided diagnosis, e.g. based on medical expert systems
• CPC Class Codes: G16H 50/30 for calculating health indices; for individual health risk assessment
• CPC Class Codes: G16Z 99/00 Subject matter not provided for in other main groups of this subclass

Abstract

A fetal parameter or anatomy is measured or detected from three-dimensional ultrasound data. An algorithm is machine-trained to detect fetal anatomy. Any machine training approach may be used. The machine-trained classifier is a joint classifier, such that one anatomy is detected using the ultrasound data and the detected location of another anatomy. The machine-trained classifier uses marginal space such that the location of anatomy is detected sequentially through translation, orientation and scale rather than detecting for all location parameters at once. The machine-trained classifier includes detectors for detecting from the ultrasound data at different resolutions, such as in a pyramid volume.

Claims

1. A method for measuring a fetal parameter from three-dimensional ultrasound data, the method comprising:

applying a machine-trained classifier to the three-dimensional ultrasound data;

detecting a first fetal anatomy as a function of the applying;

measuring a value of the fetal parameter associated with the first fetal anatomy; and

displaying the value.

2. The method of claim 1 further comprising displaying a planar reconstruction from the three-dimensional ultrasound data, the planar reconstruction showing the fetal anatomy.

3. The method of claim 1 wherein applying comprises applying the machine-trained classifier as a detector of the fetal cerebellum, fetal cisterna magna, fetal lateral ventricles, or combinations thereof.

4. The method of claim 1 wherein applying the machine-trained classifier comprises applying a joint classifier for a plurality of fetal anatomies, including the first fetal anatomy, the first fetal anatomy being detected as a function of detection of another fetal anatomy.

5. The method of claim 4 wherein applying the joint classifier comprises applying a machine-trained head detector, and then applying a machine trained detector of the first fetal anatomy, the machine-trained detector of the first fetal anatomy dependent on an output of the machine-trained head detector.

6. The method of claim 1 wherein applying the machine-trained classifier comprises applying a sequence of marginal machine-trained classifiers to the three-dimensional ultrasound data, a first of the marginal machine-trained classifiers for translation of the first fetal anatomy within a volume represented by the three-dimensional ultrasound data, a second of the marginal machine-trained classifiers for rotation of the first fetal anatomy within the volume, and a third of the marginal machine-trained classifiers for scale of the first fetal anatomy within the volume.

7. The method of claim 1 wherein applying the machine-trained classifier comprises applying a first detector to first data comprising a coarse sampling of the ultrasound data and applying a second detector to second data comprising a finer sampling of the ultrasound data than the first data, a search space of the second detector being a function of an output of the first detector.

8. The method of claim 1 wherein applying the machine-trained classifier comprises applying a machine-trained joint, marginal space classifier.

9. The method of claim 1 wherein applying the machine-trained classifier comprises applying a probabilistic boosting tree.

10. The method of claim 1 wherein measuring comprises determining a line representing the first fetal anatomy as an output of the machine-trained classifier, determining a voxel size, and calculating the distance as a function of the line and the voxel size.

11. In a computer readable storage medium having stored therein data representing instructions executable by a programmed processor for measuring a fetal parameter from three-dimensional ultrasound data, the storage medium comprising instructions for:

receiving user input requesting automatic measurement of a fetal head anatomy, the fetal parameter being for the fetal head anatomy;

detecting the fetal head anatomy with a computer learnt classifier, the computer learnt classifier operable to detect from the three-dimensional ultrasound data without user indication of a position; and

calculating a value for the fetal parameter.

12. The computer readable storage medium of claim 11 wherein detecting comprises detecting with the computer learnt classifier comprising a joint classifier such that the fetal head anatomy is detected as a function of a location of other anatomy.

13. The computer readable storage medium of claim 12 wherein detecting comprises detecting a fetal head with a first probabilistic boosting tree classifier, and detecting a cerebellum, cisterna magna, lateral ventricle or combinations thereof as the fetal head anatomy using a second probabilistic boosting tree classifier.

14. The computer readable storage medium of claim 11 wherein detecting comprises detecting with the computer learnt classifier comprising a marginal space classifier.

15. The computer readable storage medium of claim 14 wherein detecting with the marginal space classifier comprises using a translation search with a first detector of the computer learnt classifier, detecting using a scale search limited by a position output by the first detector with a second detector of the computer learnt classifier, and detecting using an orientation search limited by a scale output by the second detector and the position with a third detector of the computer learnt classifier.

16. The computer readable storage medium of claim 11 wherein detecting comprises detecting with detectors trained for different resolutions of the ultrasound data in a volume pyramid.

17. The computer readable storage medium of claim 11 wherein detecting comprises detecting a fetal head, cerebellum, cisterna magna, and lateral ventricles with a joint, hierarchal, marginal space classifier.

18. A system for measuring a fetal parameter from three-dimensional ultrasound data, the system comprising:

a memory operable to store ultrasound data representing a fetal volume;

a processor operable to apply a probabilistic model to detect a first anatomy as a function of the ultrasound data and a position of a second anatomy, the detection of the first anatomy being performed sequentially for at least two of position, orientation and scale; and

a display operable to display an image of the first anatomy, a value representing a measurement of the first anatomy or combinations thereof.

19. The system of claim 18 wherein the probabilistic model comprises a machine learned classifier.

20. The system of claim 18 wherein the second anatomy comprises a fetal skull, the first anatomy comprises a cerebellum, a cisterna magna, or a lateral ventricle, and the probabilistic model comprises a machine-trained classifier.