Apparatus and method for computer-aided diagnosis

• US 10,147,223 B2
• Filed: 04/03/2017
• Issued: 12/04/2018
• Est. Priority Date: 10/24/2013
• Status: Active Grant

• Current Assignee: Samsung Electronics Co. Ltd.
• Original Assignee: Samsung Electronics Co. Ltd.
• Inventors: Kim, Ye-Hoon, Park, Moon-Ho, Seong, Yeong-Kyeong, Kim, Jung-Hoe, Cho, Baek-Hwan
• Primary Examiner(s): CHANG, JON CARLTON
• Application Number:
  US15/477,771
• Publication Number:
  US 20170206699A1
• Time in Patent Office: 610 Days
• CPC Class Codes: A61B 5/0033 Features or image-related aspects of imaging apparatus classified in A61B5/00, e.g. for MRI, optical tomography or impedance tomography apparatus; arrangements of imaging apparatus in a room image data processing or generation G06T
• CPC Class Codes: G01R 33/4814 MR combined with ultrasound
• CPC Class Codes: G06F 2218/12 Classification; Matching
• CPC Class Codes: G06T 15/005 General purpose rendering architectures
• CPC Class Codes: G06T 15/08 Volume rendering
• CPC Class Codes: G06T 19/003 Navigation within 3D models or images
• CPC Class Codes: G06T 2200/04 involving 3D image data
• CPC Class Codes: G06T 2200/08 involving all processing steps from image acquisition to 3D model generation
• CPC Class Codes: G06T 2207/10072 Tomographic images
• CPC Class Codes: G06T 2207/10088 Magnetic resonance imaging [MRI]
• CPC Class Codes: G06T 2207/10136 3D ultrasound image
• CPC Class Codes: G06T 7/0012 Biomedical image inspection

Abstract

An apparatus and method for medical diagnostics includes receiving three-dimensional (3D) volume data of a part of a patient'"'"'s body, and generating two-dimensional (2D) slices including cross-sections of the 3D volume data cut from a cross-section cutting direction. The apparatus and the method also determine whether a lesion in each of the 2D slices is benign or malignant and output results indicative thereof, select a number of the 2D slices based on the results, and make a final determination whether the lesion is benign or malignant based on the selected 2D slices.

Claims

1. A method for medical diagnostics, comprising:

receiving three-dimensional (3D) volume data of at least a part of an object;

generating two-dimensional (2D) slices including cross-sections of the 3D volume data based on a cross-section cutting direction;

classifying a lesion in at least two of the generated 2D slices;

calculating confidence levels of results of the classification;

selecting a plurality of the at least two 2D slices based on the confidence levels of the results of the classification; and

classifying a lesion included in the 3D volume data based on at least one of the classification results of the selected 2D slices or the confidence levels of the classification results of the selected 2D slices.

2. The method of claim 1, further comprising:

outputting the classification result of the lesion included in the 3D volume data.

3. The method of claim 1, wherein the classifying of the lesion in at least two of the generated 2D slices comprises applying the at least two 2D slices to a diagnostic model to classify the lesion in the at least two 2D slices.

4. The method of claim 3, wherein the classifying of the lesion in at least two of the generated 2D slices comprises applying the at least two 2D slices to a single diagnostic model that is generated based on the cross-section cutting direction.

5. The method of claim 3, wherein the classifying of the lesion in at least two of the generated 2D slices comprises applying the at least two 2D slices to respective diagnostic models that are generated based on cross-section cutting directions of the at least two 2D slices.

6. The method of claim 1, wherein the generating of the 2D slices comprises:

generating at least one virtual plane; and

generating the 2D slices including cross-sections of the 3D volume data cut by the virtual plane.

7. The method of claim 6, wherein the generating of the at least one virtual plane comprises at least one of:

generating the at least one virtual plane by changing coefficient values of a plane equation that represents an arbitrary plane of the 3D volume data;

generating the at least one virtual plane by performing principal component analysis (PCA) on the 3D volume data;

or

generating the at least one virtual plane based on a user'"'"'s input information.

8. The method of claim 7, wherein the generating of the at least one virtual plane further comprises at least one of:

determining feature points having a predetermined feature from among voxels of the 3D volume data based on values of the voxels, and generating the at least one virtual plane based on a distribution of the feature points by performing the PCA;

calculating a first principal component vector corresponding to an axis in a direction in which a change in the 3D volume data is the greatest by performing the PCA, and generating the at least one virtual plane with reference to the first principal component vector;

or

detecting a mass included in the 3D volume data based on values of voxels of the 3D volume data, and generating the at least one virtual plane based on a distribution of points included in the mass by performing the PCA.

9. An apparatus for medical diagnostics, comprising:

a receiver configured to receive three-dimensional (3D) volume data of at least a part of an object; and

at least one processor configured to;

generate two-dimensional (2D) slices including cross-sections of the 3D volume data based on a cross-section cutting direction,

classify a lesion in at least two of the generated 2D slices,

calculate confidence levels of the results of the classification,

select a plurality of the at least two 2D slices based on the confidence levels of the results of the classification, and

classify a lesion included in the 3D volume data based on at least one of the classification results of the selected 2D slices or the confidence levels of the classification results of the selected 2D slices.

10. The apparatus of claim 9, wherein the at least one processor is further configured to output the classification result of the lesion included in the 3D volume data.

11. The apparatus of claim 9, wherein the at least one processor is further configured to classify the lesion in the at least two 2D slices by applying the at least two 2D slices to a diagnostic model.

12. The apparatus of claim 11, wherein the at least one processor is further configured to classify the lesion in the at least two 2D slices by applying the at least two 2D slices to a single diagnostic model generated based on a cross-section cutting direction.

13. The apparatus of claim 11, wherein the at least one processor is further configured to classify the lesion in the at least two 2D slices by applying the at least two 2D slices to respective diagnostic models generated based on cross-section cutting directions of the at least two 2D slices.

14. The apparatus of claim 9, wherein the at least one processor is further configured to:

generate at least one virtual plane, and

generate 2D slices of cross-sections of the 3D volume data cut by the virtual plane.

15. The apparatus of claim 14, wherein the at least one processor is further configured to generate the at least one virtual plane by changing coefficients of a plane equation of an arbitrary plane of the 3D volume data by at least one of:

performing principal component analysis (PCA) on the 3D volume data, or

based on a user'"'"'s input information.

16. The apparatus of claim 15, wherein the at least one processor is further configured to:

determine feature points with a predetermined feature from among voxels of the 3D volume data based on values of the voxels, and generate the at least one virtual plane based on a distribution of the feature points,

calculate a first principal component vector corresponding to an axis in a direction toward which a greatest change occurs in the 3D volume data by performing the PCA, and generate the at least one virtual plane based on the first principal component vector, or

detect a mass included in the 3D volume data based on values of voxels of the 3D volume data, and generate the at least one virtual plane based on a distribution of points included in the mass by performing the PCA.

17. The apparatus of claim 9, wherein the at least one processor is further configured to:

generate a feature vector using extracted features of the lesion in the at least two 2D slices, and

apply the feature vector to a diagnostic model to classify the lesion in the at least two 2D slices.

18. The apparatus of claim 9, wherein the at least one processor is further configured to:

bind two or more 2D slices into one group,

compare confidence levels of the 2D slices in the group, and

select one or more 2D slices in descending order of the confidence levels thereof.

19. The apparatus of claim 9, wherein the at least one processor is further configured to:

randomly bind the selected 2D slices into one group,

compare confidence levels thereof, and

select one or more 2D slices in descending order of the confidence levels thereof.

20. A computer program product comprising a non-transitory computer readable storage medium having a computer readable program stored therein, wherein the computer readable program, when executed on at least one processor, configures the at least one processor to:

receive three-dimensional (3D) volume data of at least a part of an object;

generate two-dimensional (2D) slices including cross-sections of the 3D volume data based on a cross-section cutting direction;

classify a lesion in at least two of the generated 2D slices;

calculate confidence levels of results of the classification;

select a plurality of the at least two 2D slices based on the confidence levels of the results of the classification; and

classify a lesion included in the 3D volume data based on at least one of the classification results of the selected 2D slices or the confidence levels of the classification results of the selected 2D slices.