Ultrasonic tissue-type classification and imaging methods and apparatus

• US 6,238,342 B1
• Filed: 05/26/1999
• Issued: 05/29/2001
• Est. Priority Date: 05/26/1998
• Status: Expired due to Term

• Current Assignee: Riverside Research Institute
• Original Assignee: Riverside Research Institute
• Inventors: Liu, Tian, Feleppa, Ernest Joseph, Lizzi, Frederic Louis
• Primary Examiner(s): Lateef, Marvin M.
• Assistant Examiner(s): Imam, Ali M.
• Application Number:
  US09/318,892
• Time in Patent Office: 734 Days
• Field of Search: 600/442, 600/407, 600/437, 600/443, 128/916, 382/132, 382/286-287, 73/625
• US Class Current: 600/437
• CPC Class Codes: A61B 17/3403 Needle locating or guiding means guides for suture needles A61B17/0482; guiding or tracking by nuclear magnetic resonance G01R33/285
• CPC Class Codes: A61B 2017/00274 Prostate operation, e.g. prostatectomy, turp, bhp treatment
• CPC Class Codes: A61B 2018/00547 Prostate
• CPC Class Codes: A61B 8/00 Diagnosis using ultrasonic, sonic or infrasonic waves imaging of objects using sonar G01S15/00
• CPC Class Codes: A61B 8/08 Detecting organic movements or changes, e.g. tumours, cysts, swellings A61B8/02 - A61B8/06 take precedence
• CPC Class Codes: G01S 7/52036 using analysis of echo signal for target characterisation
• CPC Class Codes: G01S 7/52084 related to particular user interfaces special user input means for ultrasonic medical diagnostic devices A61B8/467

Abstract

Clinical data, ultrasonic radio frequency (RF) backscatter spectral data and histological results of corresponding biopsy sites are stored in a database and are used to train a classifier suitable for real-time tissue classification and imaging. In clinical use, clinical data and ultrasonic RF backscatter data are applied as input variables to the trained classifier which assigns a likelihood of cancer (LOC) score to each pixel location in an ultrasound image. The LOC scores are then categorized by ranges, which can be established by user selected threshold values, to apply different colors or grey scale values distinguishing varying levels of suspicion (LOS) to each pixel position in real-time. This classification and display technique is especially valuable for guiding a person performing a needle biopsy of the prostate.

Claims

1. An ultrasound apparatus for performing material classification comprising:

an ultrasound scanner, said scanner acquiring RF echo signals related to material properties;

a digitizer, said digitizer responsive to said ultrasound scanner and converting the RF echo signals to digital signals representing echo signals from a plurality of spatial regions in material under examination;

a processor, said processor responsive to the digital signals from said digitizer and extracting a plurality of spectral parameters relating to the RF echo signals;

an input device, said input device for providing non-ultrasound material-related data to said processor;

a classifier, said classifier being responsive to at least a portion of said plurality of spectral parameters and at least a portion of said non-ultrasound material-related data and assigning a material classification score to the points of the ultrasound scan; and

an output device for providing data corresponding to the assigned material classification scores for at least one area corresponding to said spatial regions in said material.

2. The ultrasound apparatus as defined by claim 1, wherein the classifier includes a look-up table, said table relating each combination of said portion of spectral parameters and said portion of material-related data to a predetermined material classification score.

3. The ultrasound apparatus as defined by claim 1, wherein the classifier includes a trained neural network having a plurality of inputs for receiving said portion of spectral parameters and said portion of material-related data, said trained neural network for processing said spectral parameters and material-related data for each region to determine a material classification score for each region.

4. The ultrasound apparatus as defined by claim 1, wherein said plurality of spectral parameters include at least one of a slope characteristic, an intercept characteristic and a mid band characteristic of a linear estimation of a power spectrum of the RF echo signals.

5. The ultrasound apparatus as defined by claim 1, wherein said material is tissue and said material classification scores are related to different tissue types.

6. The ultrasound apparatus as defined by claim 5, wherein said different tissue types relate to a likelihood of cancer and said material-related parameters include a prostate specific antigen level.

7. The ultrasound apparatus as defined by claim 5, wherein the tissue classification score for the points in the ultrasound scan define a range of values, wherein:

the range of values is divided into a plurality of sub-ranges which define varying levels of suspicion of cancerous tissue;

said output device is a display for providing an ultrasound image; and

said processor assigns a unique display parameter corresponding to each level of suspicion.

8. The ultrasound apparatus as defined by claim 7, wherein the unique display parameter is a color assigned to each level of suspicion.

9. The ultrasound apparatus as defined by claim 7, wherein the unique display parameter is a grey scale level assigned to each level of suspicion.

10. An ultrasound apparatus for building a database and training a classifier for a material classification system, the apparatus comprising:

an ultrasound scanner, said scanner acquiring RF echo signals related to material properties;

a digitizer, said digitizer responsive to said ultrasound scanner and converting the RF signals to digital signals representing echo signals from a plurality of spatial regions in material under examination;

a processor, said processor responsive to the digital signals from said digitizer and extracting a plurality of spectral parameters relating to the RF signals;

an input device, said input device for providing non-ultrasound material-related data to said processor;

a non-volatile computer storage device for storing records of a database, the records including non-ultrasound material-related data inputted through the input device, spectral parameter data provided by the processor and corresponding conclusive material-type data from determinative test results;

a classifier, said classifier having a plurality of inputs and providing a material classification score output value, said classifier being operatively coupled to said database and, during a training operation, receiving at least a portion of said spectral parameters and at least a portion of said non-ultrasound material-related data as input parameters and the conclusive data as an expected output value and adapting the processing of the input parameters such that the output value substantially matches the expected output value.

11. The ultrasound apparatus as defined by claim 10, wherein the classifier includes a look-up table, said look-up table relating each combination of said portion of spectral parameters and said portion of material-related data to a predetermined memory location, and during a training operation, said memory locations are provided material classification scores corresponding to each unique combination of input parameters.

12. The ultrasound apparatus as defined by claim 10, wherein the classifier includes a neural network having a plurality of inputs for receiving said portion of spectral parameters and said portion of non-ultrasound material-related data, and processing said portion of spectral parameters and non-ultrasound material-related data to determine a material classification score for said spatial points.

13. The ultrasound apparatus as defined by claim 10, wherein said plurality of spectral parameters include at least one of a slope characteristic, an intercept characteristic and a mid band characteristic of a straight line estimation of a power spectrum of the RF echo signals.

14. The ultrasound apparatus as defined by claim 10, wherein said material is tissue and said material classification scores relate to different tissue types.

15. The ultrasound apparatus as defined by claim 14, wherein said different tissue types relate to a likelihood of cancer and said material-related parameters include a prostate specific antigen level.

16. The ultrasound apparatus as defined by claim 15, wherein said different tissue types relate to a likelihood of cancer in the prostate and said material-related parameters include a prostate specific antigen level.

17. The ultrasound apparatus as defined by claim 16, wherein said material-related parameters also include age, and said prostate specific antigen level is adjusted in accordance with age.

18. A method of performing material classification of a material under test using ultrasound, comprising:

receiving non-ultrasound material-related data for the material under test;

performing an ultrasound scan of the material under test and receiving RF echo data therefrom;

extracting a plurality of spectral parameters from the RF echo data corresponding to a plurality of spatial regions of the material under test;

applying at least a portion of the non-ultrasound material-related data and at least a portion of the spectral parameters to a classifier which assigns a score for each of the plurality of spatial regions which is an indicator of material classification;

outputting said assigned scores for said spatial regions.

19. The method of classifying as defined by claim 18, wherein said outputting step comprises displaying an image of said tissue wherein each image area corresponds to one of said spatial regions and has an image characteristic selected according to said assigned scores for said spatial region.

20. The method of classifying as defined by claim 19, wherein the material is tissue, said material-related data is clinical data and said material classification represents a likelihood of cancer.

21. The method of classifying as defined by claim 19, wherein the scores are grouped into a plurality of ranges, each range being assigned a unique image characteristic.

22. The method of classifying as defined by claim 19, wherein the unique image characteristic is color.

23. The method of classifying as defined by claim 19, wherein the unique image characteristic is grey scale value.

24. The method of classifying as defined by claim 18, wherein said step of performing an ultrasound scan, extracting spectral parameters, and assigning a score to each point are repeated over a plurality of ultrasound scan planes defining a volume of a region of tissue, and wherein said displaying step comprises forming a three dimensional image from the plurality of ultrasound scan planes.

25. A method of creating a look-up table for classifying material by material type using ultrasound imaging, comprising:

creating a database of records of subject data, for each record;

receiving and storing non-ultrasound material-related data regarding the subject;

performing an ultrasound scan of the tissue and collecting RF echo data therefrom;

extracting and storing a plurality of spectral parameters from the RF echo data at a plurality of points of the ultrasound scan; and

receiving and storing conclusive data of at least a portion of the material;

training a neural network by applying at least a portion of the non-ultrasound material-related data and at least a portion of the spectral parameters as input parameters, processing the input parameters to develop an output signal, comparing the output signal to the conclusive data to create an error value, and iteratively adjusting the processing to minimize the error value; and

applying a substantially continuous range of input parameters to the trained neural network and generating output signals as values for a look-up table indicative of tissue type.

26. The method of creating a look-up table as defined by claim 25, wherein said plurality of spectral parameters include at least one of a slope characteristic, an intercept characteristic and a mid band characteristic of a straight line estimation of the spectrum of the RF echo signals.

27. The method of creating a look-up table as defined by claim 25, wherein the material is tissue and said look-up table values represent a likelihood of cancer.

28. The method of creating a look-up table as defined by claim 27, wherein said material-related data is at least one clinical parameter related to cancer of a particular tissue type.

29. The method of creating a look-up table as defined by claim 28, wherein the at least one clinical parameter includes a prostate specific antigen level.