Method and apparatus for remote tissue identification by statistical modeling and hypothesis testing of echo ultrasound signals

• US 4,542,744 A
• Filed: 03/23/1983
• Issued: 09/24/1985
• Est. Priority Date: 03/23/1983
• Status: Expired due to Fees

• Current Assignee: North American Philips Corp. (Koninklijke Philips N.V.)
• Original Assignee: North American Philips Corp. (Koninklijke Philips N.V.)
• Inventors: Barnes, Casper W., Towfiq, Farhad
• Primary Examiner(s): Howell, Kyle L.
• Assistant Examiner(s): Jaworski, Francis J.
• Application Number:
  US06/478,172
• Time in Patent Office: 916 Days
• Field of Search: 128/660, 128/663, 73/599, 73/602, 73/625-626
• US Class Current: 600/449
• CPC Class Codes: G01N 2291/02475 Tissue characterisation
• CPC Class Codes: G01S 15/8977 using special techniques for image reconstruction, e.g. FFT, geometrical transformations, spatial deconvolution, time deconvolution
• CPC Class Codes: G01S 7/52036 using analysis of echo signal for target characterisation

Abstract

Apparatus and methods for remote identification of tissue types model the scattering of ultrasound energy from living tissue as an autoregressive or autoregressive moving average random process. Autoregressive or autoregressive moving average models of candidate tissue types are generated from pulse-echo data that is known to come from that particular tissue type. Kalman prediction error filters are used for each candidate tissue type to generate estimates of the probability that an unknown pulse echo signal belongs to the class generated by that tissue type. Unknown pulse-echo signals are filtered in a specific Kalman filter to test the hypothesis that the unknown signal belongs to the class associated with that particular Kalman filter.

Claims

1. A method for remote identification of a tissue using ultrasound, comprising the steps of:

directing ultrasound energy into an unknown tissue;

measuring values of a characteristic of echoes of the ultrasound which are scattered from the unknown tissue;

accumulating a statistically significant sample of the measured values of the characteristic;

filtering the accumulated sample to determine, for each of a plurality of predetermined mathematical models, each of which is associated with a possible tissue type, a signal which is a measure of the likelihood that the accumulated sample of measured values was generated by said mathematical model;

applying the the filtered signals as inputs of a predetermined logical decision function which selects one of the models which most likely produced the sample; and

assigning the tissue type which is associated with the chosen model as the identity of the unknown tissue.

2. The method of claim 1 wherein the step of directing energy comprises directing pulses of ultrasound energy into the tissue.

3. The method of claim 1 wherein the step of directing energy comprises directing pulses of ultrasound energy into the tissue and wherein the measured values are values of the instantaneous amplitude of the echoes.

4. The method of claim 3 wherein the step of filtering comprises calculating the logarithm of the likelihood function for each possible tissue type.

5. The method of claim 4 wherein the step of combining the filtered signals comprises choosing the mathematical model having the largest likelihood function.

6. The method of claim 4 wherein the step of combining the filtered signals comprises choosing the model having the largest a posteriori probability.

7. The method of claim 6 further comprising the step of generating an image from the values of the scattered echoes and wherein the step of assigning the tissue type comprises identifying pixels of the image with specific brightness or color values which are associated with the associated tissue type.

8. The method of claim 6, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

9. The method of claim 4 wherein there is a known penalty attached to misidentification of one or more tissue types and wherein the step of combining the filtered signals comprises choosing the mathematical model which minimizes the expected value of the penalty.

10. The method of claim 9 further comprising the step of generating an image from the values of the scattered echoes and wherein the step of assigning the tissue type comprises identifying pixels of the image with specific brightness or color values which are associated with the associated tissue type.

11. The method of claim 9, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

12. The method of claim 3 where the step of filtering determines a signal which is a measure of the likelihood that the accumulated sample of the echo amplitudes was generated by autoregressive mathematical models.

13. The method of claim 12 wherein the step of filtering comprises filtering the accumulated sample with Kalman one-step prediction error filters.

14. The method of claim 13 further comprising the step of generating an image from the values of the scattered echoes and wherein the step of assigning the tissue type comprises identifying pixels of the image with specific brightness or color values which are associated with the associated tissue type.

15. The method of claim 13, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

16. The method of claim 12 wherein the step of combining the filtered signals comprises choosing the mathematical model having the largest likelihood function.

17. The method of claim 16, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

18. The method of claim 12 wherein the step of combining the filtered signals comprises choosing the model having the largest a posteriori probability.

19. The method of claim 12 wherein there is a known penalty attached to misidentification of one or more tissue types and wherein the step of combining the filtered signals comprises choosing the mathematical model which minimizes the expected value of the penalty.

20. The method of claim 12 further comprising the step of generating an image from the values of the scattered echoes and wherein the step of assigning the tissue type comprises identifying pixels of the image with specific brightness or color values which are associated with the associated tissue type.

21. The method of claim 12, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

22. The method of claim 3 wherein the step of filtering determines a signal which is a measure of the likelihood that the accumulated sample of echo amplitudes was generated by autoregressive moving average mathematical models.

23. The method of claim 3 wherein the step of filtering comprises recursively filtering valves of the data sample as successive amplitude values are accumulated.

24. The method of claim 23 wherein the step of combining the filtered signals comprises choosing the mathematical model having the largest likelihood function.

25. The method of claim 23 wherein the step of combining the filtered signal comprises choosing the model having the largest a posteriori probability.

26. The method of claim 23 wherein there is a known penalty attached to misidentification of one or more tissue types and wherein the step of combining the filtered signals comprises choosing the mathematical model which minimizes the expected value of the penalty.

27. The method of claim 23, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

28. The method of claim 3 wherein the step of combining the filtered signals comprises choosing the mathematical model having the largest likelihood function.

29. The method of claim 28 further comprising the step of generating an image from the values of the scattered echoes and wherein the step of assigning the tissue type comprises identifying pixels of the image with specific brightness or color values which are associated with the associated tissue type.

30. The method of claim 28, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

31. The method of claim 3 further comprising the step of generating an image from the values of the scattered echoes and wherein the step of assigning the tissue type comprises identifying pixels of the image with specific brightness or color values which are associated with the associated tissue type.

32. The method of claim 3, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

33. The method of claim 1, wherein the step of filtering comprises filtering the accumulated sample in one or more filters which are characterized by numerical parameters and further comprising the step of determining numerical parameters which characterize each possible tissue type from samples of echoes which are scattered from a known sample of the particular tissue type.

34. A method for manufacturing a filter which extracts, from a statistically significant sample of the values of signals which represent detected echoes of ultrasound which are scattered from unknown tissue, a signal which is a measure of the likelihood that the accumulated sample was generated by a mathematical model which characterizes a known tissue type, comprising the steps of:

directing ultrasound energy into a sample of the known tissue;

detecting values characteristic of echoes of the ultrasound which are scattered from the known tissue;

accumulating a statistically significant sample of the detected characteristic values;

calculating, from the accumulated sample, autoregressive parameters which characterize an autoregressive process which models the sample values; and

constructing a Kalman predictive filter which incorporates the determined autoregressive parameters.

35. The method of claim 34 wherein the step of detecting values of a characteristic of the echoes comprises detecting samples of the instantaneous amplitude of the echoes.

36. The method of claim 35 wherein the known tissue type comprises a tissue selected from the class of tissues consisting of human liver tissue, human pancreas tissue, and human spleen tissue.

37. A filter manufactured by the process of claim 36.

38. A filter manufactured by the process of claim 35.

39. A filter manufactured by the process of claim 34.

40. Appartus for remote identification of a tissue comprising:

means for directing ultrasound energy into an unknown tissue;

means for detecting a characteristic of echoes of the ultrasound which are scattered from the unknown tissue;

means for accumulating a statistically significant sample of signals which represent values of the detected characteristic;

a plurality of filter means, each of which function to filter the accumulated sample to extract a signal which is a measure of the likelihood that the accumulated sample was generated by an associated, predetermined mathematical model, wherein each of said mathematical models is associated with a possible tissue type; and

means which apply the signals extracted by the filters, as inputs to a predetermined logical decision function to select the extracted signal which is associated with the model which most likely produced the sample and assign the tissue type associated with the model producing the chosen signal as the identity of the unknown tissue.

41. The apparatus of claim 40 wherein for directing energy direct pulses of ultrasound energy into the unknown tissue and wherein the means which detect the characteristic of the echoes extracts a time sequence of samples of the value of the instantaneous amplitude of the echoes.

42. The apparatus of claim 41 wherein the mathematical models associated with the filter means are autoregressive models.

43. The apparatus of claim 42 wherein the filter means comprise Kalman one step error prediction filters.

44. The apparatus of claim 43 wherein the means for combining signals choose the extracted signal having the largest value.

45. The apparatus of claim 43, further comprising means for displaying an image of at least the unknown tissue and wherein the means for combining the filtered signals functions to identify the unknown tissue by assigning a predetermined brightness and/or color value with associated pixels of the displayed image.

46. The apparatus of claim 43 wherein the filter means comprise one or more filters which are characterized by a series of autoregressive parameters which are produced by the process of:

directing pulses of ultrasound energy into a sample of a known tissue type;

detecting a sequence of values of the instantaneous amplitude of echoes of the ultrasound which are scattered from the known tissue type;

accumulating a statistically significant sample of the detected amplitudes; and

calculating values of autoregressive parameters which model the sequence of accumulated samples as an autoregressive process and incorporating the calculated values in the filter.

47. The apparatus of claim 42 wherein the filter means operate recursively to process data samples as they are accumulated.

48. The apparatus of claim 47 wherein the filter means comprise one or more filters which are characterized by a series of autoregressive parameters which are produced by the process of:

directing pulses of ultrasound energy into a sample of a known tissue type;

detecting a sequence of values of the instantaneous amplitude of echoes of the ultrasound which are scattered from the known tissue type;

accumulating a statistically significant sample of the detected amplitudes; and

calculating values of autoregressive parameters which model the sequence of accumulated samples as an autoregressive process and incorporating the calculated values in the filter.

49. The apparatus of claim 42 wherein the means for combining signals choose the extracted signal having the largest value.

50. The apparatus of claim 42 wherein the means for combining signals choose the extracted signal having the largest value.

51. The apparatus of claim 42, further comprising means for displaying an image of at least the unknown tissue and wherein the means for combining the filtered signals functions to identify the unknown tissue by assigning a pedetermined brightness and/or color value with associated pixels of the displayed image.

52. The apparatus of claim 42 wherein the filter means comprise one or more filters which are characterized by a series of autoregressive parameters which are produced by the process of:

directing pulses of ultrasound energy into a sample of a known tissue type;

detecting a sequence of values of the instantaneous amplitude of echoes of the ultrasound which are scattered from the known tissue type;

accumulating a statistically significant sample of the detected amplitudes; and

calculating values of autoregressive parameters which model the sequence of accumulated samples as an autoregressive process and incorporating the calculated values in the filter.

53. The apparatus of claim 41 wherein the mathematical models associated with the filters are autoregressive moving average models.

54. The apparatus of claim 41 wherein the means for combining signals choose the extracted signal having the largest value.

55. The apparatus of claim 54 wherein the filter means comprise one or more filters which are characterized by a series of autoregressive parameters which are produced by the process of:

directing pulses of ultrasound energy into a sample of a known tissue type;

detecting a sequence of values of the instantaneous amplitude of echoes of the ultrasound which are scattered from the known tissue type;

accumulating a statistically significant sample of the detected amplitudes; and

calculating values of autoregressive parameters which model the sequence of accumulated samples as an autoregressive process and incorporating the calculated values in the filter.

56. The apparatus of claim 41, further comprising means for displaying an image of at least the unknown tissue and wherein the means for combining the filtered signals functions to identify the unknown tissue by assigning a predetermined brightness and/or color value with associated pixels of the displayed image.

57. The apparatus of claim 41 wherein the filter means comprise one or more filters which are characterized by a series of autoregressive parameters which are produced by the process of:

directing pulses of ultrasound energy into a sample of a known tissue type;

detecting a sequence of values of the instantaneous amplitude of echoes of the ultrasound which are scattered from the known tissue type;

accumulating a statistically significant sample of the detected amplitudes; and

calculating values of autoregressive parameters which model the sequence of accumulated samples as an autoregressive process and incorporating the calculated values in the filter.

58. The apparatus of claim 40 wherein the means for combining signals choose the extracted signal having the largest value.

59. The apparatus of claim 40, further comprising means for displaying an image of at least the unknown tissue and wherein the means for combining the filtered signals functions to identify the unknown tissue by assigning a predetermined brightness and/or color value with associated pixels of the displayed image.

60. The apparatus of claim 40 wherein the filter means comprise one or more filters which are characterized by a series of autoregressive parameters which are produced by the process of:

directing pulses of ultrasound energy into a sample of a known tissue type;

detecting a sequence of values of the instantaneous amplitude of echoes of the ultrasound which are scattered from the known tissue type;

accumulating a statistically significant sample of the detected amplitudes; and

calculating values of autoregressive parameters which model the sequence of accumulated samples as an autoregressive process and incorporating the calculated values in the filter.