System and method for adaptive filtering of images based on similarity between histograms

• US 5,594,807 A
• Filed: 12/22/1994
• Issued: 01/14/1997
• Est. Priority Date: 12/22/1994
• Status: Expired due to Term

• Current Assignee: Siemens Medical Solutions USA Incorporated (Siemens AG)
• Original Assignee: Siemens Medical Systems, Inc. (Siemens AG)
• Inventors: Liu, Dong-Chyuan
• Primary Examiner(s): JOHNS, ANDREW W
• Application Number:
  US08/363,191
• Time in Patent Office: 754 Days
• Field of Search: 382/128, 382/168, 382/170, 382/171, 382/260, 382/261, 382/275, 364/413.25, 364/724.19, 128/660.01, 128/662.02, 348/607, 348/163
• US Class Current: 382/128
• CPC Class Codes: G06T 2207/10132 Ultrasound image
• CPC Class Codes: G06T 2207/30004 Biomedical image processing
• CPC Class Codes: G06T 5/40 using histogram techniques
• CPC Class Codes: G06T 5/94 based on local image properties, e.g. for local contrast enhancement

Abstract

An interrogation region, such as a portion of a patient'"'"'s body, is imaged as an unfiltered pattern of image elements, to each of which is assigned a numerical value such as brightness. One element is selected either automatically or manually as a reference element, which lies in a substantially homogeneous image portion. For ultrasonic imaging, the homogeneous portion preferably corresponds to a region of speckle noise. A reference histogram of the numerical values is then accumulated for all the elements in a region or window about the reference element. All of the remaining elements are sequentially designated as current elements, for each of which a current histogram is accumulated. Each current histogram is compared with the reference histogram using an error function for each corresponding pair of histogram bins. The error function values for all bins, which may be normalized and weighted, are used to generate a similarity value between the current and reference histograms. The more similar to the reference histogram the current histogram is, the more strongly it is assumed that the current element lies in a homogeneous portion of the image, and the greater it is smoothed. All possible error function values may be prestored to increase computation speed. The degree of smoothing may be adjusted by the user. The user preferably also may adjust a comparison factor, which determines how strongly a difference between a current and a reference bin will contribute to the similarity value.

Claims

1. A method for imaging an interrogation region, comprising the following steps:

A. generating a multi-dimensional image of the interrogation region as a pattern of image elements;

B. assigning to each image element one of a predetermined, finite number of numerical values corresponding to a predetermined imaging property of the interrogation region;

C. displaying the image on a display;

D. selecting a reference element in a substantially homogeneous portion of the displayed image;

E. accumulating in a reference histogram a reference frequency distribution of the numerical values of all image elements in a reference region, which includes the reference element;

F. sequentially designating each image element as a current element;

G. for each current element, accumulating in a current histogram a current frequency distribution of the numerical values of all image elements in a current region, which includes the current element;

H. for each current element, determining a similarity value between the reference histogram and the current histogram; and

I. displaying each current element with an updated, filtered value, which is a predetermined filtering function of the corresponding similarity value.

2. A method as defined in claim 1, in which:

A. the similarity value is generated from a similarity function, which includes a comparison factor;

B. when the comparison factor is at a first value, the similarity function has a predetermined minimum value; and

C. when the comparison factor is at a second value, the similarity function has a predetermined maximum value;

further including the following step;

D. adjusting the comparison factor until a user-defined noise removal level is reached for the displayed image.

3. A method as defined in claim 2, further including the following steps:

A. storing the reference histogram in a plurality of reference memory locations as a plurality of reference bins, in which each reference bin corresponds to a predetermined range of numerical values of the imaging property, and into each of which is accumulated the number of image elements in the reference region whose numerical values fall within the range of the corresponding bin;

B. storing the current histogram in a plurality of current memory locations as a plurality of current bins corresponding in number and numerical ranges to the reference bins of the reference histogram, and in which into each current bin is accumulated the number of image elements in the current region whose numerical values fall within the range of the corresponding current bin;

C. for each reference bin and corresponding current bin, generating an error function value;

D. generating the similarity value as a predetermined similarity function of the error function values; and

E. prestoring in a memory unit all possible error function values for arbitrary reference bins and current bins.

4. A method as defined in claim 3, further including the following steps:

A. determining the reference histogram bin in whose range a greatest number B_peak of reference elements'"'"' numerical values lie;

B. for each current histogram, determining the current histogram bin in whose range the greatest number of current elements'"'"' numerical values lie;

C. scaling all bins of the current histogram by a normalization factor and assigning to the bin in whose range the greatest number of elements'"'"' numerical values lie the value B_peak ; and

D. prestoring in the memory unit all possible normalization factors for the given number of image elements in the reference region.

5. A method as defined in claim 4, further including the following steps:

A. partitioning the histograms into left and right halves, with an equal number of bins in either half;

B. adjusting the ranges of the bins in the left and right halves, respectively, to be proportional to respective total widths of the left and right halves;

C. generating the error function values as a function of separate weights for histogram bins in the left and right halves, with the weights proportional to a reciprocal of a ratio of the total widths of the left and right halves; and

D. whereby the similarity function is a function of skewness of the histograms.

6. A method as defined in claim 1, further including the following steps:

A. displaying each current element with a respective numerical value in a range from its unchanged value to a maximum filtered value as a function of the corresponding similarity value and a smoothing factor; and

B. adjusting the smoothing factor until a user-defined noise removal level is reached for the displayed image.

7. A method as defined in claim 6, further including the following steps:

A. setting a predetermined number of similarity ranges;

B. for each similarity range and for each current element, storing a corresponding average sub-region value that is an average of the numerical values of the image elements in a corresponding sub-region, which includes the current element, of the current region; and

C. displaying each current element with a respective numerical value in a range from its unchanged value to the average sub-region value of the sub-region in whose range the similarity value of the current element lies.

8. A method as defined in claim 1, comprising the following steps:

A. defining the interrogation region as a portion of a patient'"'"'s body;

B. generating the multi-dimensional image of the interrogation region as a two-dimensional image by applying to the interrogation region an ultrasonic scan signal, sensing a corresponding ultrasonic return signal, and by converting the ultrasonic return signal into the pattern of pixels;

C. assigning to each pixel one of a predetermined, finite number of brightness values corresponding to a predetermined acoustic property of a corresponding portion of the interrogation region; and

D. selecting as the reference element a reference pixel in the substantially homogeneous portion of the displayed image such that pixels in the substantially homogeneous portion with substantially different displayed brightness values than those of adjacent pixels can be assumed to result from speckle noise.

9. A method as defined in claim 1, in which the step of selecting a reference element includes the steps of automatically comparing predetermined statistics of a plurality of unfiltered image elements and of selecting as the reference element the one of the unflitered image element whose statistics most closely correspond to a predetermined homogeneity condition.

10. An ultrasonic imaging method comprising the following steps:

A. selecting as an interrogation region a portion of a patient'"'"'s body;

B. generating a multi-dimensional image of the interrogation region as a two-dimensional image by applying to the interrogation region an ultrasonic scan signal, sensing a corresponding ultrasonic return signal, and by converting the ultrasonic return signal into the pattern of pixels;

C. assigning to each pixel one of a predetermined, finite number of brightness values corresponding to a predetermined acoustic property of a corresponding portion of the interrogation region;

D. selecting a reference pixel in a substantially homogeneous portion of the displayed image such that pixels in the substantially homogeneous portion with substantially different displayed brightness values than those of adjacent pixels can be assumed to result from speckle noise;

E. accumulating in a reference histogram a reference frequency distribution of the numerical brightness values of all pixels in a reference region, which includes the reference pixel;

F. storing the reference histogram in a plurality of reference memory locations as a plurality of reference bins, in which each reference bin corresponds to a predetermined range of numerical brightness values, and into each of which is accumulated the number of image pixels in the reference region whose numerical brightness values fall within the range of the corresponding bin;

G. sequentially designating each pixel as a current pixel;

H. for each current pixel, accumulating in a current histogram a current frequency distribution of the numerical values of all pixels in a current region, which includes the current pixel;

I. storing the current histogram in a plurality of current memory locations as a plurality of current bins corresponding in number and numerical ranges to the reference bins of the reference histogram, and in which into each current bin is accumulated the number of pixels in the current region whose numerical brightness values fall within the range of the corresponding current bin;

J. determining the reference histogram bin in whose range a greatest number B_peak of reference pixels'"'"' brightness values lie;

K. for each current histogram, determining the current histogram bin in whose range the greatest number of current pixels'"'"' numerical values lie;

L. scaling all bins of the current histogram linearly by a normalization factor and, to the bin in whose range the greatest number of pixels'"'"' numerical values lie, assigning the value B_peak ; and

M. prestoring in the memory unit all possible normalization factors for the given number of image elements in the reference region;

N. for each reference bin and corresponding current bin, generating an error function value;

O. prestoring in a memory unit all possible error function values for arbitrary reference bins and current bins.

P. fix each current pixel, generating a similarity value between the reference histogram and the current histogram as a predetermined similarity function of the error function values;

in which;

1) the similarity value is generated from a similarity function, which includes a comparison factor;

2) when the comparison factor is at a first value, the similarity function has a predetermined minimum value; and

3) when the comparison factor is at a second value, the similarity function has a predetermined maximum value;

Q. adjusting the comparison factor until a user-defined noise removal level is reached for the displayed image;

R. displaying each current pixel with a respective numerical brightness value in a range from its unchanged value to a maximum filtered value as a function of the corresponding similarity value and a smoothing factor;

S. adjusting the smoothing factor until a user-defined noise removal level is reached for the displayed image; and

T. displaying each current pixel with an updated, filtered brightness value, which is a predetermined filtering function of the corresponding similarity value.

11. A system for imaging an interrogation region, comprising:

A. a source of a scanning signal which has, as an unformed output, a received imaging signal corresponding to a predetermined imaging property of the interrogation region;

B. a signal converter that has, as an unformed input, the received imaging signal, and that has a formed output corresponding to a multi-dimensional image of the interrogation region as a pattern of image elements, with each image element having one of a predetermined, finite number of numerical values corresponding to the predetermined imaging property of the interrogation region;

C. a display means for displaying the image as the pattern of image elements;

D. reference selection means for selecting one of the displayed image elements as a reference element in a substantially homogeneous portion of the displayed image;

E. reference accumulation means for accumulating and storing reference histogram values in a memory unit for a reference frequency distribution of the numerical values of all image elements in a reference region, which includes the reference element;

F. current accumulation means for sequentially designating each image element as a current element, and, for each current element, for accumulating and storing current histogram values in the memory unit for a current frequency distribution of the numerical values of all image elements in a current region, which includes the current element; and

G. filtering means;

1) for determining a similarity value between the reference histogram values and the current histogram values; and

2) for changing for display the corresponding numerical value for each current pixel to an updated value as a predetermined filtering function of the corresponding similarity value.

12. A system as defined in claim 11, further comprising:

A. comparison adjustment means for setting a histogram comparison factor from a minimum to a maximum value, in which;

B. the similarity value is a predetermined similarity function of the comparison factor such that, when the comparison factor is at a first value, the similarity function has a predetermined minimum value, and, when the comparison factor is at a second value, the similarity function has a predetermined maximum value.

13. A system as defined in claim 11, further comprising:

A. smoothness adjustment means for setting a histogram smoothness factor from a minimum to a maximum value, in which;

B. when the smoothness factor is at a first value, a maximum possible degree of change between each current pixel and its updated value is at a predetermined minimum; and

C. when the smoothness factor is at a second value, the maximum possible degree of change between each current pixel and its updated value is at a predetermined maximum.

14. A system as defined in claim 11, in which the source of the scanning signal is an ultrasonic transducer.

15. A system as defined in claim 11, in which the reference selection means is an automatic reference identification means for comparing predetermined statistics of a plurality of unfiltered image elements and for selecting as the reference element the one of the unflitered image element whose statistics most closely correspond to a predetermined homogeneity condition.