Ultrasound imaging method/technique for speckle reduction/suppression in an improved ultra sound imaging system

Abstract

The present invention relates to an improved ultrasound imaging method/technique for speckle reduction/suppression in an ultra sound imaging system in which scan conversion and speckle reduction is performed simultaneously in the scan conversion stage avoiding any kind of conventional interpolation. An improved method for speckle reduction in an ultrasound imaging system and an improved ultra sound imaging system for speckle reduction is provided in the present invention. The method comprises steps of receiving in a processor means raw data samples as an input comprising image signals with noises from a logarithmic amplifier, processing the received image signals for scan conversion and speckle reduction in the processor means so as to get pixel value from the raw data samples and to perform speckle reduction so as to provide speckle filtered output image.

Claims

1. A method for speckle reduction in an ultrasound imaging system, said method comprising steps of: receiving, at a processor, data samples comprising image signals with noise from a logarithmic amplifier; and processing, via the processor, said image signals by simultaneously performing scan conversion and speckle reduction so as to get a pixel values from said data samples and to perform speckle reduction to provide a speckle filtered output image; wherein, said pixel values at raster grid points in a rectangular coordinate system are determined using a speckle reduction by means of speckle reduction filter implemented along with scan conversion; wherein processing of said image signals to get the pixel values at each point where radial lines cut the horizontal grid lines, comprising: evaluation a plurality of nearest points on said radial lines with respect to said cut point where one of said plurality of said radial line cuts a one of said horizontal grid line: assigning sample values to said nearest points, said sample values lying substantially around said cut point; and imposing said speckle reduction by means of a single scale spatial filter to compute a one of said pixel values for said cut point; wherein imposing said speckle reduction comprises applying a high pass filter technique with edge enhancement to enhance at least one of positive edge slope or both positive and negative edge slope; wherein said high pass filter technique has positive and negative weight co-efficient determined by: $w\left(i,j\right)=\mathrm{INT}\left[w\left(K+1,K+1\right)-d\frac{20*{\log }_{10}\left(1+m+\right)}{{\log }_{10}\left(m\right)}\right]$ adapted to enhance positive edge slope or both positive and negative edge slope, where m and are a local mean and standard deviation inside a 2K+1 by 2K+1 window, d is the distance of the point (i, j) from the center of the window (K+1, K+1), and INT [x] returns the nearest integer to x.

2. The method of claim 1, wherein said step of processing to get the pixel values from the data samples is performed at each of a plurality of points where radial lines cut the horizontal grid lines after: determining a plurality of radial lines in said rectangular co-ordinate system; and determining a plurality of rectangular grids comprising vertical and horizontal grid lines in said rectangular co-ordinate system.

3. The method of claim 1 wherein said step of processing to get the pixel values from the data samples at each of said raster grid points comprises performing, for each one of said raster grid points: receiving said plurality of points where said plurality of radial lines cut said plurality of horizontal grid lines; evaluating a plurality of nearest points from said plurality of points where said plurality of radial lines cut said plurality of horizontal grid lines with respect to said one raster grid point; assigning sample values to said evaluated nearest points; and imposing speckle reduction techniques using a single scale spatial filter technique to compute a one of said pixel values for said one raster grid point.

4. The method of claim 1, wherein said single scale spatial filter technique comprises at least one of a linear filter technique or a non-linear filter technique.

5. The method of claim 1, wherein said step for applying said high pass filter technique to enhance both positive and negative edge slope comprises: (i) determining the weight co-efficient for each of the pixel values; (ii) evaluating a weighted median of the pixel values within a window using the weight co-efficient adapted to obtain the positive edge slopes; (iii) controlling the sharpness in the positive edge slope directions by adjusting a control parameter .sub.1 to yield a first image; (iv) inverting the pixel values for the window followed by the step (ii) obtain the negative edge slopes; (v) controlling the sharpness in the negative edge slope directions by adjusting control parameter .sub.2 to yield a second image; and (vi) combining the images obtained from step (iii) and step (v).

6. An apparatus for improving speckle reduction in an ultrasound imaging system, said apparatus comprising: a processor; a computer-readable medium, having stored thereon a plurality of instructions for causing the processor to perform the steps of: receiving data samples as an input, the data samples comprising image signals with noises from a logarithmic amplifier; and processing said image signals by simultaneously performing scan conversion and speckle reduction so as to get pixel values from said data samples and to perform speckle reduction to provide a speckle filtered output image; and wherein said pixel values at raster grid points in a rectangular coordinate system are determined using speckle reduction by means of an improved speckle reduction filter implemented along with scan conversion; wherein said step of processing of said image signals to get the pixel values from the data samples is performed at each of a point where radial lines cut the horizontal grid lines, comprising: evaluating a plurality of nearest points on said radial lines with respect to said cut point where a one of said plurality of radial line cuts a one of said horizontal grid line; assigning sample values to said nearest points, said sample values lying substantially around said cut point; and imposing said speckle reduction by means of a single scale spatial filter to compute a one of said pixel values for said cut point; wherein imposing said speckle reduction comprises applying a high pass filter technique with edge enhancement to enhance at least one of positive edge slope or both positive and negative edge slope; wherein said high pass filter technique has positive and negative weight co-efficient determined by: $w\left(i,j\right)=\mathrm{INT}\left[w\left(K+1,K+1\right)-d\frac{20*{\log }_{10}\left(1+m+\right)}{{\log }_{10}\left(m\right)}\right]$ adapted to enhance positive edge slope or both positive and negative edge slope, where m and are a local mean and standard deviation inside a 2K+1 by 2K+1 window, d is the distance of the point (i, j) from the center of the window (K+1, K+1), and INT [x] returns the nearest integer to x.

7. The apparatus of claim 6, wherein said processing to get the pixel values from the data samples is performed at each of a plurality of points where radial lines cut the horizontal grid lines after: determining a plurality of radial lines in said rectangular co-ordinate system; and determining a plurality of rectangular grids comprising vertical and horizontal grid lines in said rectangular co-ordinate system.

8. The apparatus of claim 6 wherein said step of processing to get the pixel values from the data samples at each of said raster grid points comprises performing, for each one of said raster grid points: receiving said plurality of points where said plurality of radial lines cut said plurality of horizontal grid lines; evaluating a plurality of nearest points from said plurality of points where said plurality of radial lines cut said plurality of horizontal grid lines with respect to said one raster grid point; assigning sample values to said evaluated nearest points; and imposing speckle reduction techniques using a single scale spatial filter technique to compute a one of said pixel values for said one raster grid point.

9. The apparatus of claim 6, wherein said single scale spatial filter technique comprises at least one of a linear filter technique or a non-linear filter technique.

10. The apparatus of claim 1, wherein said applying said high pass filter technique to enhance both positive and negative edge slope further comprises steps of: determining the weight co-efficient for each of the pixel values; (ii) evaluating a weighted median of the pixel values within a window using the weight co-efficient to obtain the positive edge slopes; (iii) controlling the sharpness in the positive edge slope directions by adjusting a control parameter .sub.1 to yield a first image; (iv) inverting the pixel values for the window followed by the step (ii) obtain the negative edge slopes; (v) controlling the sharpness in the negative edge slope directions by adjusting control parameter .sub.2 to yield a second image; and (vi) combining the images obtained from step (iii) and step (v).

Description

BRIEF DESCRIPTION OF THE ACCOMPANYING DRAWING

(1) Other features as well as the advantages of the invention will be clear from the following description.

(2) In the appended drawing:

(3) FIG. 1a illustrates schematic Block diagram of B-mode ultrasound imaging system.

(4) FIG. 1b illustrates schematic block diagram of the proposed speckle reduction technique where scan conversion and speckle reduction is performed simultaneously.

(5) FIG. 2 illustrates scan-conversion geometry.

(6) FIG. 3 illustrates geometry of first stage computation

(7) FIG. 4 illustrates pixel geometry for raster grid point computation.

(8) FIG. 5 illustrates reconstructed phantom images in different stages for different filtering methods.

(9) FIG. 6 illustrates plot of Quality metrics of different methods for comparison of performance

(10) FIG. 7 illustrates SR reconstructed images for different techniques.

(11) FIG. 8 illustrates Plot of Quality metrics of different speckle reduction methods with SR reconstruction.

(12) FIG. 9(a), (b) illustrates Final output images by applying median filtering technique on the ultrasound simulated phantom image at different stages.

(13) FIG. 10 illustrates Image of the noisy scan data.

(14) FIG. 11 illustrates Scan converted noisy image after only scan conversion (without filtering).

(15) FIG. 12 illustrates High-pass adaptive weighted median filtering with edge enhancement

DETAILED DESCRIPTION OF THE ACCOMPANYING DRAWINGS

(16) In the following detailed description, reference is made to the accompanying drawings that form a part hereof, and illustrate the best mode presently contemplated for carrying out the invention. The invention is described in reference to specific embodiment and such description should not be considered to a limitation of the present invention. However, such description should not be considered as any limitation of scope of the present mechanism. The structure of the system thus conceived is susceptible of numerous modifications and variations, all the details may furthermore be replaced with elements having technical equivalence. In practice the materials and dimensions may be any according to the requirements, which will still be comprised within its true spirit.

(17) FIG. 1a discloses the simplified schematic block diagram of a typical diagnostic conventional B-mode ultrasound imaging system. The speckle reduction filter is employed here after log compression of the demodulated output. Interpolation is then performed on the filtered log compressed signal for scan conversion and the signal is prepared for display after some post processing tasks. The speckle reduction techniques are applied on envelope detected raw scan-data, log compressed data at the preprocessing stage before scan conversion or scan converted data at post-processing stage.

(18) FIG. 1b is the simplified block diagram of the proposed new paradigm of the speckle reduction technique. In the new technique, all the blocks perform same operations as in the case of old conventional technique except the preprocessing, post-processing and the scan conversion block. Here the speckle reduction is shifted from the preprocessing or post-processing block to scan conversion block since speckle reduction is performed simultaneously with scan conversion.

(19) FIG. 2.: In the present technique/method, the speckle reduction scan conversion method is employed simultaneously avoiding the conventional interpolation. A few so-called single scale spatial speckle-reduction filtering methods (linear and nonlinear such as Lee, Kuan, Median) are chosen to test the performance of the improved method/technique. The method for speckle reduction scan-conversion is described with the help of a diagram of scan-conversion geometry as in FIG. 2. Ultrasound data samples obtained from the logarithm amplifier are placed on rectangular raster along radial lines. A few sample points are placed in FIG. 2 as solid triangular points for the ease of illustration. Now, for scan conversion it needs to be found the pixel value on the rectangular grids from the available data. To perform this, the inventors have first found out the pixel value at the points where radial lines cut the horizontal grid lines. For example, three successive radial lines (Line j1, Line j, and Line j+1) are considered. The pixel value at point P is found out, where the radial line, Line j cuts the horizontal grid line. The three nearest points around P along the Line j is found. These points are D, E and F. Suppose E is the nearest point of P along Line j. Hence, next two nearest points are F and D, respectively. The nearest sample value as s(nearest, j) is assigned. Consequently, the other two points D and F as s(nearest 1, j) and s(nearest +1, j) respectively is also assigned. In a similar the other six points (A, B, C, G, H and I), three from each Line j1 and Line j+1 is found out. These six points are: s(nearest1, j1), s(nearest, j1), s(nearest +1, j1), s(nearest 1, j+1), s(nearest, j+1), s(nearest +1, j+1) respectively. Around the point P we get nine sample values as a local window from which the pixel value at P is calculated.

(20) To calculate the pixel value at P, different single scale spatial filter (linear or nonlinear) based so-called popular speckle reduction algorithm is imposed. For illustration, the Lee filter technique is used. Lee filter technique is already discussed in the literature survey. The parameter k of Lee filter can be determined from the variance and the mean of the local window. Then the pixel value p at the point P can be calculate as,
p=s+k[s(nearest, j)s](26)
where s is the average value of the pixels within the local window. s is calculated by adding all the pixel values within the window which contains the pixels designated by s(nearest, j), s(nearest+1, j), . . . etc. as described and dividing the result by the number of the pixels within the windows.

(21) k is different for different linear filtering techniques (such as Lee, Kuan etc.) and it can be calculated from the statistics of the local window. For nonlinear filters such as median, weighted median or adaptive weighted median filters k is not defined. For these filters, the median value is calculated from the pixel values of the local window using simple median calculation technique or weighted median calculation technique and it is mentioned earlier section of this document.

(22) After computation of all the pixel values at the points where radial lines cut the horizontal gridlines, the geometry will be converted as shown in FIG. 3 below: The computed points are denoted as solid circles. P.sub.1, P.sub.2, P.sub.3 . . . are such points.

(23) Now, with available of the points P1, P.sub.2, P.sub.3 . . . the raster grid points of the raster scan is computed.

(24) FIG. 4 discloses the procedure of computation of the pixel values at the raster grid points. In FIG. 4, pixel values at the points P.sub.1, P.sub.2, P.sub.3 . . . are already calculated in the first stage. In the next step, the pixel values at the raster grid points Q.sub.1, Q.sub.2, Q.sub.3 . . . etc is computed. In the example the raster grid point Q.sub.5 in the i.sup.th row and j.sup.th column is considered. Also the pixel values at the points P.sub.k, k=1, 2, 3 . . . are represented with two index variables is considered. Three nearest points of Q.sub.5 along i.sup.th row are determined. P.sub.7, P.sub.6 and P.sub.8 are such three nearest points. P.sub.7 is the nearest one and P.sub.6 and 8 are the next two successive nearest points. The pixel value of P.sub.7 as p(i, nearest) is assigned. Then other two nearest points can be assigned as p(i, nearest1) and p(i, nearest+1), respectively. Similarly, the three nearest points from previous row other three from next row is found out. For finding three nearest points from the previous row i.e. (i1).sup.th row, the grid point Q.sub.2 of the same column and (i1).sup.th row and search three nearest points around Q.sub.2 along the row is found out. These points are assigned as p(i1, nearest 1), p(i1, nearest11) and p(i1, nearest1+1). And in a similar way, three nearest points from next row i.e. (i+1).sup.th row is found out. The points as p(i+1, nearest 2), p(i+1, nearest 21) and p(i+1, nearest 2+1) are assigned. Finally, the pixel value at the grid point Q.sub.5 can be computed from these nine points as
q=p+k[p(i, nearest)p](27)
where p average value of the pixels within the window.

(25) The average value p is calculated by adding the pixel values within the windows which are designated by p(i, nearest), p(i, nearest+1), . . . etc. and dividing the result by the number of pixels within the local window.

(26) Different single scale spatial filtering techniques are applied within this improved method where filtering and scan conversion is done simultaneously.

(27) The pixel values at the grid points in rectangular co-ordinate system are calculated using filtering technique from the neighbor pixel values. It fulfills the requirement of scan conversion, and at the same time, it gives the speckle filtered output image. Hence interpolation stage in the scan conversion process is avoided.

(28) In the geometrical portrait, Q points are the grid points in the rectangular co-ordinate system. To generate a speckle filtered ultrasound image that is displayed in the conventional video monitor which supports rectangular co-ordinate system and therefore, first the pixel values at the grid points Q is calculated. To evaluate the values at the pixel points Q, in the present technique the pixel values at the points P is calculated as an intermediate stage using filtering algorithm avoiding interpolation. After calculating the pixel values at the point P the pixel values at Q is calculated by using the pixel values at the points P applying filtering algorithm again.

(29) In the present invention, to evaluate the pixel value of a grid point on the rectangular raster, the nearest pixel value from the raw data is used and the noise reduction algorithm on that nearest pixel value is applied. Hence scan conversion and speckle reduction are performed simultaneously.

(30) FIG. 5 discloses Simulation results: The reconstructed image of a simulated phantom for each case. The comparisons of quality metrics for the evaluation of the quality of the reconstructed images are shown in FIG. 6 according to table 1.

(31) It is observed that the quality of the reconstructed image is the best if filtering and scan conversion are performed simultaneously. This present improved technique also reduces the functional blocks of the ultrasound imaging systems. It is verified that it is also valid in case of super-resolution. SR reconstructed images by the above methods are shown in FIG. 7 and the performance in terms of quality metrics is shown in FIG. 8. Filtering operation for all the reconstructed images is done with 33 window.

(32) The present adaptive weighted median filtering technique is also applied to noiseless signal to verify whether the present filter provides a considerable good output or not.

(33) FIG. 9(a) shows the original noiseless image and FIG. 9(b) shows the output of the present filter.

(34) FIG. 12 demonstrate the high-pass filtering with positive slope and negative slope edge enhancement. This algorithm increases the sharpness and the contrast of the image. The parameters .sub.1 and .sub.2 are the controls parameters which controls the sharpness in the positive and negative slope directions as per requirement.

(35) Since it is high pass in nature, it is able to preserve image details, which is most important criteria in the medical ultrasound image. The negative values of the weights make the filter high pass in nature. The filter preserves both positive and negative-slope edges of the image. The sharpness control factor controls sharpness and the positive and negative-slope edge enhancing capability.

(36) The quality metrics of the output of the present filter with noiseless image is given in table 1.

(37) TABLE-US-00001 TABLE 1 Quality metrics of the output image of the present filter when input image is noise free image MSE 71.0386 PSNR 29.6159 Q 0.9986

(38) The quality metric also confirms that the filter does not hamper much the noise free image.

(39) MSE: Mean Square Error

(40) PSNR: peak signal to noise ratio

(41) Q: Universal quality index

(42) The different techniques are compared with the help of quality metrics. The value of the quality metrics imply that the invention provides the better quality of the image and the reconstructed image is closer to the original image than the other methods.

(43) The invention explores a new paradigm where the old popular speckle reduction techniques can be used to obtain better quality of output image. The same equations for Lee, Kuan or Median filtering techniques are used. But they must be used before scan conversion or during scan conversion instead of using them after scan conversion. The MSE, PSNR, Q shows better results when the conventional filtering techniques are used during scan conversion.

(44) Further it is found that the improved AWM based speckle reduction technique which gives better quality of image if it is applied in old popular speckle reduction techniques like Lee, Kuan or Median filtering algorithm.

(45) It is observed that though speckle reduction before scan conversion and during scan conversion performs better than the speckle reduction after scan conversion, the best technique is the speckle reduction during scan conversion. This is because it gives the best noise reduction capability and decrease in computational burden.

(46) Expectedly, the method for speckle reduction in an ultrasound imaging system and system for speckle reduction disclosed herein will find many useful applications in diverse technical fields. Examples of such applications include not only: ultrasound imaging for medical diagnostic and non-destructive evaluation but also SAR imaging, PET/SPECT and other modalities, etc.

(47) It is understood that the systems and methods of the illustrative embodiments may be modified in a variety of ways which will become readily apparent to those skilled in the art, and having the benefit of the novel teachings disclosed herein. All such modifications and variations of the illustrative embodiments thereof shall be deemed to be within the scope and spirit of the present invention as defined by the claims to invention appended hereto.