METHOD FOR DENOISING TIME SERIES IMAGES OF A MOVED STRUCTURE FOR MEDICAL DEVICES

Abstract

Embodiments provide a method for denoising time series images of a moved structure for a medical device. A movement detector detects the moved structure. The movement detector obtains a measurement of the similarity of two images that each represent the same section of the moved structure. The two images originate from two different time series images. A ratio between spatial and temporal denoising is defined for the section as a function of the measurement of the similarity.

Claims

1. A method for denoising a plurality of time series images of a moved structure for a medical device, the method comprising: detecting, by a movement detector, the moved structure; obtaining, by the movement detector, a measurement of a similarity of two images, each of the two images representing a same section of the moved structure, wherein the two images originate from two different time series images of the plurality of time series images; and defining a ratio between a spatial and a temporal denoising for the section as a function of the measurement of the similarity.

2. The method of claim 1, further comprising: segmenting the two images by at least band passes before the measurement of the similarity is obtained.

3. The method of claim 2, wherein the segmenting comprises a Laplace segmentation or an à trous segmentation.

4. The method of claim 3, wherein the segmenting uses an edge preserving kernel.

5. The method of claim 2, wherein obtaining the measurement of the similarity comprises using a comparison of respective bandpass signals of one of the band passes for the two images.

6. The method of claim 5, wherein the ratio between spatial and temporal denoising is applied according to the following formula:
BP_filtered(T).sub.x,y=(1−α.sub.x,y).Math.BP_denoise_spatial(T).sub.x,y+α.sub.x,y.Math.BP_filtered(T−n).sub.x+dx,y+dy wherein BP_filtered(T).sub.x,y is a denoised bandpass signal of a first of time series signals, wherein α.sub.x,y is a value between 0 and 1 and represents the ratio between spatial and temporal denoising, wherein the value 1 denotes solely temporal denoising and the value 0 denotes solely spatial denoising; wherein BP_denoise_spatial(T).sub.x,y represents a spatially denoised bandpass signal; and wherein BP_filtered(T−n).sub.x+dx,y+dy represents a filtered past bandpass signal of one image of the plurality of time series images preceding another image of the plurality of time series images.

7. The method of claim 1, wherein the measurement of the similarity is obtained using a comparison of respective output signals of at least two low passes for the two images.

8. The method of claim 6, wherein the value α is calculatable using a cross fading function, and wherein the cross fading function is a linear function, an arctangent function, or a linear and arctangent function.

9. The method of claim 1, wherein the medical device is an X-ray device or an infrared recording device.

10. An apparatus for denoising time series images of a moved structure for a medical device, the apparatus comprising: a denoising device; and a movement detector configured to: detect the moved structure; obtain a measurement of a similarity of two images, each of the two images representing a same section of the moved structure, wherein the two images originate from two different time series images; and define a ratio between a spatial denoising and a temporal denoising for the denoising device as a function of the measurement.

11. The apparatus of claim 10, the movement detector is further configured to segment the two images by band pass before the measurement of the similarity is obtained.

12. The apparatus of claim 11, wherein segmenting comprises a Laplace segmentation or an à trous segmentation.

13. The apparatus of claim 12, wherein the segmenting uses an edge preserving kernel.

14. The apparatus of claim 11, wherein the movement detector is configured to obtain the measurement of the similarity by a comparison of a respective bandpass signal of one of a band passes for the two images.

15. The apparatus of claim 14, wherein the ratio between spatial and temporal denoising is applied by the denoising device according to the following formula:
BP_filtered(T).sub.x,y=(1−α.sub.x,y).Math.BP_denoise_spatial(T).sub.x,y+α.sub.x,y.Math.BP_filtered(T−n).sub.x+dx,y+dy wherein BP_filtered(T).sub.x,y is a denoised bandpass signal of a first of the plurality of time series signals, wherein α.sub.x,y is a value between 0 and 1 and represents the ratio between spatial and temporal denoising, wherein the value 1 denotes solely temporal denoising and the value 0 denotes solely spatial denoising; wherein BP_denoise_spatial(T).sub.x,y is a spatially denoised bandpass signal; and wherein BP_filtered(T−n).sub.x+dx,y+dy is a filtered past bandpass signal of one of the second time series images preceding the first.

16. The apparatus of claim 10, wherein the movement detector is configured to obtain the measurement of the similarity using a comparison of respective output signals of at least two low passes for the two images.

17. The apparatus of claim 15, wherein the value α may be calculated by the movement device using a cross fading function; wherein the cross fading function is a linear function, an arctangent function, or a linear function and an arctangent function.

18. The apparatus of claim 10, wherein the medical device is an X-ray device or an infrared recording device.

19. The method of claim 3, wherein obtaining the measurement of the similarity comprises using a comparison of a respective bandpass signal of one of a band passes for the two images.

20. The method of claim 4, wherein obtaining the measurement of the similarity comprises using a comparison of a respective bandpass signal of one of a band passes for the two images.

Description

BRIEF DESCRIPTION OF THE FIGURES

[0021] FIG. 1 depicts a block diagram of the method according to an embodiment.

[0022] FIG. 2 depicts a detailed section of FIG. 1 according to an embodiment.

[0023] FIG. 3 depicts a function profile for the value α according to an embodiment.

DETAILED DESCRIPTION

[0024] FIG. 1 depicts a block diagram of how a denoised time series image 3, 4, 5 is achieved from an image of a moved structure 9. The method corresponds to a corresponding apparatus having denoising device and movement detector. The moved structure 9 may be segmented into bandpass signals 8. The segmentation may use, for example, a Laplace segmentation or by the application of an à trous segmentation by an adaptive edge-preserving kernel. A bandpass segmentation may include a mean freedom with bandpass signals 8. Segmentation in a segmentation plane 10 into bandpass signals 8 has the property that noise, for example, in X-ray systems, is Poisson distributed and convoluted with the system modulation transfer function. The noise, but also other signals, may be analyzed and processed in different spatial frequencies. Band passes may be used. In FIG. 1 the method is carried out, for example, on only one bandpass signal 8. The method may additionally or alternatively also be applied to the other bandpass signal 8. As an alternative to segmentation into bandpass signals 8, segmentation into low-pass signals may be used. After segmentation of the moved structure 9 into different bandpass signals 8, pre-denoising is performed by a local denoiser 6 in a spatial denoising plane 17. The spatially denoised bandpass signal 12 may be produced, for example, by a measurement of the local variance in the bandpass signal 8, the comparison of this measurement with the expected variance in the noise and the reduction in local coefficients in the bandpass signal.

[0025] The spatial or spatially denoised or smoothed bandpass signal 12 is filtered further by the method 1 for movement compensation and/or movement detection. FIG. 2 depicts the method 1 in a block diagram. A movement detector 2 detects the moved structure 9 from the spatial denoised bandpass signal 12, by comparing the bandpass signal 12 spatially denoised by an image signal 14, that originates from an image buffer, at time T=T0 (with T as a whole number) with a past bandpass signal 11a at time T=T0−1, that originates from an image signal 15 at time T=T0−1 from the image buffer. Displacement vectors are produced that may be used for a decision variable 21 (cf. FIG. 3). A measurement of the similarity between the spatial denoised bandpass signal 12 at time T=T0 and the past bandpass signal 11a at time T=T0−1 may be calculated, for example, by the sum of absolute differences. In the example where the sum of the absolute differences is used as a measurement of the similarity, the displacement vector may then be given by a minimum of corresponding measurements for the similarity of the image points of a vicinity of the respective image point. The displacement vectors may also firstly be sought on low-frequency bandpass signals 8 to increase quality.

[0026] Once the movement detector 2 has calculated a degree α 23 of admixing of spatial denoising 6 in relation to temporal denoising 7 using the decision variable 21, a denoised bandpass signal 11 is output at time T=T0 14. The value of α 23 may be between 0 or 1. The value of α is described in FIG. 3, for example, as a linear transition.

[0027] In the embodiment depicted in FIG. 1 further denoised past bandpass signals 11b from the past T=T0−n 16 are also used in the method 1 in relation to the denoised past bandpass signal 11a. The use of a plurality of denoised past bandpass signals 11b provides an increase in the accuracy of the movement detection and simultaneously improves denoising.

[0028] To obtain a denoised time series image 3, the individual, partially denoised bandpass signals 11 are combined to form an image. The signals may be combined in a spatiotemporal denoising plane 18.

[0029] FIG. 3 depicts an example of how using the similarity of the sections of the moved structure 9, the movement detector chooses a decision variable 21 and calculates the ratio between spatial denoising 6 and temporal denoising 7, e.g. the value α 23. The value α 23 between 0 and 1 is depicted, in particular on the abscissa, where 0 is the origin. The decision variable 21 may be found on the ordinate, and is produced, for example, from the sums of the absolute differences of a displacement of an image point of the spatial denoised bandpass signal 12 and the past bandpass signal 11a. The profile of α 23 is depicted, by way of example, as a graph 22. The profile is linear. A low value may be found with a slight difference between the spatial denoised bandpass signal 12 and the past bandpass signal 11a, whereby α lies at 1. The degree of admixing of spatial denoising 6 is very low, for example, 0. Purely temporal denoising occurs. With a high absolute difference, α is very low, in particular (close to) 0, and high temporal denoising 7 is not applied. The decision variable 21 depicts the sum of the absolute differences. Other clearances are possible. Furthermore, other functions, that are not linear, for example an arctangent function, may be used as a graph 22. The transition point 19 (e.g. threshold value), that depicts, for example, an α 23 of 0.5, that is used for differentiation, above which clearance spatial denoising 6 or temporal denoising 7 may be applied, is used for a more detailed description of the graph 22. The steepness of the transition 20 describes the size range in which a mixing of the temporal denoiser 7 and the spatial denoiser 6 occurs and in which α 23 is not equal to 1 and not equal to 0.

[0030] It is to be understood that the elements and features recited in the appended claims may be combined in different ways to produce new claims that likewise fall within the scope of the present invention. Thus, whereas the dependent claims appended below depend from only a single independent or dependent claim, it is to be understood that these dependent claims may, alternatively, be made to depend in the alternative from any preceding or following claim, whether independent or dependent, and that such new combinations are to be understood as forming a part of the present specification.

[0031] While the present invention has been described above by reference to various embodiments, it may be understood that many changes and modifications may be made to the described embodiments. It is therefore intended that the foregoing description be regarded as illustrative rather than limiting, and that it be understood that all equivalents and/or combinations of embodiments are intended to be included in this description.