CN103471811A - Method for measuring detective quantum efficiency and system thereof - Google Patents

Abstract

The invention discloses a method for measuring detective quantum efficiency and a system for measuring detective quantum efficiency. The method comprises the following steps: a digital detector is used to receive exposure of a target to be detected so as to obtain an image; an edge of the target to be detected is extracted from the image, edge spread function ESF is obtained, the ESF undergoes derivation to obtain line spread function LSF, Fourier transform of the LSF is carried out, and normalization of absolute value of the Fourier transform result at zero frequency is carried out to obtain modulation transfer function MTF; noise power spectrum NPS is calculated; and according to the MTF and the NPS, detective quantum efficiency DQE is calculated. The technical scheme provided by the invention is stable and is easy to implement.

Description

Measure method and the system of detective quantum efficiency

Technical field

The present invention relates to the x-ray imaging technical field, particularly a kind of method and a kind of system of measuring detective quantum efficiency of measuring detective quantum efficiency.

Background technology

Detective quantum efficiency (DQE) is a key index, and is widely used in assessing the performance performance of digital detector in arm holder for radiographic imaging apparatus.The calculating of DQE depends on the measurement of modulation transfer function (MTF) and noise power spectrum (NPS).Although International Electrotechnical Commission (IEC) and some are organized and issued about how arranging and implement the specific standards of DQE test macro, still have some can discuss and improved place.

The Chinese patent application CN102204828A that the application people is University Of Tianjin discloses a kind of modulation transfer function accurate measurement method of digital X-ray imaging system, comprise: place the imitative body of fat, gather multiple image, and the noise power spectrum of computing system overall noise; Place for covering the stereotype of detector, gather the electronic noise power spectrum of multiple image computing system; Place line to card, gather multiple image, by each image superposed average, obtain line to the card the average image; By line, the card the average image is obtained to intensity profile map, calculate line to mtf value corresponding to card integer space frequency; Utilize linear regression method, calculating noise power spectrum and modulation transfer function related coefficient η, obtain the MTF change curve under the imaging system spatial frequency range.

Summary of the invention

In view of this, the present invention proposes a kind of method of measurement detective quantum efficiency more stable, that more easily implement.The present invention also proposes a kind of system of measuring detective quantum efficiency, so as more stable, more easily implement to measure detective quantum efficiency.

According to one embodiment of the invention, a kind of method of measuring detective quantum efficiency DQE is provided, the method comprises:

Use digital detector to receive the exposure to a target to be checked, obtain image;

Extract an edge of described target to be checked in described image, obtain edge spread function ESF, edge spread function ESF differentiate obtains line spread function LSF, described line spread function LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtain modulation transfer function;

Calculating noise power spectrum NPS;

According to described modulation transfer function, described noise power spectrum NPS, calculate described detective quantum efficiency DQE.

Alternatively, the step of extracting the edge of described target to be checked comprises: select the region of interest in image; By image binaryzation; Detect the edge of target to be checked, and extract this edge.

Preferably, by after image binaryzation, further eliminate the noise of region of interest.

Alternatively, utilize Ou Su Otsu method by image binaryzation, and/or, utilize Hough Hough method to detect the edge of target to be checked.

Alternatively, the step of calculating noise power spectrum NPS comprises:

The Region Segmentation of a calculating noise power spectrum foursquare region of interest that pluralizes will be used in image;

Linearization process is carried out in the emerging zone of described sense, and calculate the mean value of these region of interest;

Described mean value is carried out to two-dimension fourier transform, and calculate two-dimentional noise power spectrum;

From described two-dimentional noise power spectrum, intercepting obtains the one dimension noise power spectrum.

According to another embodiment of the present invention, a kind of system of measuring detective quantum efficiency DQE is provided, this system comprises:

An image acquiring device, receive the formed image of the exposure of a target to be checked for obtaining by digital detector;

A modulation transfer function calculation element, for extract an edge of described target to be checked at described image, obtain edge spread function ESF, edge spread function ESF differentiate obtains line spread function LSF, described line spread function LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtain modulation transfer function;

A noise power spectrum calculation element, for calculating noise power spectrum NPS;

A detective quantum efficiency calculation element, for according to described modulation transfer function, described noise power spectrum NPS, calculate described detective quantum efficiency DQE.

Alternatively, described modulation transfer function calculation element comprises an edge detection unit, and this edge detection unit, for selecting the region of interest of image, by image binaryzation, is eliminated the noise of region of interest, detects the edge of target to be checked, and extracts this edge.

Alternatively, described noise power spectrum calculation element comprises:

A cutting unit, for being used for image the Region Segmentation of a calculating noise power spectrum foursquare region of interest that pluralizes;

A linearizer, for carrying out linearization process to the emerging zone of described sense;

A mean value unit, for calculating the mean value of these region of interest;

A Fourier transform unit, for carrying out two-dimension fourier transform to described mean value;

An one dimension noise power spectrum computing unit, for calculating two-dimentional noise power spectrum, from described two-dimentional noise power spectrum, intercepting obtains the one dimension noise power spectrum.

From such scheme, can find out, because the method that the OTSU method the present invention proposes based on threshold value combines with morphological operation is calculated MTF, this method can make to have robustness and accuracy for the knife-edge detection of MTF calculation services.In addition, in the NPS calculation stages, proposed one fast image remove linear method, further improved the speed of measuring DQE.So technical scheme of the present invention is more stable, more easily implement.

The accompanying drawing explanation

Below will the person of ordinary skill in the art is more clear that above-mentioned and other feature and advantage of the present invention by with reference to accompanying drawing, describing the preferred embodiments of the present invention in detail, in accompanying drawing:

Fig. 1 is the target to be checked of using in one embodiment of the invention.

The layout schematic diagram that Fig. 2 is imaging system in one embodiment of the invention.

Fig. 3 is for calculating according to an embodiment of the invention the method for MTF.

Fig. 4 is for calculating according to an embodiment of the invention the method for NPS.

Wherein Reference numeral is as follows:

10: target 20 to be checked: tungsten plate 30: zone to be analyzed

40:X radiographic source 50:X beam 60: the surface of detector

S210 – S250: the step of method

S510 – S560: the step of method

Embodiment

For making the purpose, technical solutions and advantages of the present invention clearer, the present invention is described in more detail by the following examples.

For the DQE (u, v) of dependent Frequency, be defined as follows shown in the formula (1) of face:

$\mathrm{DQE}(u,v)={\mathrm{MTF}}^2(u,v)\frac{W_{\mathrm{in}}(u,v)}{W_{\mathrm{out}}(u,v)}{\mathrm{MTF}}^2(u,v)\frac{K_a{\mathrm{SNR}}_{\mathrm{in}}^2}{W_{\mathrm{out}}(u,v)}\left(1\right)$

In formula (1), DQE (u, v) is Quantum detection efficiency, the modulation transfer function that MTF (u, v) is the digital X-ray imaging device; K _afor the air kerma recorded, unit is uGy; for square signal to noise ratio (S/N ratio) of every air kerma, unit is 1/ (mm ²uGy); W _in(u, v) and W _out(u, v) is respectively power input and output power; U and v are horizontal and vertical frequency.

If test condition and X-ray beam quality are determined, the W calculated for different computing method so _in(u, v) is identical.Propose many diverse ways in the past few decades and calculated MTF and W _out.

In general, can calculate mtf value by the Fourier transform of a line spread function (Line Spread Function, LSF).The method of calculating LSF has slit method and knife-edge method usually.In the slit method, LSF is that the slit detection angled by the pel array of the image with obtaining obtained.Then according to the different distance of slit, drawing out signal intensity, the combination of measurements of all row forms LSF.In knife-edge method, the derivative of knife-edge spread function (EdgeSpread Function, ESF) also can produce line spread function.

The noise power spectrum NPS(f of dependent Frequency) be defined in spatial frequency domain the variable of a random signal with frequency change, express for convenient two dimension, usually also remember that NPS (f) is for the W in formula (1) _out(u, v).

The W of dependent Frequency _out(u, v) is defined as follows:

$W_{\mathrm{out}}(u_n,v_k)=\underset{N_x,N_y,M\→\∞}{\mathrm{Lim}}\frac{N_x{N}_y}{M.{\mathrm{\Δ}}_x{\mathrm{\Δ}}_y}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\left|\underset{i=1}{\overset{N_x}{\mathrm{\Σ}}}\underset{j=1}{\overset{N_y}{\mathrm{\Σ}}}(I(x_i,y_j)-S(x_i,y_j))\exp (-2\mathrm{\πi}(u_n{x}_i+v_k{y}_j))\right|}^2---\left(2\right)$

In formula, Δ _xΔ _yfor the product of pel spacing in the horizontal and vertical directions; N _xn _yproduct for number of pixels in area-of-interest in the horizontal and vertical directions; The number that M is area-of-interest; I(x _i, y _j) be at location of pixels (x _i, y _j) image intensity located; S(x _i, y _j) be I (x _i, y _j) mean value, Lim is the function of asking the limit, ∑ be the summation function.

Next describe the embodiment of the present invention and measure the method for detective quantum efficiency (DQE).

The embodiment of the present invention can be used target to be checked as shown in Figure 1.As shown in Figure 1, target 10 to be checked comprises a tungsten plate 20.Also marked to be analyzed regional 30 in Fig. 1.

Fig. 2 has provided the layout of imaging system in one embodiment of the invention.As shown in Figure 2, this imaging system comprises target to be checked as shown in Figure 1 of an X-ray tube 40,10, a digital detector (surface 60 that has shown this detector in Fig. 2).Wherein, X-ray tube 40 sends X-ray beam 50, and this X-ray beam 50, through after target 10 to be checked, is detected device and receives.Wherein, target 10 to be checked is placed on the place ahead of digital detector at a certain angle.

The method of measuring according to an embodiment of the invention DQE comprises:

Step S100, used imaging system as shown in Figure 2, uses X-ray tube to be exposed to a detection means 10, utilizes digital detector to receive X ray, thereby obtain the raw data of image.

Step S200, in raw data, an edge of extraction target to be checked, obtain edge spread function ESF.

Step S300, edge spread function ESF differentiate obtains line spread function LSF.

Step S400, carry out Fourier transform to line spread function LSF, therefore, through level and smooth ESF, with obtaining LSF, is distinguishing.LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtain modulation transfer function.

Step S500, calculating noise power spectrum NPS.

Step S600, according to modulation transfer function, noise power spectrum NPS, utilize formula (1) to calculate detective quantum efficiency DQE.

In said method, can first calculate MTF, also can first calculate NPS, can also calculate MTF and NPS simultaneously, embodiments of the invention do not require this.

In above-mentioned steps S200, one embodiment of the invention proposes the method for a powerful extraction object edge to be checked.As shown in Figure 3, this method of extracting object edge to be checked comprises:

Step S210, according to the raw data of image, obtain image.

Step S220, in image, select region of interest.

Step S230, by the image binaryzation of region of interest.Preferably use and utilize Ou Su (Otsu) method to carry out binaryzation.

Afterwards, application Otsu threshold method carries out binary conversion treatment to image.The Otsu method is to utilize the image threshold based on shape automatically to generate histogrammic method.The supposition of this algorithm will be carried out the pixel that the image of Threshold segmentation comprises two kinds, class internal variance minimum when then calculating best threshold value and making to separate this two classifications.Carry out morphological operation for reducing exceptional value on bianry image.This step is for eliminating noise, and the data of therefore processing are stable in knife-edge detects.

Step S240, further can eliminate the noise of region of interest after binaryzation.

Step S250, utilize Hough (Hough) method to detect the edge of blade, thereby extract the edge of target to be checked.Therefore extract the calculating that the knife-edge profile carries out MTF from raw data according to the Hough testing result.

The process of calculating NPS in the step of the calculating NPS of above-mentioned steps S500 can comprise the steps:

Step S510, will be used for the Region Segmentation of a calculating noise power spectrum NPS foursquare region of interest that pluralizes.

The part area of the digital X ray detector of analyzing for NPS is divided into square region.For example, each ROI that calculates the single sample of NPS is 256 * 256 pixel sizes.

Step S520, carry out linearization process to feeling emerging zone.

Step S530, calculate the mean value of these region of interest.

Step S540, carry out two-dimension fourier transform to mean value, and for example, fast Fourier changes (FFT).

Step S550, calculate two-dimentional noise power spectrum.

Step 560, from two-dimentional noise power spectrum, intercepting obtains one dimension noise power spectrum NPS.

Following formula is for calculating two-dimentional NPS.

$W_{\mathrm{out}}(u_n,v_k)=\frac{{\mathrm{\Δ}}_x{\mathrm{\Δ}}_y}{M\·256\·256}\underset{m=1}{\overset{M}{\mathrm{\Σ}}}{\left|\underset{i=1}{\overset{256}{\mathrm{\Σ}}}\underset{j=1}{\overset{256}{\mathrm{\Σ}}}(I(x_i,y_j)-S(x_i,y_j))\exp (-2\mathrm{\πi}(u_n{x}_i+v_k{y}_j))\right|}^2---\left(3\right)$

In formula, Δ _xΔ _yfor the product of pel spacing in the horizontal and vertical directions; The number that M is area-of-interest; I(x _i, y _j) be at location of pixels (x _i, y _j) image intensity located; S(x _i, y _j) be I (x _i, y _j) mean value.

Thereby intercept two-dimentional NPS along the spatial frequency domain axle and obtain one dimension NPS.For the complicacy of calculating, applied the average of removing the image of linear trend in the method proposed under the prerequisite of not losing the NPS precision simultaneously.

Another embodiment of the present invention has also proposed the system of a kind of DQE of measurement, and this system comprises: an image acquiring device, a modulation transfer function calculation element, a noise power spectrum calculation element and a detective quantum efficiency calculation element.

Wherein, image acquiring device receives the formed image of the exposure of a target to be checked for obtaining by digital detector.

The modulation transfer function calculation element, for extract an edge of target to be checked at image, obtains ESF, and to ESF, differentiate obtains LSF, and LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtains MTF.

The noise power spectrum calculation element is for calculating NPS.

The detective quantum efficiency calculation element, for according to MTF, NPS, calculates DQE.

Preferably, the modulation transfer function calculation element can comprise an edge detection unit, and this edge detection unit, for selecting the region of interest of image, by image binaryzation, is eliminated the noise of region of interest, detects the edge of target to be checked, and extracts this edge.

On the other hand, the noise power spectrum calculation element preferably can comprise: a cutting unit, a linearizer, a mean value unit, a Fourier transform unit and an one dimension noise power spectrum computing unit.

Wherein, cutting unit is for being used for image the Region Segmentation of a calculating noise power spectrum foursquare region of interest that pluralizes.

Linearizer is for carrying out linearization process to the emerging zone of described sense.

The mean value unit is for calculating the mean value of these region of interest.

Fourier transform unit for example, for carrying out two-dimension fourier transform, Fast Fourier Transform (FFT) to described mean value.

One dimension noise power spectrum computing unit is for calculating two-dimentional noise power spectrum, and from described two-dimentional noise power spectrum, intercepting obtains the one dimension noise power spectrum.

The invention discloses a kind of method and a kind of system of measuring detective quantum efficiency of measuring detective quantum efficiency.Described method comprises: use digital detector to receive the exposure to a target to be checked, obtain image; Extract an edge of described target to be checked in described image, obtain edge spread function ESF, edge spread function ESF differentiate obtains line spread function LSF, described line spread function LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtain modulation transfer function; Calculating noise power spectrum NPS; According to described modulation transfer function, described noise power spectrum NPS, calculate described detective quantum efficiency DQE.Characteristics stable, easy enforcement that technical scheme of the present invention has.

The foregoing is only preferred embodiment of the present invention, in order to limit the present invention, within the spirit and principles in the present invention not all, any modification of doing, be equal to replacement, improvement etc., within all should being included in protection scope of the present invention.

Claims (8)

1. a method of measuring detective quantum efficiency DQE, the method comprises:

Use digital detector to receive the exposure to a target to be checked, obtain image;

Extract an edge of described target to be checked in described image, obtain edge spread function ESF, edge spread function ESF differentiate obtains line spread function LSF, described line spread function LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtain modulation transfer function;

Calculating noise power spectrum NPS;

According to described modulation transfer function, described noise power spectrum NPS, calculate described detective quantum efficiency DQE.

2. method according to claim 1, is characterized in that, the step of extracting the edge of described target to be checked comprises:

Select the region of interest in image;

By image binaryzation;

Detect the edge of target to be checked, and extract this edge.

3. method according to claim 2, is characterized in that, by after image binaryzation, further eliminates the noise of region of interest.

4. method according to claim 2, is characterized in that, utilizes Ou Su Otsu method by image binaryzation, and/or, utilize Hough Hough method to detect the edge of target to be checked.

5. method according to claim 1, is characterized in that, the step of calculating noise power spectrum NPS comprises:

The Region Segmentation of a calculating noise power spectrum foursquare region of interest that pluralizes will be used in image;

Linearization process is carried out in the emerging zone of described sense, and calculate the mean value of these region of interest;

Described mean value is carried out to two-dimension fourier transform, and calculate two-dimentional noise power spectrum;

From described two-dimentional noise power spectrum, intercepting obtains the one dimension noise power spectrum.

6. a system of measuring detective quantum efficiency DQE, this system comprises:

An image acquiring device, receive the formed image of the exposure of a target to be checked for obtaining by digital detector;

A modulation transfer function calculation element, for extract an edge of described target to be checked at described image, obtain edge spread function ESF, edge spread function ESF differentiate obtains line spread function LSF, described line spread function LSF is carried out to Fourier transform, in zero frequency, the absolute value of Fourier transform result is carried out to normalization, obtain modulation transfer function;

A noise power spectrum calculation element, for calculating noise power spectrum NPS;

A detective quantum efficiency calculation element, for according to described modulation transfer function, described noise power spectrum NPS, calculate described detective quantum efficiency DQE.

7. system according to claim 6, it is characterized in that, described modulation transfer function calculation element comprises an edge detection unit, this edge detection unit is for selecting the region of interest of image, by image binaryzation, eliminate the noise of region of interest, detect the edge of target to be checked, and extract this edge.

8. system according to claim 6, is characterized in that, described noise power spectrum calculation element comprises:

A cutting unit, for being used for image the Region Segmentation of a calculating noise power spectrum foursquare region of interest that pluralizes;

A linearizer, for carrying out linearization process to the emerging zone of described sense;

A mean value unit, for calculating the mean value of these region of interest;

A Fourier transform unit, for carrying out two-dimension fourier transform to described mean value;

An one dimension noise power spectrum computing unit, for calculating two-dimentional noise power spectrum, from described two-dimentional noise power spectrum, intercepting obtains the one dimension noise power spectrum.

Images