WO2018218441A1

WO2018218441A1 - Spectrogram analysis method, apparatus and device, and computer readable storage medium

Info

Publication number: WO2018218441A1
Application number: PCT/CN2017/086370
Authority: WO
Inventors: 马忠伟; 胡鹏; 冯磊; 赵新; 李高隆
Original assignee: 北京悦琦创通科技有限公司
Priority date: 2017-05-27
Filing date: 2017-05-27
Publication date: 2018-12-06
Also published as: CN107979987A; CN107979987B

Abstract

A spectrogram analysis method (100), apparatus (600) and device (700), and a computer readable storage medium. The spectrogram analysis method comprises: acquiring a Doppler spectrogram to be analyzed (S110); recognizing, from the Doppler spectrogram to be analyzed, spectrogram features to be analyzed (S120); and analyzing the spectrogram features to be analyzed by means of a trained classifier to obtain abnormal information, the abnormal information comprising abnormal spectral information about whether abnormal spectrum features exist, and the abnormal spectrum features comprising one or more of the following: stealing, vortex, turbulence and short dash (S130). The spectrogram analysis method (100), apparatus (600) and device (700) and the computer readable storage medium automatically recognize the abnormal spectrum features on the basis of the spectrogram features of the transcranial Doppler spectrogram, such that labor costs can be saved, the efficiency and accuracy of characteristic recognition can be improved, thereby facilitating the improvement of disease diagnosis correctness, and having great application value and wide market prospect.

Description

Spectral analysis method, device and device, and computer readable storage medium

Technical field

The present invention relates to the field of ultrasound Doppler blood flow detection technology, and more particularly to a spectrum analysis method, apparatus and apparatus, and computer readable storage medium.

Background technique

Transcranial Doppler (TCD) blood flow analysis is a method for evaluating the physiological characteristics of different blood flow states through non-invasive examination. Ultrasound Transcranial Doppler Flow Analyzer is a customized ultrasound device designed for transcranial ultrasound examination. Ultrasound transcranial Doppler blood flow analyzer is a product that appeared in the early 1980s to diagnose cerebrovascular disease and help to check for cerebral vascular narrowing, obstruction, poor blood flow or cerebral hemorrhage. The application of Doppler spectrum analysis technology can provide information such as blood flow waveform, blood flow velocity (peak velocity, average velocity) and blood flow disorder for clinical diagnosis, which is very important for the early detection of cerebrovascular diseases.

Ultrasound transcranial Doppler flowmetry uses an in vitro ultrasound probe to transmit ultrasound to the cerebral vessels through the gap or "window" of the skull. A Doppler effect (Doppler shift) is generated between the ultrasonic wave and the blood flow, and the reflected ultrasonic wave returns to the probe, and the data is processed by the processor in the analyzer to obtain corresponding information. Using the Doppler effect, the ultrasound transcranial Doppler flow analyzer can detect information such as blood flow velocity in blood vessels.

The existing transcranial Doppler device mainly generates a transcranial Doppler spectrogram (may be referred to as "spectrum"), and the operator analyzes the spectrum to identify abnormal spectral features in the spectrum (eg, stealing) Blood, eddy currents, turbulence, short horizontal lines, etc.), and then give a diagnosis. First of all, this will increase the workload of the operator, and the operator needs to manually identify the features one by one. Secondly, the operator's feature recognition is greatly affected by the mental state, and leakage recognition may occur when fatigue or mood is low. Thirdly, the clinical diagnosis problem is very complicated, and the technical requirements of the operator are high, which requires more clinical training. Therefore, there is a need for a method of automatically analyzing a spectrum.

Summary of the invention

The present invention has been made in consideration of the above problems. The present invention provides a spectrogram analysis method, apparatus and apparatus, and computer readable storage medium.

According to an aspect of the invention, a spectrogram analysis method is provided. The method comprises: obtaining a Doppler spectrum map to be analyzed; identifying a spectrum feature to be analyzed from the Doppler spectrum to be analyzed; and analyzing the spectrum characteristics by using the trained classifier to obtain abnormal information, wherein The abnormal information includes abnormal spectral information about whether an abnormal spectral feature exists, and the abnormal spectral features include one or more of stealing blood, eddy current, turbulence, and dash.

Illustratively, the abnormal spectral features include blood stealing, and the spectral features to be analyzed include waveform features related to the waveform of the Doppler spectrogram to be analyzed, and identifying the features to be analyzed from the Doppler spectrogram to be analyzed includes: The maximum envelope is identified in the Doppler spectrogram to be analyzed; the cardiac cycle is divided according to the change rule of the identified maximum envelope; and the change rule according to the identified maximum envelope in any cardiac cycle is determined to be analyzed. Whether the waveform of the Platz spectrum is reversed to obtain waveform characteristics.

Illustratively, identifying the to-be-analyzed spectral feature from the Doppler spectrogram to be analyzed further includes: encapsulating the identified maximum envelope before dividing the cardiac cycle according to the changed variation of the identified maximum envelope smooth.

Illustratively, the anomalous spectral features include eddy currents and/or dashes, and the spectral features to be analyzed include energy distribution features associated with the energy distribution of the Doppler spectrogram to be analyzed, identified from the Doppler spectrogram to be analyzed The characteristics of the spectrum to be analyzed include: averaging the energy of the blood flow signal in the Doppler spectrogram to be analyzed to obtain an effective average value; and finding a target region whose energy is higher than the effective average in the Doppler spectrum to be analyzed The morphology of the target region is analyzed to obtain morphological features; the symmetry of the target region relative to the baseline of the Doppler spectrogram to be analyzed is analyzed to obtain symmetry features; wherein the energy distribution features include morphological features and symmetry features.

Illustratively, in the case that the abnormal spectral features include eddy currents, identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed further includes: identifying a minimum envelope from the Doppler spectral image to be analyzed to obtain Whether the frequency window has a frequency window feature, wherein the spectrum feature to be analyzed further includes a frequency window feature.

Illustratively, the abnormal spectral features include turbulence, and the spectral features to be analyzed include flow velocity energy characteristics related to the flow velocity and energy relationship of the Doppler spectrogram to be analyzed, and the spectral characteristics to be analyzed are identified from the Doppler spectrogram to be analyzed. The method comprises: determining a correspondence relationship between blood flow velocity and energy according to a Doppler spectrum to be analyzed; performing curve fitting with blood flow velocity and energy as variables; and calculating a slope of the fitted curve; wherein the flow velocity energy characteristic includes a slope .

Illustratively, identifying the spectral feature to be analyzed from the Doppler spectrogram to be analyzed further comprises: identifying a minimum envelope from the Doppler spectrogram to be analyzed to obtain a frequency window characteristic regarding whether the frequency window exists, wherein The spectral features to be analyzed also include frequency window features.

Illustratively, before acquiring the Doppler spectrogram to be analyzed, the method further comprises: obtaining an initial Doppler spectrogram; and if the initial Doppler spectrogram is based on two or more blood vessels superimposed together The Doppler signal is generated, and the initial Doppler spectrum is decomposed into two or more sub-Doppler spectrograms corresponding to two or more blood vessels one by one; obtaining the Doppler spectrum to be analyzed The graph includes determining one of two or more sub-Doppler spectrograms as a Doppler spectrum to be analyzed.

Illustratively, before acquiring the Doppler spectrogram to be analyzed, the method further comprises: obtaining an initial Doppler spectrogram; if the initial Doppler spectrogram is based on two or more blood vessels superimposed together The indication information generated by the Pull signal and the blood flow direction of two or more blood vessels in the initial Doppler spectrum is the same, and the indication information for instructing the operator to re-implement the transcranial Doppler examination is output.

Illustratively, before acquiring the Doppler spectrogram to be analyzed, the method further comprises: acquiring an initial Doppler spectrogram; if the initial Doppler spectrogram is generated based on Doppler signals of the two blood vessels superimposed together And in the initial Doppler spectrogram, the blood flow direction of the two blood vessels is reversed and there is an abnormality in the portion of the spectrum corresponding to the at least one blood vessel in the initial Doppler spectrogram, and the output is used to instruct the operator to re-implement the transcranial Instructions for Doppler inspection.

Illustratively, before identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed, the method further comprises: performing noise reduction on the Doppler spectrogram to be analyzed.

Illustratively, performing noise reduction on the Doppler spectrogram to be analyzed includes: extracting, by filtering, a blood flow signal whose energy in the Doppler spectrogram to be analyzed is higher than a preset energy threshold, to obtain a to-be-analyzed after noise reduction Doppler spectrogram.

Illustratively, performing noise reduction on the Doppler spectrogram to be analyzed includes: averaging the overall energy of the Doppler spectrogram to be analyzed to obtain a mean value of the spectrum; and calculating a lower spectrum in the Doppler spectrum to be analyzed The mean and variance of the energy of the mean; the adaptive energy threshold is set according to the mean and the variance; and the blood flow signal of the energy to be analyzed in the Doppler spectrogram to be higher than the adaptive energy threshold is extracted by filtering to obtain a drop Doppler spectrogram to be analyzed after noise.

Illustratively, before the spectral feature to be analyzed is identified from the Doppler spectrogram to be analyzed, the method further comprises: performing interference filtering on the Doppler spectrogram to be analyzed by filtering to remove the Doppler spectrum to be analyzed. Interference signal.

Illustratively, the abnormality information further includes: status information as to whether the blood flow is normal as a whole and/or direction information as to whether the blood flow direction is reversed.

Illustratively, the method further includes: obtaining a positive sample Doppler spectrum map and a negative sample Doppler spectrum map, wherein the positive sample Doppler spectrum map includes anomalous spectral features included in the Doppler spectrum map to be analyzed Type-consistent specific anomalous spectral features, negative sample Doppler spectrograms do not contain specific anomalous spectral features; positive sample spectral features are identified from positive sample Doppler spectrograms, and negative is identified from negative sample Doppler spectrograms Sample spectral features; and training the classifier model using positive sample spectral features and negative sample spectral features to obtain a trained classifier.

According to another aspect of the present invention, a spectrum analysis apparatus is provided, comprising: a spectrum acquisition module to be analyzed for acquiring a Doppler spectrum to be analyzed; and a feature recognition module to be analyzed for using a Doppler spectrum to be analyzed. Identifying spectral features to be analyzed; and analyzing a module for analyzing the spectral features to be analyzed by the trained classifier to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether the abnormal spectral features exist, Abnormal spectral features include one or more of stealing blood, eddy currents, turbulence, and dashes.

According to another aspect of the present invention, a spectrum analysis apparatus is provided, comprising: a memory for storing a program; a processor for running a program; wherein, when the program is run in the processor, the method is configured to: Analyze the Doppler spectrum; identify the features of the spectrum to be analyzed from the Doppler spectrum to be analyzed; and analyze the characteristics of the analyzed spectrum using the trained classifier to obtain abnormal information, wherein the abnormal information includes Abnormal spectrum information for the presence of spectral features, including one or more of stealing blood, eddy currents, turbulence, and dashes.

Illustratively, the spectroscopic analysis device is a device independent of the ultrasonic transcranial Doppler blood flow analyzer for acquiring a Doppler signal to obtain a Doppler spectrogram to be analyzed, or the spectroscopic analysis device is an ultrasound transcranial Doppler blood flow analyzer.

According to another aspect of the present invention, a computer readable storage medium is provided. A program is stored on a storage medium, and the program is used at runtime to perform the following steps: acquiring a Doppler spectrum to be analyzed; and analyzing a Doppler spectrum from the spectrum to be analyzed. Identifying spectral features to be analyzed; and analyzing the spectral features to be analyzed using the trained classifier to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether abnormal spectral features exist, and the abnormal spectral features include stealing blood One or more of eddy currents, turbulence, and dashes.

The method, device and device and computer readable storage medium according to embodiments of the present invention automatically identify abnormal spectral features based on spectral features of a transcranial Doppler spectrogram, which can save labor costs and improve the efficiency and accuracy of feature recognition. Sex, which helps to improve the correctness of disease diagnosis, has great application value and broad market prospects.

DRAWINGS

The above as well as other objects, features and advantages of the present invention will become more apparent from the embodiments of the invention. The drawings are intended to provide a further understanding of the embodiments of the invention, In the figures, the same reference numerals generally refer to the same parts or steps.

1 shows a schematic flow chart of a spectrogram analysis method according to an embodiment of the present invention;

2 shows a schematic diagram of a transcranial Doppler spectrogram according to an example of the present invention;

Figure 3 is a schematic view showing the flow of blood in a blood vessel;

Figures 4a-4i show schematic diagrams of transcranial Doppler spectrograms in different blood flow states;

Figure 5a shows a schematic diagram of an initial Doppler spectrum of Doppler signal generation based on two blood vessels superimposed together, according to one example;

Figure 5b shows a schematic diagram of an initial Doppler spectrum of Doppler signal generation based on two blood vessels superimposed together, according to another example;

Figure 6 shows a schematic block diagram of a spectroscopic analysis apparatus in accordance with one embodiment of the present invention;

Figure 7 shows a schematic block diagram of a spectroscopic analysis device in accordance with one embodiment of the present invention.

detailed description

In order to make the objects, the technical solutions and the advantages of the present invention more apparent, the exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. It is apparent that the described embodiments are only a part of the embodiments of the present invention, and are not to be construed as limiting the embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention described herein without departing from the scope of the invention are intended to fall within the scope of the invention.

In order to solve the above problems, an embodiment of the present invention provides a spectrum analysis method and apparatus, and a storage medium. According to the spectrum analysis method of the embodiment of the invention, abnormal spectrum features (such as stealing blood, eddy current, turbulence, dash, etc.) can be automatically identified, which can save manpower and improve the accuracy of feature recognition.

Hereinafter, a spectroscopic analysis method according to an embodiment of the present invention will be described with reference to FIG. FIG. 1 shows a schematic flow chart of a spectral analysis method 100 in accordance with one embodiment of the present invention. As shown in FIG. 1, the spectrogram analysis method 100 includes the following steps.

In step S110, a Doppler spectrum map to be analyzed is acquired.

The Doppler spectrogram to be analyzed refers to a transcranial Doppler spectrogram that can be obtained by any suitable, existing or future ultrasound transcranial Doppler flow analyzer. Each time the probe of the ultrasound transcranial Doppler flow analyzer emits an ultrasound, it is equivalent to performing a sampling on the time axis. The sampling rate of the sample is Fs. The original data collected by the ultrasound transcranial Doppler blood flow analyzer is a one-dimensional Doppler signal f(t) that changes with time. The Doppler signal is essentially a non-stationary signal, which mainly reflects the frequency domain characteristics and time variation. The frequency will also change. The signal f(t) can be processed using a short time Fourier transform method. Short-time Fourier transform is a commonly used signal processing method. Its idea is to choose a time-frequency localized window function g(t), assuming that the window function g(t) is stationary in a short time interval. For smoothness, the moving window function g(t) is such that f(t)g(t) is a stationary signal for different finite time widths, thereby calculating the power spectrum at different times.

The engineering power spectrum is usually calculated by the fast Fourier transform. According to the sampling theorem, the highest analyzable frequency is half the sampling rate. Assuming Fs is the sampling rate, the range of power spectrum energy analysis is -Fs/2~Fs/2, which is the analyzable range of Doppler frequency offset. When there is no Doppler signal, that is, when there is no blood flow, the energy will concentrate around 0, and the 0 frequency offset is usually called the baseline. It can be obtained from the Doppler formula that the frequency offset is proportional to the flow velocity (ie, the blood flow velocity), and the larger the frequency offset is, the larger the flow velocity is. Since the velocity of sound waves in the soft tissue of the human body is relatively constant, about 1540 m / s, the blood flow velocity can be quantitatively calculated based on the frequency offset. Due to the backscattering principle of ultrasound, the more red blood cells in a certain flow rate interval, the stronger the Doppler signal at this frequency offset, and the greater the value of a point on the power spectrum.

Exemplarily, each power spectrum can be converted into a display line, and the power spectrum can be converted into an image that is easily recognized by the human eye through pseudo color mapping, and the brightness distribution on the image can reflect the distribution of blood flow velocity in the blood vessel. For example, by combining multiple power spectra over a period of time, a three-dimensional image can be generated (luminance is also considered as a dimension), that is, a Doppler spectrogram to be analyzed, wherein the abscissa represents time and the ordinate represents frequency offset or Flow rate, brightness indicates energy intensity. Fig. 2 shows a schematic diagram of a transcranial Doppler spectrogram according to an example of the present invention, in which the ordinate represents a frequency offset.

The maximum velocity (or frequency offset) at each time point in the transcranial Doppler spectrogram is a curve called the maximum envelope, which is an important feature. In the case where the ordinate indicates the flow rate, the highest point of the maximum envelope is the systolic velocity (Vs), the lowest point of the maximum envelope is the end-diastolic flow velocity (Vd), and the average value of the maximum envelope is the average flow velocity ( Vm). The pulsitility index (PI) is an important parameter to measure the waveform characteristics, PI = (Vs - Vd) / Vm. Generally speaking, the larger the PI value, the greater the vascular resistance; on the contrary, the smaller the vascular resistance. For intracranial vessels, the resistance is usually small and the PI value is in the range of 0.5 to 1.0.

At step S120, the spectral features to be analyzed are identified from the Doppler spectrogram to be analyzed.

The spectral feature to be analyzed is a feature related to one of parameters such as waveform, flow velocity, energy, or a combination of a plurality of parameters in the Doppler spectrogram to be analyzed. Illustratively, the spectral features to be analyzed may include and Waveform-related waveform characteristics of the Doppler spectrogram to be analyzed, energy distribution characteristics related to the energy distribution of the Doppler spectrogram to be analyzed, and flow velocity energy characteristics related to the flow velocity and energy relationship of the Doppler spectrogram to be analyzed One or more of them.

In step S130, the spectral feature to be analyzed is analyzed by using a trained classifier to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether an abnormal spectral feature exists, and the abnormal spectral features include blood stealing and eddy current. One or more of turbulence, turbulence, and dash.

It should be noted that the abnormal information is not limited to the above abnormal spectrum information. When there are abnormal spectral features such as stealing blood, turbulence, eddy current and short horizontal lines in the blood vessels, the corresponding transcranial Doppler spectrogram and the transcranial Doppler spectrum in normal blood flow state (referred to as "normal spectrum") ) is different, and the existence of anomalous spectral features can be identified by the unique spectral features brought about by each anomalous spectral feature. The different manifestations of transcranial Doppler spectrograms in the presence of abnormal spectral features are described below.

The transcranial Doppler spectrogram reflects the blood flow in the blood vessels. Blood flow is initiated by the heart. After the heart is ejected, the blood flow velocity in the artery rises rapidly. When the aortic valve is closed, the blood flow velocity will decrease significantly due to the loss of the main ejection force. However, since the aorta stores a large amount of blood and continues to provide a certain pressure, there is still some blood flow during the systole.

Normally, blood moves in the blood vessels in the form of laminar flow. Laminar flow is a flow state of a fluid that acts as a laminar flow with its mass moving smoothly along a direction parallel to the tube axis. The flow rate of the fluid is greatest at the center of the tube and is minimal near the vessel wall. The ratio of the average flow rate of the fluid within the tube to the maximum flow rate is equal to or substantially equal to 0.5. Figure 3 shows a schematic diagram of blood flowing in a blood vessel. The blood flow state at the A position shown in Fig. 3 is a laminar flow.

Cerebral artery stenosis is a common clinical condition, and a common cause is plaque in the blood vessel wall. When the plaque increases to a certain extent, it will occupy most of the space inside the blood vessel, causing a fundamental change in blood flow. When blood flows to the area where the plaque is located (see the B position shown in Fig. 3), since the flow space is rapidly reduced, it encounters a large resistance, which requires more pressure to pass the plaque, so there is only a small amount Part of the blood can pass through the plaque, which can be defined as high resistance and low flow.

A small portion of the blood that can pass through the plaque, its flow rate is fast due to an increase in pressure, see the blood flow state at the C position shown in FIG.

When blood flows through the plaque, the blood will rapidly diffuse in all directions due to the rapid increase of the blood vessel space, see the blood flow state at the D position shown in FIG. Since there is a low pressure zone behind the plaque, it is easy to produce blood rotation, which is called vortex (Vortex). If the degree of arterial stenosis is high and the blood flow rate is too fast, it is easy to produce turbulence. The basic feature of turbulence is flow The randomness of body microphore movement. The turbulent micelles not only have lateral pulsations, but also have a reverse motion with respect to the total motion of the fluid, so the trajectory of the fluid micelles is extremely disordered and changes rapidly with time.

When the blood flows forward, it will gradually return to laminar flow. Since the blood flowing through the plaque will decrease, the resistance is relatively low, showing the characteristics of low resistance and low flow. See the blood at the E position shown in Fig. 3. Flow status.

The normal spectrum corresponding to normal blood flow has specific spectral features. When there are some abnormal conditions in the blood flow (such as abnormal spectral features such as eddy current, turbulence, dash or blood stealing), transcranial Doppler spectrum The graph may show some special spectral features.

Figures 4a-4i show schematic diagrams of transcranial Doppler spectrograms in different blood flow states. Figure 4a is a schematic illustration of a transcranial Doppler spectrum in a normal blood flow state. For the normal spectrum, the pulsation index is generally between 0.5 and 1.0. Generally, during the systolic phase, the blood flow will be concentrated in the high flow velocity region, and the signal in the low velocity region is weak or even missing. Generally, the missing portion of the signal is called the frequency window. (as in the triangle area in Figure 4a). Figure 4b is a schematic diagram of a typical transcranial Doppler spectrum with a high resistance waveform, characterized by a small difference between the systolic phase of the systolic phase and the normal spectrum, but the diastolic flow rate is lower, the Vd value is lower, and the pulsation index is lower. Big. Figure 4c is a schematic diagram of a typical transcranial Doppler spectrum with a low-resistance waveform, characterized by a small difference between the systolic phase of the systolic phase and the normal spectrum, but with a higher diastolic flow velocity, a higher Vd value, and a higher pulsatile index. small. Figure 4d is a schematic diagram of a transcranial Doppler spectrogram in which a waveform change occurs. Figure 4d is a typical waveform change. As shown in Figure 4d, the systolic waveform is reversed (ie, the blood flows in the opposite direction) and there is no significant change in diastolic phase. Figure 4e is a schematic diagram of a typical transcranial Doppler spectrum containing eddy currents. The base spectrum is similar to the normal spectrum, but during systole, there is a symmetric (or roughly symmetrical) low-velocity blood flow signal near the baseline. Strong. Figure 4f is a schematic diagram of a typical transcranial Doppler spectrogram containing turbulence. The turbulence does not have a distinct boundary like eddy currents, but rather has a larger energy near the baseline, and the farther away from the baseline, the lower the energy. Figure 4g is a transcranial Doppler spectrum with short horizontal lines. The dash is also caused by arterial stenosis, but its characteristics are different from those of eddy currents, and its flow velocity is relatively stable. Figure 4h is a schematic diagram of a transcranial Doppler spectrogram containing a variety of abnormal spectral features (turbulence and dashes). Figure 4i is a schematic illustration of a typical transcranial Doppler spectrogram containing interfering signals.

It can be seen from Fig. 4a-4i that if there are abnormal spectral features, the transcranial Doppler spectrogram will show some spectral features different from the normal spectral features, which may be waveform features, energy distribution features, Flow rate energy characteristics, etc. Therefore, these unique spectral features (ie, spectral features to be analyzed) can be identified from the Doppler spectrogram to be analyzed, and whether abnormal spectral features exist is determined based on the identified spectral features.

In the prior art, the operator manually analyzes the transcranial Doppler spectrogram to identify abnormal spectral features, and the present invention proposes a method for automatically identifying abnormal spectral features. According to the spectrum analysis method of the embodiment of the present invention, the abnormal spectrum feature is automatically identified based on the spectral features of the transcranial Doppler spectrogram, which can save labor cost, and can improve the efficiency and accuracy of feature recognition, thereby contributing to improvement The correctness of disease diagnosis has great application value and broad market prospects.

According to an embodiment of the invention, the abnormal spectral feature comprises stealing blood, the spectral feature to be analyzed comprises a waveform feature related to the waveform of the Doppler spectrogram to be analyzed, and step S120 comprises: identifying the maximum value from the Doppler spectrogram to be analyzed Envelope; dividing the cardiac cycle according to the change rule of the identified maximum envelope; and determining whether the waveform of the Doppler spectrogram to be analyzed is reversed according to the variation rule of the identified maximum envelope in any cardiac cycle Get waveform features.

Stealing blood can be identified based on waveform changes in the Doppler spectrogram to be analyzed. The blood flow in the human body is generally changed according to the cardiac cycle. When the heart is ejecting blood, the blood flow velocity in the blood vessel is accelerated; when the heart is in the diastolic phase, the blood flow velocity in the blood vessel is slowed down. Therefore, data analysis can be performed in units of cardiac cycles, and data analysis is performed once per cardiac cycle. Therefore, waveform changes can also take into account data within a certain cardiac cycle. The cardiac cycle can be divided according to the variation law of the maximum envelope. Those skilled in the art can understand the manner of dividing the cardiac cycle, and this article does not describe it.

The following describes the identification of blood stealing by taking the Subclavian Steal Syndrome (SSS) as an example. When one side of the subclavian artery is occluded at the proximal end, it will cause insufficient blood supply to the ipsilateral vertebral artery, and the blood of the contralateral vertebral artery will flow to cause the vertebral artery blood flow to reverse. It should be noted that the blood flow direction of the normal vertebral artery is away from the probe, and if it flows in the opposite direction, it will flow toward the probe. The reverse flow pattern varies depending on the degree of arterial stenosis. For moderate stenosis, only the reverse phase of the maximum systolic velocity phase is observed; for severe stenosis, the entire systolic phase may be reversed while the diastolic phase remains normal; for complete occlusion, the entire blood flow direction is reversed.

The direction of the waveform in the transcranial Doppler spectrogram can represent the direction of blood flow. Referring back to Fig. 4d, the waveform of the systolic phase is reversed, and the waveform of the diastolic phase is normal, indicating that the blood flow is reverse flow during the systolic phase, and the blood flow is positively flowing during the diastolic phase.

The maximum envelope of the wave can represent the waveform. Compared to the normal spectrum, if a waveform change occurs in the transcranial Doppler spectrogram, the waveform change can be easily perceived from the direction of the maximum envelope in the transcranial Doppler spectrogram. Referring back to Figure 4d again, the maximum value of the systolic period is below the baseline, indicating that the waveform is reversed, and the maximum envelope of the diastolic phase is normal, indicating that the waveform is positive. Therefore, the maximum envelope of the Doppler spectrogram to be analyzed can be identified first, and the waveform direction is determined according to the maximum envelope. Positive or negative (corresponding to whether the blood flow direction is positive or negative).

Waveform features can be obtained by the above analysis for the maximum envelope. Waveform features can include whether the waveform is reversed. If the waveform does not reverse, it can be considered that there is no blood stealing, and if the waveform is reversed, it can be considered that there is blood stealing.

According to the embodiment of the present invention, before the cardiac cycle is divided according to the change rule of the identified maximum envelope, step S120 may further include: performing envelope smoothing on the identified maximum envelope.

The maximum envelope can be smoothed based on a priori knowledge that the blood flow velocity is continuously changing to improve the correctness of the feature recognition. For example, the upper limit of human heart rate is generally 300 times / minute, so the frequency is about 5 Hz, and after considering various harmonic components, low-pass filtering with a cutoff frequency of 35 Hz can generally retain the main components, while suppressing noise as much as possible. . In addition, if more details need to be retained, low-pass filtering can be performed with a cutoff frequency of 75 Hz, but due to the presence of power frequency interference, it is necessary to consider increasing the 50 Hz notch. The lower limit of human heart rate is generally 30 times/min, so the frequency is about 0.5 Hz. It can be considered that the signal below this frequency is of no value. Therefore, a high-pass filter with a cutoff frequency of, for example, 0.5 Hz can be designed to filter the low-frequency components of no value. .

In addition, median filtering is also an effective envelope smoothing method. Median filtering is a nonlinear signal processing technique based on the theory of sorting statistics that can effectively suppress noise. The basic principle of median filtering is to replace the value of a point in a digital image or a sequence of numbers with the median of the values of the points in the neighborhood of the point, so that the value of the point is close to the surrounding pixel values, thereby eliminating isolated noise. point.

Envelope smoothing is beneficial to improve the classification of cardiac cycle, the extraction of blood flow characteristic parameters (such as Vs, Vd, PI, etc.) and the accuracy of spectral feature recognition, which is beneficial to more accurately identify abnormal spectral features.

According to an embodiment of the invention, the abnormal spectral features include eddy currents and/or short horizontal lines, and the spectral features to be analyzed include energy distribution features related to energy distribution conditions of the Doppler spectrogram to be analyzed, and step S120 may include: The energy of the blood flow signal in the Pule spectrogram is averaged to obtain an effective average value; the target region whose energy is higher than the effective average is found in the Doppler spectrum to be analyzed; the morphology of the target region is analyzed to obtain the morphology Feature; analyzing the symmetry of the target region relative to the baseline of the Doppler spectrogram to be analyzed to obtain a symmetry feature; wherein the energy distribution feature includes morphological features and symmetry features.

Eddy currents and dashes can be identified based on the energy distribution of the Doppler spectrogram to be analyzed.

The eddy current is the rapid spin of the blood. Because of the circular motion, the Doppler angle continuously changes, and the flow velocity distribution is relatively uniform, concentrated near the baseline, and both positive and negative (return to Figure 4e). Eddy currents are generally formed during systole and when blood flows at high speed through the plaque. The eddy current signal is usually superimposed on the base spectrum. The base spectrum is close to the normal spectrum, so the transcranial Doppler spectrum containing the eddy current can be understood as the superposition of the eddy current signal and the normal spectrum. Normal spectra are usually weaker near the baseline during systole, and may even be frequency windows. Transcranial Doppler spectrograms containing eddy currents have symmetrical (or roughly symmetrical) low-velocity blood flow signals near the baseline during systole, where the energy is stronger. Therefore, the transcranial Doppler spectrogram containing eddy currents is significantly different in energy distribution from the normal spectrum, and the existence of eddy currents can be identified based on this difference.

Since the eddy current is concentrated near the baseline and the energy is much larger than the energy of the normal spectrum, it is possible to focus on finding the region with higher energy in the Doppler spectrum to be analyzed, for example, in the Doppler spectrum to be analyzed, the energy is higher than a certain value. The target area of the threshold. If the target area exists and its shape and symmetry satisfy the vortex shape and symmetry requirements, then eddy currents can be considered to exist. For example, if the shape of the target area is elliptical or circular as shown in Figure 4e, and the target area is symmetric (or substantially symmetrical) with respect to the baseline of the Doppler spectrogram to be analyzed, then eddy currents may be considered to exist. The threshold for dividing the target area may be set as needed. Illustratively, the threshold may be an average of the energy of the blood flow signal in the Doppler spectrum to be analyzed. The morphological features described herein can include the shape of the target area. Morphological features may also include the location of the target region in the cardiac cycle (eg, whether it is during systole, during diastolic or systolic and diastolic phases). Illustratively, the morphological feature may be represented by a coordinate representation of the target region in the coordinate system of the Doppler spectrogram to be analyzed (eg, represented by time coordinates and flow velocity coordinates occupied by the target region).

According to an embodiment of the invention, after determining the morphological features and symmetry features of the target region, the morphological features and the symmetry features may be input to the classifier to determine if eddy currents are present. The classifier is trained in advance. During the training process, a large number of known transcranial Doppler spectrograms containing eddy currents can be used as positive samples, and these transcranial Doppler spectrograms can be analyzed to obtain the morphological and symmetry characteristics of the respective target regions. And use the morphological and symmetry features obtained by the analysis to train the classifier.

Similar to the eddy current, the energy of the dash line is also significantly stronger than the normal spectrum, so the dash line can also be used in a similar manner to the eddy current. The dash line is mainly inconsistent with the eddy current. Referring to the vortex shown in Fig. 4e and the dash shown in Fig. 4g, the target area corresponding to the eddy current is continuous in flow rate, and the target area corresponding to the short horizontal line is discontinuous in flow rate, distributed at several constant flow rates. There may be harmonic components at the same time. In addition, short horizontal lines generally appear during systole and occasionally continue to diastole. Therefore, the eddy current and the dash line can be distinguished based on the morphological characteristics of the identified target area.

According to an embodiment of the present invention, in the case that the abnormal spectral feature includes eddy current, step S120 may further include: identifying a minimum value envelope from the Doppler spectrogram to be analyzed to obtain a frequency window characteristic regarding whether the frequency window exists, wherein The spectral features to be analyzed also include frequency window features.

The main significance of the minimum envelope is the identification of the frequency window. Referring back to FIG. 4a, the frequency window is a triangular region in the transcranial Doppler spectrogram. As can be seen from the figure, the edge line of the frequency window is the minimum envelope of the transcranial Doppler spectrogram. Therefore, it is possible to determine whether or not the frequency window exists by identifying the minimum envelope.

The normal spectrum is a frequency window, and if there is a eddy current, there is usually no frequency window. Therefore, the frequency window can be used as an auxiliary judgment basis for the existence of eddy current. The frequency window can improve the recognition accuracy of the eddy current to a certain extent.

According to an embodiment of the invention, the abnormal spectral feature includes turbulence, and the spectral feature to be analyzed includes a flow velocity energy characteristic related to a flow velocity and an energy relationship of the Doppler spectrogram to be analyzed, and step S120 may include: according to the Doppler spectrogram to be analyzed Determining the correspondence between blood flow velocity and energy; performing curve fitting with blood flow velocity and energy as variables; and calculating a slope of the fitted curve; wherein the flow velocity energy characteristic includes a slope.

The flow rate and energy distribution of turbulence are significantly different from eddy currents. When turbulence is present, the blood flow velocity is extremely chaotic and the normal spectrum distribution disappears. Generally speaking, the energy near the baseline is the strongest, but not strictly symmetrical, and then the farther from the baseline, the lower the energy. When turbulent flow is present, the maximum flow rate Vs increases significantly, typically up to 180 cm/sec.

Based on the above characteristics of turbulence, a way can be devised to identify turbulence from the Doppler spectrogram to be analyzed. Illustratively, it may be considered to identify turbulence based on the relationship between blood flow velocity and energy. As described above, the frequency offset is proportional to the blood flow velocity, and the blood flow velocity at each moment can be easily determined based on the Doppler spectrum to be analyzed. It should be noted that the blood flow velocity at each moment is not a single value, but rather a plurality of values distributed over a range. In addition, the Doppler spectrogram to be analyzed has energy information at each time and each frequency offset, and thus, a one-to-one correspondence between blood flow velocity and energy can be obtained. Illustratively, the blood flow velocity can be used as the abscissa, the energy is used as the ordinate to establish a coordinate system, and each point at which the blood flow velocity and energy are coordinates is marked in the coordinate system. Curve fitting can then be performed based on the points marked in the coordinate system and the slope of the fitted curve can be determined.

For the normal spectrum, the curve of blood flow velocity and energy is smoother, the slope is roughly positive, and the energy increases with the increase of blood flow velocity. When turbulence is present, the slope is negative at the initial stage of the fitted curve, ie the energy decreases as the blood flow velocity increases. Therefore, depending on the relationship between blood flow velocity and energy, it is possible to distinguish between turbulent flow and no turbulence.

According to an embodiment of the present invention, step S120 may further include: identifying a minimum value envelope from the Doppler spectrum to be analyzed to obtain a frequency window feature regarding whether the frequency window exists, wherein the spectrum feature to be analyzed further includes a frequency window. feature.

Similar to eddy currents, in the presence of turbulence, there is usually no frequency window. Therefore, the frequency window can also serve as an auxiliary judgment basis for the existence of turbulence. The frequency window can improve the recognition accuracy of turbulence to a certain extent.

According to an embodiment of the present invention, before step S110, the method 100 may further include: acquiring an initial Doppler spectrum map; and if the initial Doppler spectrum map is based on Doppler of two or more blood vessels superimposed together If the signal is generated, the initial Doppler spectrum map is decomposed into two or more sub-Doppler spectrograms corresponding to two or more blood vessels one by one; step S110 may include: determining two or One of more than two sub-Doppler spectrograms is the Doppler spectrogram to be analyzed.

The main blood vessels in the brain include: middle cerebral artery, anterior cerebral artery, posterior cerebral artery, vertebral artery and basilar artery. Since the blood vessels of the neck and the blood vessels of the brain are directly connected, the relevant common carotid artery, internal carotid artery and external carotid artery are also blood vessels that can be examined by an ultrasound transcranial Doppler blood flow analyzer. For each blood vessel, a corresponding Doppler signal can be obtained. If the initial Doppler spectrogram is generated based on a single blood vessel Doppler signal, the initial Doppler spectrogram can be directly used as the subsequent Doppler spectrogram obtained in step S110 for subsequent spectral feature recognition. And analysis steps.

It should be noted that since Doppler sampling has a certain depth, a typical value is about 10 mm, so when the distance between two blood vessels is less than this depth, it may happen that two blood vessels are simultaneously sampled. There are two specific cases of vascular signal overlap, as shown in Figures 5a and 5b. Figure 5a shows a schematic diagram of an initial Doppler spectrogram based on Doppler signal generation of two blood vessels superimposed together, according to one example, and Figure 5b shows two blood vessels based on superimposed together according to another example Schematic diagram of the initial Doppler spectrogram generated by the Doppler signal. Fig. 5a shows an initial Doppler spectrum obtained in the case of reversal of blood flow of two blood vessels, and Fig. 5b shows initial Doppler obtained in the case where blood flows of two blood vessels are in the same direction. Le spectrogram.

Illustratively, in the case where the Doppler signals of the two blood vessels are superimposed to generate an initial Doppler spectrum (ie, the blood flow signals of the two blood vessels in the initial Doppler spectrum are superimposed), first Identify the boundary of the hemorrhagic flow signal and then separately process the portion of the spectrum associated with the blood flow signal. In general, in the portion of the spectrum where the two blood vessel flow rates are consistent, the energy has a superimposed effect, so the spectral energy here is significantly larger than the portion of the spectrum associated with the non-overlapping signal. Non-overlapping signals can be used as noise, and the portion of the spectrum associated with the blood flow signal can be identified by filtering. Subsequently, the portion of the spectrum associated with the blood flow signal is decomposed to obtain a sub-Doppler spectrum corresponding to each blood vessel. Any sub-Doppler spectrogram may be used as a subsequent spectral feature recognition and analysis step as the Doppler spectrogram to be analyzed acquired in step S110. It should be understood that steps can be implemented separately for each sub-Doppler spectrogram S120 and S130, to obtain abnormal information corresponding to each sub-Doppler spectrum map.

When the number of blood vessels from which the Doppler signal is generated by the initial Doppler spectrum is more than two, the same manner can be used for the decomposition, and the description is omitted.

According to an embodiment of the present invention, before step S110, the method 100 may further include: acquiring an initial Doppler spectrum; if the initial Doppler spectrum is based on Doppler of two or more blood vessels superimposed together The signal is generated and the blood flow direction of the two or more blood vessels is the same in the initial Doppler spectrogram, and the indication information for instructing the operator to re-implement the transcranial Doppler examination is output.

According to the embodiment described above, in the case where the blood flow signals of the two blood vessels in the initial Doppler spectrum are superimposed, the blood flow signals of the two blood vessels can be automatically decomposed. For the case where the blood flow signals of the two blood vessels in the initial Doppler spectrogram are inversely superimposed (as shown in Fig. 5a), since the flow velocity is relatively independent, the desired decomposition result can be relatively easily obtained by the above decomposition method. However, if the blood flow signals of the two blood vessels are superimposed in the same direction in the initial Doppler spectrogram (as shown in Figure 5b), it may not be possible to obtain the desired decomposition result, which is not conducive to subsequent spectral feature recognition and The analysis can therefore optionally prompt the operator to re-acquire the unsuperimposed Doppler signal, so as to improve the accuracy of subsequent spectral feature recognition and analysis, and obtain more accurate abnormal spectral feature recognition results.

According to an embodiment of the present invention, before step S110, the method 100 may further include: acquiring an initial Doppler spectrum; if the initial Doppler spectrum is generated based on Doppler signals of two blood vessels superimposed together and In the initial Doppler spectrogram, the blood flow direction of the two blood vessels is reversed and there is an abnormality in the portion of the spectrum corresponding to the at least one blood vessel in the initial Doppler spectrogram, and the output is used to instruct the operator to re-implement transcranial Doppler Le check the instructions.

For the case where the blood flow signals of the two blood vessels in the initial Doppler spectrogram are inversely superimposed (as shown in Fig. 5a), if the spectral portions on both sides of the baseline are consistent with the normal spectrum, the spectrum may be prompted. The graph is normal. If the portion of the spectrum on at least one side of the baseline is abnormal, the operator may optionally be prompted to reacquire the Doppler signal to obtain a new initial Doppler spectrum. It is only possible to obtain the correct feature recognition result if the acquired initial Doppler spectrum is correct. Methods for adjusting transcranial Doppler examinations can include changing the angle and/or position of the probe, adjusting the depth of the sample, reducing the volume, and the like.

According to an embodiment of the present invention, before step S120, the method 100 may further include: performing noise reduction on the Doppler spectrum to be analyzed.

The data collected by the ultrasound transcranial Doppler flow analyzer is the superposition of noise and signal. Therefore, the transcranial Doppler spectrum obtained after Fourier transform is also the result of noise and signal superposition. As analyzed by the principle of ultrasound imaging, the noise on the transcranial Doppler spectrogram is generally uniformly distributed white noise. Doppler When the intensity of the signal is higher than the intensity of the background noise, the signal can be identified. The stronger the signal, the greater the difference in distribution between the signal and the noise, and the easier it is to separate.

The noise reduction may include noise reduction for the image in which the Doppler spectrogram to be analyzed is located and noise reduction for the noise in the Doppler spectrogram to be analyzed. The noise reduction of the image in which the Doppler spectrogram to be analyzed is located can be achieved by smoothing filtering. One useful filter is a Gaussian filter, which is a weighted average of the entire image. The pixel value of each pixel is obtained by weighted averaging of itself and other pixel values in the neighborhood. The Gaussian filter is a linear filter, and the larger the order, the better the filtering effect. In addition, nonlinear median filtering and other processing methods can be used to obtain similar effects. The noise reduction of the image in which the Doppler spectrogram to be analyzed is located can reduce the noise variance of the Doppler spectrogram to be analyzed, thereby further improving the separability of the signal and noise.

Noise reduction for noise in the Doppler spectrogram to be analyzed can be achieved using two example approaches described below.

In an example, performing noise reduction on the Doppler spectrum to be analyzed may include: extracting, by using a filtering method, a blood flow signal whose energy in the Doppler spectrum to be analyzed is higher than a preset energy threshold to obtain a drop. Doppler spectrogram to be analyzed after noise.

For well-designed transcranial Doppler flow analysis systems (including ultrasound transcranial Doppler flow analyzers), under certain parameters, the noise is uniformly distributed white noise at various frequencies in the transcranial Doppler spectrogram The mean and variance are the same. In one example, a threshold (ie, a preset energy threshold) can be set by the mean and variance of the background noise, and the spectral signal above this threshold can be classified as a blood flow signal and will be below this threshold. The spectral signal is defined as noise.

In another example, performing noise reduction on the Doppler spectrogram to be analyzed may include: averaging the overall energy of the Doppler spectrogram to be analyzed to obtain a spectral average; calculating a low in the Doppler spectrum to be analyzed The mean and variance of the energy of the average of the spectrum; the adaptive energy threshold is set according to the mean and the variance; and the blood flow signal of the energy to be analyzed in the Doppler spectrum to be higher than the adaptive energy threshold is extracted by filtering, The Doppler spectrum to be analyzed after noise reduction is obtained.

An adaptive noise calculation method can be used to distinguish between signals and noise. For example, the overall energy of the analyzed Doppler spectrogram can be averaged first, and then the spectral signal whose energy is lower than the average is used as noise, and the mean and variance of the determined noise are calculated. A threshold (ie, an adaptive energy threshold) can then be set using the calculated mean and variance, with which the signal and noise are differentiated. In this example, each transcranial Doppler spectrogram can have its own adaptive energy threshold, so this noise reduction method can better remove noise, which can improve the reliability of abnormal spectral feature recognition.

According to an embodiment of the present invention, before step S120, the method 100 may further include: performing interference filtering on the Doppler spectrogram to be analyzed by filtering, to remove the interference signal in the Doppler spectrogram to be analyzed.

During the acquisition of the transcranial Doppler spectrogram, it is difficult for the subject and the operator to remain stationary for a long time, so occasional small body movements (such as coughing) may cause image interference. In addition, various physical conditions may interfere with the image, such as strong electromagnetic interference. This source of interfering signals is not part of the subject and therefore has a negative impact on spectral analysis. In general, the interference signal (returning to Figure 4i) is significantly different from the normal blood flow signal. For example, the energy range and energy distribution of the two are different, and the duration of the interference signal is short, the shape is high, and there is no periodicity. . Illustratively, the interfering signal can be identified in the following manner: if the signal strength is increased by 6 decibels in a short period of time, the duration is less than 100 milliseconds, and the frequency range is wide (eg, more than 80% of the analyzable range), then It is short-term interference. Correct identification of interference can ensure that the data of subsequent analysis is basically a valid blood flow signal, and finally achieve correct abnormal spectral feature recognition. After the interference is identified, the interference can be filtered by nonlinear filtering. After interference filtering, a relatively smooth envelope (mainly the maximum envelope) can be obtained, which facilitates the implementation of subsequent envelope recognition and envelope smoothing steps.

According to an embodiment of the present invention, the abnormality information may further include: state information regarding whether the blood flow is normal as a whole and/or direction information regarding whether the blood flow direction is reversed.

If the Doppler spectrogram to be analyzed is a normal spectrum, the blood flow is normal as a whole, and the state information may be information indicating that the blood flow is normal as a whole. If abnormal spectral features such as stealing blood, eddy currents, turbulence, or dash lines or other abnormalities are identified from the Doppler spectrogram to be analyzed, the state information may be information indicating an abnormality in the blood flow as a whole. As described above, the blood flow direction can be determined by the direction of the maximum envelope, and will not be described again here. If the blood flow direction is positive, the direction information may be information indicating that the blood flow direction is positive, and if the blood flow direction is reverse, the direction information may be information indicating that the blood flow direction is reverse.

According to an embodiment of the invention, the method 100 may further include: acquiring a positive sample Doppler spectrum and a negative sample Doppler spectrum, wherein the positive sample Doppler spectrum includes the Doppler spectrum to be analyzed The specific spectral characteristics of the abnormal spectrum feature are consistent, the negative sample Doppler spectrum does not contain specific anomalous spectral features; the positive sample spectral features are identified from the positive sample Doppler spectrogram, and the negative sample Doppler spectrogram The negative sample spectral features are identified; and the classifier model is trained using positive sample spectral features and negative sample spectral features to obtain a trained classifier.

In one example, a positive sample Doppler spectrogram can contain blood stealing, and a positive sample spectral feature can Including waveform features of a positive sample Doppler spectrogram, the negative sample spectral features may include waveform features of a negative sample Doppler spectrogram. Correspondingly, the spectral features to be analyzed may include waveform features of the Doppler spectrogram to be analyzed, and the waveform characteristics of the analyzed Doppler spectrogram are analyzed by the trained classifier, and the Doppler spectrum to be analyzed may be obtained. Does the map contain information on stealing blood? In this case, the trained classifier can be used to identify blood stealing.

In one example, the positive sample Doppler spectrogram may comprise eddy currents, the positive sample spectral features may comprise energy distribution features of a positive sample Doppler spectrogram, and the negative sample spectral features may comprise negative sample Doppler spectrograms Energy distribution characteristics. Correspondingly, the spectral features to be analyzed may include the energy distribution characteristics of the Doppler spectrogram to be analyzed, and the energy distribution characteristics of the Doppler spectrogram to be analyzed by the trained classifier may be analyzed, and the Doppler to be analyzed may be obtained. Whether the Le spectrogram contains information about the eddy current. In this case, the trained classifier can be used to identify eddy currents.

In one example, the positive sample Doppler spectrogram may comprise a dash line, the positive sample spec feature may comprise an energy distribution characteristic of a positive sample Doppler spectrogram, and the negative sample spec feature may comprise a negative sample Doppler spectrum The energy distribution characteristics of the graph. Correspondingly, the spectral features to be analyzed may include the energy distribution characteristics of the Doppler spectrogram to be analyzed, and the energy distribution characteristics of the Doppler spectrogram to be analyzed by the trained classifier may be analyzed, and the Doppler to be analyzed may be obtained. Whether the Le spectrogram contains information on the dash. In this case, the trained classifier can be used to identify the dash.

In one example, the positive sample Doppler spectrogram may comprise turbulence, the positive sample spectro feature may comprise a flow velocity energy signature of the positive sample Doppler spectrogram, and the negative sample spectrogram may comprise a negative sample Doppler spectrogram Flow rate energy characteristics. Correspondingly, the spectral features to be analyzed may include the flow velocity energy characteristics of the Doppler spectrogram to be analyzed, and the flow rate energy characteristics of the Doppler spectrogram to be analyzed by the trained classifier are analyzed, and the Doppler to be analyzed may be obtained. Whether the Le spectrogram contains turbulent information. In this case, the trained classifier can be used to identify turbulence.

In one example, a positive sample Doppler spectrogram can include four anomalous spectral features such as stealing, eddy current, turbulence, and dash, and positive sample spectral features can include waveform characteristics, energy of a positive sample Doppler spectrogram Distribution characteristics and flow velocity energy characteristics, negative sample spectral features may include waveform characteristics, energy distribution characteristics, and flow velocity energy characteristics of the negative sample Doppler spectrum. Correspondingly, the spectral features to be analyzed may include waveform features, energy distribution characteristics, and flow velocity energy characteristics of the Doppler spectrogram to be analyzed, and the waveform characteristics and energy distribution characteristics of the Doppler spectrogram are analyzed by using the trained classifier. Analysis with the flow rate energy characteristics can obtain information about whether the Doppler spectrogram to be analyzed contains blood stealing, eddy current, turbulence, and dash. In this case, the trained classifier can be used to identify blood stealing at the same time, Four anomalous spectral features of eddy currents, turbulence and dashes.

The application range of the spectral analysis method described herein is not limited to the above example, and various types can be trained by changing the types of abnormal spectral features included in the sample Doppler spectrogram used in the classifier training process. The classifier, thus utilizing the trained classifier, can identify a variety of different anomalous spectral features. The training and application of the classifier can be set as needed. Illustratively, where the spectral features to be analyzed (or positive sample spectral features, or negative sample spectral features) for inputting the classifier include a plurality of features (eg, including waveform features and energy distribution features), Combine multiple features before entering the classifier.

Preferably, during the training of the classifier, the normal spectrum is selected as the negative sample Doppler spectrum. Illustratively, the classifiers described herein may be implemented using any suitable classifier that may be present or may occur in the future, such as Bayesian classifiers, support vector machines, neural networks, and decision trees. The use of classifiers makes it easy, convenient, and fast to identify anomalous spectral features.

Those skilled in the art will appreciate that the above-mentioned abnormal spectral features such as blood stealing, eddy current, turbulence, and dash are mainly used as a reference index for symptoms such as arterial stenosis, and these abnormal spectral features and blood flow parameters (including blood flow velocity) are usually required. The data such as the pulsation index and the spectrum form are combined to determine whether symptoms such as arterial stenosis exist. Therefore, the above-mentioned abnormal spectral characteristics are similar to the blood flow parameters, and it is impossible to accurately determine the presence or absence of symptoms such as arterial stenosis and the location and extent of arterial stenosis based on the abnormal spectral characteristics alone.

According to another aspect of the present invention, a spectroscopic analysis apparatus is provided. FIG. 6 shows a schematic block diagram of a spectral analysis device 600 in accordance with one embodiment of the present invention.

As shown in FIG. 6, the spectrum analysis apparatus 600 according to an embodiment of the present invention includes a spectrum acquisition module 610 to be analyzed, a feature identification module 620 to be analyzed, and an analysis module 630. The various modules may perform the various steps/functions of the spectral analysis method described above in connection with Figures 1-5, respectively. Only the main functions of the respective components of the spectrum analysis device 600 will be described below, and the details already described above are omitted.

According to an embodiment of the invention, the abnormal spectral feature includes blood stealing, and the spectral feature to be analyzed includes a waveform feature related to the waveform of the Doppler spectrogram to be analyzed, and the feature identification module 620 to be analyzed includes: a maximum envelope identification sub-module, For identifying a maximum envelope from a Doppler spectrum to be analyzed; a period dividing sub-module for dividing a cardiac cycle according to a change rule of the identified maximum envelope; and a waveform feature obtaining sub-module for identifying The variation of the maximum envelope in any cardiac cycle determines whether the waveform of the Doppler spectrogram to be analyzed is inverted to obtain waveform features.

According to an embodiment of the present invention, the feature identification module 620 to be analyzed further includes: an envelope smoothing submodule, And performing envelope smoothing on the identified maximum envelope before the period dividing sub-module divides the cardiac cycle according to the changed rule of the identified maximum envelope.

According to an embodiment of the invention, the abnormal spectral features include eddy currents and/or short horizontal lines, and the spectral features to be analyzed include energy distribution features related to the energy distribution of the Doppler spectrogram to be analyzed, and the feature identification module 620 to be analyzed includes: An effective average calculation sub-module for averaging the energy of the blood flow signal in the Doppler spectrogram to be obtained to obtain an effective average value; a search sub-module for finding energy in the Doppler spectrogram to be analyzed a target region above the effective average; a morphological analysis sub-module for analyzing the morphology of the target region to obtain morphological features; a symmetry analysis sub-module for analyzing the target region relative to the baseline of the Doppler spectrogram to be analyzed Symmetry to obtain symmetry features; wherein the energy distribution features include morphological features and symmetry features.

According to an embodiment of the present invention, in the case that the abnormal spectrum feature includes eddy current, the feature to be analyzed module 620 further includes: a first minimum envelope identification sub-module, configured to identify the minimum value package from the Doppler spectrum to be analyzed. And obtaining a frequency window feature about whether a frequency window exists, wherein the spectral feature to be analyzed further includes a frequency window feature.

According to an embodiment of the invention, the abnormal spectral feature includes turbulence, and the spectral feature to be analyzed includes a flow velocity energy characteristic related to a flow velocity and an energy relationship of the Doppler spectrogram to be analyzed, and the feature identification module 620 to be analyzed includes: a flow velocity energy relationship determiner a module for determining a correspondence between blood flow velocity and energy according to a Doppler spectrogram to be analyzed; a curve fitting sub-module for curve fitting with blood flow velocity and energy as variables; and a slope calculation sub-module for The slope of the fitted curve is calculated; wherein the flow energy characteristic includes a slope.

According to an embodiment of the present invention, the feature identification module 620 to be analyzed further includes: a second minimum envelope identification sub-module, configured to identify a minimum envelope from the Doppler spectrum to be analyzed to obtain whether the frequency window exists. A frequency window feature, wherein the spectral feature to be analyzed further includes a frequency window feature.

According to an embodiment of the invention, the apparatus 600 further includes: a first initial spectrum acquisition module for acquiring an initial Doppler spectrum; and an decomposition module for if the initial Doppler spectrum is based on two superimposed Or the Doppler signal generated by more than two blood vessels, the initial Doppler spectrum is decomposed into two or more sub-Doppler spectrograms corresponding to two or more blood vessels one-to-one; The spectrum acquisition module 610 to be analyzed includes a spectrum determination sub-module for determining one of two or more sub-Doppler spectrograms as a Doppler spectrum to be analyzed.

According to an embodiment of the present invention, the apparatus 600 further includes: a second initial spectrum acquisition module, configured to acquire an initial Doppler spectrum; and a first indication output module, configured to: if the initial Doppler spectrum is based on a stack The Doppler signals of two or more blood vessels added together and the blood flow direction of two or more blood vessels in the initial Doppler spectrum is the same, the output is used to indicate the operator re Instructions for performing a transcranial Doppler examination.

According to an embodiment of the present invention, the apparatus 600 further includes: a third initial spectrum acquisition module, configured to acquire an initial Doppler spectrum; and a second indication output module, configured to: if the initial Doppler spectrum is based on superimposed The Doppler signal generated by the two blood vessels and in the initial Doppler spectrogram, the blood flow direction of the two blood vessels is reversed and there is an abnormality in the portion of the spectrum corresponding to the at least one blood vessel in the initial Doppler spectrogram, then The indication is used to instruct the operator to re-implement the transcranial Doppler examination.

According to an embodiment of the invention, the apparatus 600 further includes: a noise reduction module, configured to: after the feature identification module 620 to be analyzed identifies the spectrum feature to be analyzed from the Doppler spectrum to be analyzed, to analyze the Doppler spectrum noise.

According to the embodiment of the present invention, the noise reduction module includes: a first filtering sub-module, configured to extract, by using a filtering method, a blood flow signal whose energy in the Doppler spectrum image to be analyzed is higher than a preset energy threshold, to obtain a noise reduction method. Doppler spectrogram to be analyzed.

According to an embodiment of the invention, the noise reduction module comprises: a spectral average calculation sub-module for averaging the overall energy of the Doppler spectrogram to be analyzed to obtain a spectral average; a mean variance calculation sub-module for Calculating a mean and a variance of energy below the average of the spectrum in the Doppler spectrogram to be analyzed; a threshold setting sub-module for setting an adaptive energy threshold based on the mean and the variance; and a second filtering sub-module for The blood flow signal whose energy is higher than the adaptive energy threshold in the Doppler spectrogram to be analyzed is extracted by filtering to obtain a Doppler spectrum to be analyzed after noise reduction.

According to an embodiment of the present invention, the apparatus 600 further includes: an interference filtering module, configured to analyze the Doppler spectrum by filtering before the feature identification module 620 to be analyzed identifies the feature to be analyzed from the Doppler spectrum to be analyzed. The figure performs interference filtering to remove the interference signal in the Doppler spectrum to be analyzed.

According to an embodiment of the present invention, the abnormality information further includes: state information regarding whether the blood flow is normal as a whole and/or direction information as to whether the blood flow direction is reversed.

According to an embodiment of the invention, the apparatus 600 further includes: a sample spectrum acquisition module, configured to acquire a positive sample Doppler spectrum map and a negative sample Doppler spectrum map, wherein the positive sample Doppler spectrum map includes and is to be analyzed The specific spectral characteristics of the abnormal spectral features included in the Pulcher spectrogram are consistent, the negative sample Doppler spectrum does not contain specific anomalous spectral features; the sample feature recognition module is used to identify positive from the positive sample Doppler spectrogram Sample spectral features and identifying negative samples from negative sample Doppler spectrograms Spectral features; and a training module for training the classifier model with positive sample spectral features and negative sample spectral features to obtain a trained classifier.

Those of ordinary skill in the art will appreciate that the elements and algorithm steps of the various examples described in connection with the embodiments disclosed herein can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the solution. A person skilled in the art can use different methods for implementing the described functions for each particular application, but such implementation should not be considered to be beyond the scope of the present invention.

FIG. 7 shows a schematic block diagram of a spectroscopic analysis device 700 in accordance with one embodiment of the present invention. The spectrum analysis device 700 includes a memory 710 and a processor 720.

The memory 710 stores program code (i.e., program) for implementing respective steps in the spectrogram analysis method according to an embodiment of the present invention.

The processor 720 is configured to execute program code stored in the memory 710 to perform respective steps of a spectrum analysis method according to an embodiment of the present invention, and to implement the spectrum analysis apparatus 600 according to an embodiment of the present invention. The spectrum acquisition module 610 to be analyzed, the feature identification module 620 to be analyzed, and the analysis module 630.

In one embodiment, when the program code is running in the processor 720, the method is configured to: acquire a Doppler spectrum map to be analyzed; and identify a spectrum feature to be analyzed from the Doppler spectrum to be analyzed. And using the trained classifier to analyze the analyzed spectral features to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether the abnormal spectral features exist, and the abnormal spectral features include blood stealing, eddy current, turbulence, and short horizontal One or more of the lines.

In one embodiment, the abnormal spectral features include blood stealing, and the spectral features to be analyzed include waveform features associated with waveforms of the Doppler spectrogram to be analyzed, the program code being used to execute when executed in the processor 720 The step of identifying the feature to be analyzed from the Doppler spectrogram to be analyzed includes: identifying a maximum envelope from the Doppler spectrogram to be analyzed; and dividing the cardiac cycle according to the changed rule of the identified maximum envelope; And determining whether the waveform of the Doppler spectrogram to be analyzed is reversed according to the variation rule of the identified maximum envelope in any cardiac cycle to obtain a waveform feature.

In one embodiment, the program code is in the process prior to the step of dividing the cardiac cycle according to a change rule of the identified maximum envelope when the program code is run in the processor 720 The step of identifying the feature to be analyzed from the Doppler spectrogram to be analyzed when executed in the 720 is further included: performing envelope smoothing on the identified maximum envelope.

In one embodiment, the anomalous spectral features include eddy currents and/or dashes, spectral signatures to be analyzed An energy distribution feature associated with an energy distribution condition of the Doppler spectrogram to be analyzed, the program code identifying a spectrum to be analyzed from the Doppler spectrogram to be analyzed for execution when executed in the processor 720 The characteristic step comprises: averaging the energy of the blood flow signal in the Doppler spectrogram to be obtained to obtain an effective average value; and searching for a target region whose energy is higher than the effective average in the Doppler spectrum to be analyzed; The morphology of the target region is obtained to obtain morphological features; the symmetry of the target region relative to the baseline of the Doppler spectrogram to be analyzed is analyzed to obtain symmetry features; wherein the energy distribution features include morphological features and symmetry features.

In one embodiment, the step of identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed for execution of the program code when the program code is run in the processor 720, in the case where the abnormal spectral features include eddy currents The method further includes: identifying a minimum envelope from the Doppler spectrogram to be analyzed to obtain a frequency window feature regarding whether the frequency window exists, wherein the spectral feature to be analyzed further includes a frequency window feature.

In one embodiment, the anomalous spectral features include turbulence, and the spectral features to be analyzed include flow velocity energy characteristics associated with flow rate and energy relationships of the Doppler spectrogram to be analyzed, the program code being run in the processor 720 The step of identifying the feature to be analyzed from the Doppler spectrum to be analyzed for performing includes: determining a correspondence between blood flow velocity and energy according to the Doppler spectrum to be analyzed; using blood flow velocity and energy as variables Curve fitting; and calculating the slope of the fitted curve; wherein the flow rate energy characteristic includes a slope.

In one embodiment, the step of identifying the spectral feature to be analyzed from the Doppler spectrogram to be analyzed when the program code is executed in the processor 720 further comprises: analyzing the Doppler spectrum from the spectrum to be analyzed. The minimum envelope is identified in the figure to obtain frequency window features as to whether the frequency window is present, wherein the spectral features to be analyzed also include frequency window features.

In one embodiment, the program code is also run in the processor 720 prior to the step of acquiring the Doppler spectrogram to be analyzed for execution of the program code in the processor 720 For performing the following steps: obtaining an initial Doppler spectrogram; and if the initial Doppler spectrogram is generated based on Doppler signals of two or more blood vessels superimposed together, initial Doppler will be generated The spectrogram is decomposed into two or more sub-Doppler spectrograms that correspond one-to-one with two or more blood vessels; the acquisition to be performed by the program code when executed in the processor 720 The step of the Doppler spectrogram includes determining one of the two or more sub-Doppler spectrograms as the Doppler spectrogram to be analyzed.

In one embodiment, the program code is at the processor 720 prior to the step of acquiring a Doppler spectrogram to be analyzed for execution when the program code is run in the processor 720. The middle run is also used to perform the following steps: obtaining an initial Doppler spectrogram; if the initial Doppler spectrogram is generated based on Doppler signals of two or more blood vessels superimposed together and initially If the blood flow directions of two or more blood vessels in the Pulcher spectrogram are the same, an indication is output for instructing the operator to re-implement the transcranial Doppler examination.

In one embodiment, the program code is also run in the processor 720 prior to the step of acquiring the Doppler spectrogram to be analyzed for execution of the program code in the processor 720 For performing the following steps: obtaining an initial Doppler spectrogram; if the initial Doppler spectrogram is generated based on Doppler signals of two blood vessels superimposed together and two blood vessels in the initial Doppler spectrogram The blood flow direction is reversed and there is an abnormality in the portion of the spectrum corresponding to the at least one blood vessel in the initial Doppler spectrogram, and the indication information for instructing the operator to re-implement the transcranial Doppler examination is output.

In one embodiment, prior to the step of identifying a spectral feature to be analyzed from the Doppler spectrogram to be analyzed for execution of the program code in the processor 720, the program code is The processor 720 is also used to perform the following steps: the Doppler spectrogram is to be analyzed for noise reduction.

In one embodiment, the step of performing noise reduction on the Doppler spectrogram to be performed when the program code is executed in the processor 720 includes: filtering the energy in the Doppler spectrogram to be analyzed A blood flow signal higher than a preset energy threshold is extracted to obtain a Doppler spectrum to be analyzed after noise reduction.

In one embodiment, the step of performing noise reduction on the Doppler spectrogram to be performed for execution of the program code in the processor 720 comprises: averaging the overall energy of the Doppler spectrogram to be analyzed To obtain the average value of the spectrum; calculate the mean and variance of the energy below the average of the spectrum in the Doppler spectrum to be analyzed; set the adaptive energy threshold according to the mean and variance; and filter the Doppler to be analyzed The blood flow signal whose energy is higher than the adaptive energy threshold in the Le spectrogram is extracted to obtain a Doppler spectrum to be analyzed after noise reduction.

In one embodiment, prior to the step of identifying a spectral feature to be analyzed from the Doppler spectrogram to be analyzed for execution of the program code in the processor 720, the program code is The processor 720 is further configured to perform the following steps: performing interference filtering on the Doppler spectrogram to be analyzed by filtering to remove the interference signal in the Doppler spectrogram to be analyzed.

In one embodiment, the abnormality information further includes: status information as to whether the blood flow is normal as a whole and/or direction information as to whether the blood flow direction is reversed.

In one embodiment, the program code, when run in the processor 720, is further configured to perform the steps of: acquiring a positive sample Doppler spectrum and a negative sample Doppler spectrum, wherein the positive sample Doppler The spectrogram contains specific anomalous spectral features consistent with the type of anomalous spectral features included in the Doppler spectrogram to be analyzed. The negative sample Doppler spectrogram does not contain specific anomalous spectral features; identifying positive from the positive sample Doppler spectrogram Sample spectral features, and identifying negative sample spectral features from negative sample Doppler spectrograms; and using the positive sample spectral features and negative sample spectral features to train the classifier model to obtain a trained classifier.

Moreover, according to an embodiment of the present invention, there is also provided a computer readable storage medium on which program instructions (ie, programs) are stored, which are used to execute the program when the program instructions are executed by a computer or a processor Corresponding steps of the spectroscopic analysis method of the embodiments of the invention, and for implementing respective modules in the spectroscopic analysis apparatus according to an embodiment of the present invention. The storage medium may include, for example, a memory card of a smart phone, a storage unit of a tablet, a hard disk of a personal computer, a read only memory (ROM), an erasable programmable read only memory (EPROM), a portable compact disk read only memory. (CD-ROM), USB memory, or any combination of the above storage media.

In one embodiment, the computer program instructions, when executed by a computer or processor, can cause a computer or processor to implement various functional modules of a spectroscopic analysis device in accordance with embodiments of the present invention, and/or can be implemented in accordance with the present invention. The spectral analysis method of the example.

In one embodiment, the computer program instructions are operative to perform the steps of: acquiring a Doppler spectrum to be analyzed; identifying spectral features to be analyzed from the Doppler spectrum to be analyzed; and utilizing the trained The classifier analyzes the analyzed spectral features to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether the abnormal spectral features exist, and the abnormal spectral features include one of stealing blood, eddy current, turbulence, and dash A variety.

In one embodiment, the anomalous spectral features include blood stealing, and the spectral features to be analyzed include waveform features associated with waveforms of the Doppler spectrogram to be analyzed, the computer program instructions being executed at runtime for more than one to be analyzed The step of identifying the characteristics of the spectrum to be analyzed in the Pullet spectrogram includes: identifying a maximum envelope from the Doppler spectrogram to be analyzed; dividing the cardiac cycle according to the changed rule of the identified maximum envelope; and according to the identified The variation of the maximum envelope in any cardiac cycle determines whether the waveform of the Doppler spectrogram to be analyzed is inverted to obtain waveform features.

In one embodiment, the computer program instructions are executed at runtime prior to the step of dividing the cardiac cycle according to the variation of the identified maximum envelope during execution of the computer program instructions Steps for identifying the features of the spectrum to be analyzed in the Doppler spectrogram to be analyzed It also includes performing envelope smoothing on the identified maximum envelope.

In one embodiment, the anomalous spectral features include eddy currents and/or dashes, and the spectral features to be analyzed include energy distribution characteristics associated with energy distribution conditions of the Doppler spectrogram to be analyzed, the computer program instructions being at runtime The step of identifying the characteristics of the spectrum to be analyzed from the Doppler spectrogram to be analyzed for performing includes averaging the energy of the blood flow signal in the Doppler spectrogram to be obtained to obtain an effective average value; The Doppler spectrogram maps the target region whose energy is higher than the effective average value; analyzes the shape of the target region to obtain the morphological feature; analyzes the symmetry of the target region relative to the baseline of the Doppler spectrogram to be analyzed to obtain symmetry Features; wherein the energy distribution features include morphological features and symmetry features.

In one embodiment, in the case where the abnormal spectral features include eddy currents, the step of identifying, by the computer program instructions, the spectral features to be analyzed from the Doppler spectrogram to be analyzed for execution at the runtime further comprises: waiting The minimum value envelope is identified in the Doppler spectrogram to obtain a frequency window feature for the presence or absence of the frequency window, wherein the spectral feature to be analyzed further includes a frequency window feature.

In one embodiment, the anomalous spectral features include turbulence, and the spectral features to be analyzed include flow velocity energy characteristics associated with flow rate and energy relationships of the Doppler spectrogram to be analyzed, the slaves used by the computer program instructions to execute at runtime The step of identifying the characteristics of the spectrum to be analyzed in the Doppler spectrogram to be analyzed includes: determining a correspondence relationship between blood flow velocity and energy according to the Doppler spectrogram to be analyzed; performing curve fitting with blood flow velocity and energy as variables; The slope of the fitted curve is calculated; wherein the flow rate energy characteristic includes a slope.

In one embodiment, the step of identifying, by the computer program instructions, the spectral features to be analyzed from the Doppler spectrogram to be analyzed for execution at runtime further comprises: identifying a minimum value from the Doppler spectrogram to be analyzed An envelope is obtained to obtain a frequency window feature as to whether a frequency window is present, wherein the spectral feature to be analyzed further includes a frequency window feature.

In one embodiment, prior to the step of acquiring the Doppler spectrogram to be analyzed for execution by the computer program instructions at runtime, the computer program instructions are further configured to perform the following steps at runtime: obtaining initial Doppler a spectrogram; and if the initial Doppler spectrogram is generated based on Doppler signals of two or more blood vessels superimposed together, the initial Doppler spectrogram is decomposed into two or more Two or more sub-Doppler spectrograms corresponding to the two blood vessels one by one; the step of obtaining the Doppler spectrogram to be analyzed performed by the computer program instructions at runtime includes: determining two One of the more than two sub-Doppler spectrograms is the Doppler spectrogram to be analyzed.

In one embodiment, the acquisition of the computer program instructions for execution at runtime Before the step of analyzing the Doppler spectrogram, the computer program instructions are also used at runtime to perform the steps of: obtaining an initial Doppler spectrogram; if the initial Doppler spectrogram is based on two or more superimposed The output is used to instruct the operator to re-implement the transcranial Doppler examination when the Doppler signals of the two vessels are generated and the blood flow directions of the two or more vessels are the same in the initial Doppler spectrogram. Instructions.

In one embodiment, prior to the step of acquiring the Doppler spectrogram to be analyzed for execution by the computer program instructions at runtime, the computer program instructions are further configured to perform the following steps at runtime: obtaining initial Doppler Lespectogram; if the initial Doppler spectrogram is generated based on the Doppler signals of the two vessels superimposed together and the blood flow direction of the two vessels is reversed in the initial Doppler spectrogram and at the initial Doppler If there is an abnormality in the portion of the spectrum corresponding to the at least one blood vessel in the Le spectrogram, the indication information for instructing the operator to re-implement the transcranial Doppler examination is output.

In one embodiment, the computer program instructions are also executed at runtime prior to the step of identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed for execution by the computer program instructions The following steps: To analyze the Doppler spectrogram for noise reduction.

In one embodiment, the step of performing noise reduction on the Doppler spectrogram to be performed by the computer program instruction during execution comprises: filtering the energy in the Doppler spectrogram to be analyzed higher than a preset energy by filtering The threshold blood flow signal is extracted to obtain a Doppler spectrum map to be analyzed after noise reduction.

In one embodiment, the step of the computer program instructions to perform noise reduction on the Doppler spectrogram to be performed performed at runtime comprises: averaging the overall energy of the Doppler spectrogram to be analyzed to obtain a spectrum Average; calculate the mean and variance of the energy below the average of the spectrum in the Doppler spectrum to be analyzed; set the adaptive energy threshold according to the mean and variance; and filter the energy in the Doppler spectrum to be analyzed A blood flow signal higher than the adaptive energy threshold is extracted to obtain a Doppler spectrum spectrum to be analyzed after noise reduction.

In one embodiment, the computer program instructions are also executed at runtime prior to the step of identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed for execution by the computer program instructions The following steps: performing interference filtering on the analyzed Doppler spectrogram by filtering to remove the interference signal in the Doppler spectrum to be analyzed.

In one embodiment, the computer program instructions are also used to perform the following steps at runtime: Obtaining a positive sample Doppler spectrogram and a negative sample Doppler spectrogram, wherein the positive sample Doppler spectrogram contains a specific abnormal spectral characteristic consistent with the abnormal spectral feature type included in the Doppler spectrogram to be analyzed, negative The sample Doppler spectrogram does not contain specific anomalous spectral features; the positive sample spectral features are identified from the positive sample Doppler spectrogram, and the negative sample spectral features are identified from the negative sample Doppler spectrogram; and positive samples are utilized Spectral features and negative sample spectral features train the classifier model to obtain a trained classifier.

In one embodiment, the spectroscopic analysis device 700 is a device that is independent of the ultrasound transcranial Doppler flow analyzer for acquiring a Doppler signal to obtain a Doppler spectrogram to be analyzed, or a spectrogram analysis device 700 It is the ultrasound transcranial Doppler blood flow analyzer.

In one example, the spectroscopic analysis device can be any stand-alone device with computing power, such as a personal computer, mobile terminal, server, or the like. The spectral analysis device can communicate with the transcranial Doppler flow analyzer by wire or wirelessly, and receive the transcranial Doppler spectrum acquired by the ultrasound transcranial Doppler flow analyzer (including Doppler to be analyzed). Spectrogram, initial Doppler spectrogram, positive sample Doppler spectrogram and negative sample Doppler spectrogram), and spectral analysis. The spectroscopic analysis device can also obtain transcranial Doppler spectrograms from other locations, such as downloading from a network, and the like. The use of a separate device to implement the spectral analysis device enables spectral analysis to be implemented on remote devices, while also facilitating faster processing speeds.

In another example, the spectroscopic analysis device can be an ultrasound transcranial Doppler flow analyzer. The processor of the spectroscopic analysis device may be a signal processing module in an ultrasound transcranial Doppler flow analyzer, and the memory of the spectroscopic analysis device may be a storage module in an ultrasound transcranial Doppler flow analyzer. The initial Doppler signal (which is the ultrasonic echo signal) acquired by the probe of the ultrasound transcranial Doppler flow analyzer is subjected to a series of operations such as acoustic-electrical conversion, amplification, analog-to-digital conversion (ADC), and demodulation. Converted to a valid Doppler signal, the Doppler signal is sent to the signal processing module for processing. In the signal processing module, steps of generating a transcranial Doppler spectrogram and performing spectrogram analysis based on the transcranial Doppler spectrogram may be performed. The Ultrasound Transcranial Doppler Flow Analyzer has an off-the-shelf processing module, so it is very easy to integrate the spectral analysis functions described in this paper into the ultrasound transcranial Doppler flow analyzer, which enables low cost implementation. The upgrade of the ultrasound transcranial Doppler flow analyzer can also enable the ultrasound transcranial Doppler flow analyzer to achieve more functions.

Each module in the spectrogram analyzing apparatus 600 according to an embodiment of the present invention may be implemented by a processor of an electronic device that performs spectrogram analysis according to an embodiment of the present invention running computer program instructions stored in a memory, or may be Computer of a computer program product of an embodiment of the invention The computer instructions stored in the readable storage medium are implemented by the computer when it is run.

Although the example embodiments have been described herein with reference to the drawings, it is understood that the foregoing exemplary embodiments are only illustrative, and are not intended to limit the scope of the invention. A person skilled in the art can make various changes and modifications without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the present invention as claimed.

In the several embodiments provided by the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. For example, the device embodiments described above are merely illustrative. For example, the division of the unit is only a logical function division. In actual implementation, there may be another division manner, for example, multiple units or components may be combined or Can be integrated into another device, or some features can be ignored or not executed.

In the description provided herein, numerous specific details are set forth. However, it is understood that the embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques are not shown in detail so as not to obscure the understanding of the description.

Similarly, the various features of the present invention are sometimes grouped together into a single embodiment, figure, in the description of exemplary embodiments of the invention, in the description of the exemplary embodiments of the invention. Or in the description of it. However, the method of the present invention should not be construed as reflecting the intention that the claimed invention requires more features than those specifically recited in the appended claims. Rather, as the invention is reflected by the appended claims, it is claimed that the technical problems can be solved with fewer features than all of the features of a single disclosed embodiment. Therefore, the claims following the specific embodiments are hereby explicitly incorporated into the embodiments, and each of the claims as a separate embodiment of the invention.

It will be understood by those skilled in the art that all features disclosed in the specification, including the accompanying claims, the abstract and the drawings, and all methods or devices so disclosed, may be employed in any combination, unless the features are mutually exclusive. Process or unit combination. Each feature disclosed in this specification (including the accompanying claims, the abstract and the drawings) may be replaced by alternative features that provide the same, equivalent or similar purpose.

In addition, those skilled in the art will appreciate that, although some embodiments described herein include certain features that are included in other embodiments and not in other features, combinations of features of different embodiments are intended to be within the scope of the present invention. Different embodiments are formed and formed. For example, in the claims, any one of the claimed embodiments can be used in any combination.

The various component embodiments of the present invention may be implemented in hardware, or in a software module running on one or more processors, or in a combination thereof. Those skilled in the art will appreciate that some or all of the functionality of some of the spectral analysis devices in accordance with embodiments of the present invention may be implemented in practice using a microprocessor or digital signal processor (DSP). The invention can also be implemented as a device program (e.g., a computer program and a computer program product) for performing some or all of the methods described herein. Such a program implementing the invention may be stored on a computer readable medium or may be in the form of one or more signals. Such signals may be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

It is to be noted that the above-described embodiments are illustrative of the invention and are not intended to be limiting, and that the invention may be devised without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses shall not be construed as a limitation. The word "comprising" does not exclude the presence of the elements or steps that are not recited in the claims. The word "a" or "an" The invention can be implemented by means of hardware comprising several distinct elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means can be embodied by the same hardware item. The use of the words first, second, and third does not indicate any order. These words can be interpreted as names.

The above is only the specific embodiment of the present invention or the description of the specific embodiments, and the scope of the present invention is not limited thereto, and any person skilled in the art can easily within the technical scope disclosed by the present invention. Any changes or substitutions are contemplated as being within the scope of the invention. The scope of the invention should be determined by the scope of the claims.

Claims

A method of spectral analysis, comprising:

Obtaining a Doppler spectrum to be analyzed;

Identifying spectral features to be analyzed from the Doppler spectrogram to be analyzed;

Using the trained classifier to analyze the spectral feature to be analyzed to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether an abnormal spectral feature exists, the abnormal spectral feature including blood stealing and eddy current One or more of turbulence, turbulence, and dash.
The method of claim 1 wherein said abnormal spectral features comprise said blood stealing, said spectral features to be analyzed comprising waveform features associated with said waveforms of said Doppler spectrogram to be analyzed, said Identifying the characteristics of the spectrum to be analyzed in the Doppler spectrum to be analyzed includes:

Identifying a maximum envelope from the Doppler spectrum to be analyzed;

Dividing the cardiac cycle according to the change rule of the identified maximum envelope;

Determining whether the waveform of the Doppler spectrogram to be analyzed is reversed according to a variation rule of the identified maximum envelope in any cardiac cycle to obtain the waveform feature.
The method according to claim 2, wherein said identifying a feature to be analyzed from said Doppler spectrum to be analyzed is further prior to said dividing a cardiac cycle according to a variation rule of said identified maximum envelope include:

Envelope smoothing is performed on the identified maximum envelope.
The method of claim 1 wherein said abnormal spectral features comprise eddy currents and/or dashes, said spectral features to be analyzed comprising energy associated with said energy distribution of said Doppler spectrogram to be analyzed The distribution feature, the identifying the feature to be analyzed from the Doppler spectrum to be analyzed includes:

And averaging the energy of the blood flow signal in the Doppler spectrum to be analyzed to obtain an effective average value;

Finding a target area whose energy is higher than the effective average value in the Doppler spectrum to be analyzed;

Analyzing the morphology of the target area to obtain morphological features;

Analyzing a symmetry of the target region relative to a baseline of the Doppler spectrogram to be analyzed to obtain a symmetry feature;

Wherein the energy distribution feature comprises the morphological feature and the symmetry feature.
The method of claim 4, wherein, in the case that the abnormal spectral feature comprises eddy current, the identifying the spectral feature to be analyzed from the Doppler spectrogram to be analyzed further comprises:

Identifying a minimum envelope from the Doppler spectrogram to be analyzed to obtain a frequency window feature as to whether a frequency window exists, wherein the spectral feature to be analyzed further includes the frequency window feature.
The method of claim 1 wherein said abnormal spectral features comprise turbulence, said spectral characteristics to be analyzed comprising flow velocity energy characteristics associated with said flow rate and energy relationship of said Doppler spectrogram to be analyzed, said Identifying the characteristics of the spectrum to be analyzed from the Doppler spectrum to be analyzed includes:

Determining a correspondence between blood flow velocity and energy according to the Doppler spectrum to be analyzed;

Curve fitting with the blood flow velocity and the energy as variables;

Calculating the slope of the fitted curve;

Wherein the flow rate energy characteristic comprises the slope.
The method of claim 6, wherein the identifying the feature to be analyzed from the Doppler spectrum to be analyzed further comprises:

Identifying a minimum envelope from the Doppler spectrogram to be analyzed to obtain a frequency window feature as to whether a frequency window exists, wherein the spectral feature to be analyzed further includes the frequency window feature.
The method of claim 1 wherein

Before the obtaining the Doppler spectrum to be analyzed, the method further includes:

Obtaining an initial Doppler spectrum; and

If the initial Doppler spectrogram is generated based on Doppler signals of two or more blood vessels superimposed together, decomposing the initial Doppler spectrogram into two or more Two or more sub-Doppler spectrograms corresponding to one another in two vessels;

The acquiring the Doppler spectrum to be analyzed includes:

Determining one of the two or more sub-Doppler spectrograms is the Doppler spectrogram to be analyzed.
The method of claim 1, wherein before the obtaining the Doppler spectrogram to be analyzed, the method further comprises:

Obtain an initial Doppler spectrum map;

If the initial Doppler spectrogram is generated based on Doppler signals of two or more blood vessels superimposed together and the two or more than two in the initial Doppler spectrogram The blood flow direction of the blood vessels is the same, and the indication information for instructing the operator to re-implement the transcranial Doppler examination is output.
The method of claim 1 wherein said obtaining a Doppler spectrogram to be analyzed The method further includes:

Obtain an initial Doppler spectrum map;

If the initial Doppler spectrogram is generated based on Doppler signals of two blood vessels superimposed together and in the initial Doppler spectrogram the blood flow directions of the two blood vessels are reversed and If there is an abnormality in the portion of the spectrum corresponding to the at least one blood vessel in the initial Doppler spectrogram, the indication information for instructing the operator to re-implement the transcranial Doppler examination is output.
The method of claim 1, wherein before the identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed, the method further comprises:

Denoising the Doppler spectrum to be analyzed.
The method of claim 11 wherein said denoising said Doppler spectrogram to be analyzed comprises:

The blood flow signal of the energy to be analyzed in the Doppler spectrum to be analyzed is higher than the preset energy threshold by filtering to obtain the Doppler spectrum to be analyzed after noise reduction.
The method of claim 11 wherein said denoising said Doppler spectrogram to be analyzed comprises:

And averaging the overall energy of the Doppler spectrum to be analyzed to obtain a spectral average;

Calculating a mean and a variance of energy in the Doppler spectrum to be analyzed that is lower than an average of the spectrum;

Setting an adaptive energy threshold based on the mean and the variance;

The blood flow signal whose energy is higher than the adaptive energy threshold in the Doppler spectrum to be analyzed is extracted by filtering to obtain the Doppler spectrum to be analyzed after noise reduction.
The method of claim 1, wherein before the identifying the spectral features to be analyzed from the Doppler spectrogram to be analyzed, the method further comprises:

The interference filtering is performed on the Doppler spectrum to be analyzed by filtering to remove the interference signal in the Doppler spectrum to be analyzed.
The method of claim 1, wherein the abnormality information further comprises: status information as to whether the blood flow is normal as a whole and/or direction information as to whether the blood flow direction is reversed.
The method of claim 1 wherein the method further comprises:

Obtaining a positive sample Doppler spectrogram and a negative sample Doppler spectrogram, wherein the positive sample Doppler spectrogram includes a specific anomaly consistent with an abnormal spectral feature type included in the Doppler spectrogram to be analyzed a spectral characteristic, the negative sample Doppler spectrum map not including the specific abnormal spectral feature;

Identifying positive sample spectral features from the positive sample Doppler spectrogram and identifying negative sample spectral features from the negative sample Doppler spectrogram;

The classifier model is trained using the positive sample spectral feature and the negative sample spectral feature to obtain the trained classifier.
A spectrum analysis device comprising:

a spectrum acquisition module to be analyzed for acquiring a Doppler spectrum to be analyzed;

a feature identification module to be analyzed, configured to identify a spectral feature to be analyzed from the Doppler spectrum to be analyzed;

An analysis module, configured to analyze the spectral feature to be analyzed by using a trained classifier to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether an abnormal spectral feature exists, the abnormal spectral feature Includes one or more of stealing blood, eddy currents, turbulence, and dashes.
A spectral analysis device comprising:

Memory for storing programs;

a processor for running the program;

Wherein, when the program runs in the processor, it is used to perform the following steps:

Obtaining a Doppler spectrum to be analyzed;

Identifying spectral features to be analyzed from the Doppler spectrogram to be analyzed;

Using the trained classifier to analyze the spectral feature to be analyzed to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether an abnormal spectral feature exists, the abnormal spectral feature including blood stealing and eddy current One or more of turbulence, turbulence, and dash.
The apparatus according to claim 18, wherein said spectral analysis device is independent of an ultrasonic transcranial Doppler blood flow analyzer for acquiring a Doppler signal to obtain said Doppler spectrum to be analyzed The device, or the spectroscopic analysis device, is the ultrasound transcranial Doppler blood flow analyzer.
A computer readable storage medium having stored thereon a program for performing the following steps at runtime:

Obtaining a Doppler spectrum to be analyzed;

Identifying spectral features to be analyzed from the Doppler spectrogram to be analyzed;

Using the trained classifier to analyze the spectral feature to be analyzed to obtain abnormal information, wherein the abnormal information includes abnormal spectral information about whether an abnormal spectral feature exists, the abnormal spectral feature including blood stealing and eddy current One or more of turbulence, turbulence, and dash.