Quantitative classification imaging method for small micro-motion/venous vessel micro-bubble contrast perfusion function in ultrasonic brainTechnical Field
The invention belongs to the field of ultrasonic imaging, and relates to a quantitative classification imaging method for the perfusion function of micro-bubble contrast of blood vessels of middle, small and micro-motion/veins of brain.
Background
The current application of Magnetic Resonance Imaging (MRI), positron emission computed tomography (PET) imaging techniques in brain imaging mainly includes imaging brain tissues and vascular structures, and developing to brain functional imaging, such as combining specific contrast agents to perform hemodynamic characterization analysis, but these imaging techniques have disadvantages of poor time resolution and spatial resolution, and failure to obtain local fine or perfusion characteristics of small brain microvessels.
Compared with other medical imaging technologies, the ultrasonic imaging has the advantages of non-invasiveness, convenience in use, low cost, small damage to human bodies and the like, and the existing transcranial Doppler imaging technology in brain imaging is clinically applied at present, so that the ultrasonic imaging has the advantages of time resolution, bedside examination, no radiation and the like. But can only image the big and middle blood vessels in the cranium to reflect the blood flow velocity change of the big and middle blood vessels of the brain, and can not realize the identification of the perfusion characteristics of local tiny or small cerebral blood vessels. Meanwhile, due to the shielding and attenuation effects of skull sound, the sensitivity and resolution of brain ultrasonic imaging, such as transcranial Doppler imaging, are poor, and the instantaneity and the depth are mutually restricted.
The Nakagami echo statistical model has the advantage of independence of acoustic parameters, and can recover or attenuate echo amplitude changes caused by sound field changes so as to reflect inherent scattering sub-ultrasonic echo characteristics. Whereas the cranial sound shielding and attenuation effect essentially alters the intracranial sound pressure and sound field distribution. Therefore, in theory, the Nakagami parameter has the possibility of improving the skull sound shielding and attenuation effects, but the combination of the Nakagami echo statistical model for carrying out ultrasonic local fine or brain small microvascular function real-time imaging, thereby providing multi-level and multi-parameter characterization for the brain, and still being the technical problem to be solved urgently.
Disclosure of Invention
The invention aims to provide a quantitative classification imaging method for the perfusion function of micro-motion/venous vascular microbubble radiography in an ultrasonic brain.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
1) Collecting brain ultrasonic echo signals through a temporal window by using an ultrasonic probe;
2) Carrying out microbubble wavelet decorrelation full-frequency domain detection on the ultrasonic echo signals acquired in the step 1) according to the transcranial attenuated microbubble mother wavelet, and then carrying out Nakagami echo statistical parameter imaging to obtain a cerebral blood vessel three-dimensional (3D) echo statistical parameter contrast image sequence matrix;
3) Reconstructing a cerebral blood vessel three-dimensional echo statistical parameter contrast image sequence matrix along a time sequence, then carrying out component analysis and characterizing perfusion components through machine learning, establishing association of cerebral medium, small and microvascular perfusion areas on the obtained time change curves of a plurality of perfusion components, and separating cerebral medium, small and microvascular distribution respectively;
4) Respectively extracting perfusion characteristics according to the distribution of the middle cerebral blood vessels, the small cerebral blood vessels and the micro blood vessels, and then calculating the perfusion parameters of the time class, the intensity class and the ratio class of the middle cerebral blood vessels, the small cerebral blood vessels and the micro blood vessels;
5) And (3) coding and imaging the cerebral vessels in the middle cerebral, small and micro-vascular distribution according to each perfusion parameter, and separating arteries from veins.
Preferably, the step 1) specifically includes the following steps:
1.1 Placing the acoustic lens into a packaging body with one side matched with the surface of the temporal bone, and enabling the low-frequency ultrasonic phased array probe to extend into the packaging body from the other side and be in butt joint with the acoustic lens through a coupling agent in the packaging body;
1.2 A low-frequency ultrasonic phased array probe is adopted, ultrasonic waves are transmitted to the brain through an acoustic lens (the acoustic lens is positioned through a packaging body) at the outer side of the temporal window, and ultrasonic echo signals are received and collected by the low-frequency ultrasonic phased array probe, so that original radio frequency data are obtained.
Preferably, in the step 2), the transcranial attenuated microbubble mother wavelet is obtained by constructing a sound field driven cerebrovascular microbubble mother wavelet according to a distribution of a sound field of the transcranial acoustic attenuation.
Preferably, in the step 2), the Nakagami echo statistical parameter imaging specifically includes the following steps: the method comprises the steps of carrying out microbubble wavelet de-correlation full-frequency domain detection on original radio frequency data by scanning lines, and then obtaining cerebrovascular microbubble wavelet de-correlation data, taking reference window lengths with different multiples as window sizes, and respectively estimating cerebrovascular microbubble radiography echo statistical parameters under each window size, wherein the reference window length is n times of the wavelength of ultrasonic waves emitted by a low-frequency ultrasonic phased array probe, and n=1, 3,5 …; and coding and imaging the brain blood vessel microbubble radiography echo statistical parameters.
Preferably, the step 3) specifically includes the following steps:
3.1 Performing two-dimensional linear transformation on the obtained three-dimensional echo statistical parameter contrast image sequence matrix of the cerebral blood vessel according to a time sequence to obtain a reconstructed two-dimensional (2D) matrix;
3.2 Performing principal component analysis on the two-dimensional matrix obtained in the step 3.1) to obtain a plurality of perfusion components; in the principal component analysis process, the values of elements in the two-dimensional matrix are assumed to be the linear combination of a principal component vector matrix and a space weight matrix;
3.3 Calculating time change curves for different perfusion components respectively, and then establishing association with perfusion areas of middle, small and micro blood vessels of the brain respectively according to the difference of the time change curves of the perfusion components.
Preferably, the step 4) specifically includes the following steps:
4.1 Layering the perfusion areas of the middle, small and micro blood vessels of the brain, wherein each layer takes a multiplied by a pixel as the window size, and single pixel as the window moving step length, and the perfusion time intensity curves of the middle, small and micro blood vessels of the brain are respectively extracted, wherein a=1-9;
4.2 According to the perfusion time intensity curve extracted in the step 4.1), calculating to obtain the perfusion parameters of time, intensity and ratio of the small, medium and micro blood vessels in the brain;
Preferably, the time-like perfusion parameter is one or more of a flushing time, a flushing time and a peak reaching time; the intensity perfusion parameter is one or more of peak value and area under the curve; the ratio type perfusion parameter is one or more of the flushing rate and the flushing rate.
Preferably, the step 5) specifically includes the following steps:
according to the perfusion characteristic difference of contrast microbubbles in the arteries and veins and the numerical values of perfusion parameters of time class, intensity class and ratio class, pseudo-color coding is respectively carried out, and the artery and vein separation is carried out on the middle cerebral, small cerebral and microvasculature by adopting an intensity double-threshold method and adjusting the upper threshold value and the lower threshold value of pseudo-color coding images.
The beneficial effects of the invention are as follows:
After the microbubble wavelet decorrelation full-frequency domain detection and Nakagami echo statistical parameter imaging are carried out on the original radio frequency data, the invention utilizes the brain blood vessel three-dimensional image matrix to carry out perfusion component characterization and separates the distribution of the middle, small and micro blood vessels, thereby carrying out perfusion parameter imaging on the craniocerebral artery and vein layering through the respective time, intensity and ratio perfusion parameters of the middle, small and micro blood vessels, and providing multi-level and multi-parameter characterization for brain lesions.
Furthermore, the invention adopts the phased array probe with the temporal window coupling accessory (comprising the packaging body and the acoustic lens) to acquire the original radio frequency data, compared with the case that the temporal window coupling accessory is not used, the invention can increase the transcranial propagation of ultrasonic waves, acquire more intracranial ultrasonic echo signals and improve the intracranial ultrasonic imaging resolution.
Drawings
FIG. 1 is a schematic diagram of an ultrasonic probe transmitting ultrasonic waves and acquiring echo signals in an embodiment of the present invention; wherein: (A) A coupling diagram of the low-frequency ultrasonic phased array probe and the temporal window; (B) schematic of a coupling fitting constructed with an acoustic lens; 1-phased array probe, 2-encapsulation body, 3-temporal bone, 4-acoustic lens.
Fig. 2 is a flow chart of quantitative classification imaging of small micro-motion/venous vascular microbubble contrast perfusion function in an ultrasonic brain according to an embodiment of the present invention.
Detailed Description
The invention is described in further detail below with reference to the drawings and examples. The examples are given solely for the purpose of illustration and are not intended to limit the scope of the invention.
The invention provides a quantitative classification imaging method for the perfusion function of micro-motion/venous vascular microbubble radiography in an ultrasonic brain. The imaging method is based on an ultrasonic perfusion function imaging principle, uses a low-frequency craniocerebral ultrasonic phased array probe with a temporal window coupling accessory to transmit ultrasonic waves, and then acquires echo signals; then, the obtained original radio frequency data is subjected to microbubble wavelet decorrelation full-frequency domain detection and echo statistical parameter imaging, so that not only is the problem of ultrasonic signal attenuation caused by skull solved, but also a three-dimensional (3D) echo statistical parameter contrast image sequence matrix of cerebral vessels (including middle, small and micro-blood vessels of the brain) is obtained. Then reconstructing the contrast image sequence matrix according to the time sequence, and utilizing a machine learning method to perform component analysis and characterization separation, and then separating the distribution of small blood vessels and medium blood vessels in the cranium. Finally, respectively extracting the time intensity curves of the perfusion time and intensity curves of the middle, small and micro blood vessels of the cranium, calculating to obtain respective time type, intensity type and ratio type parameters, and carrying out coding imaging and hemodynamic analysis on the artery and vein of the cranium.
Referring to fig. 2, the quantitative classification imaging method for the small micro-micro motion/vein blood vessel micro-bubble contrast perfusion function in the ultrasonic brain specifically comprises the following steps:
Step one, coupling the phased array probe with a temporal window acoustic lens device, and receiving and collecting ultrasonic echo signals of cerebral vascular microbubble radiography imaging through a temporal window.
(1.1) Construction of a temporal window acoustic lens device (temporal window coupling fitting as probe, fig. 1): designing an acoustic lens 4 with a concave surface along the thickness direction of the array of the low-frequency ultrasonic phased array probe 1, placing the acoustic lens 4 into a packaging body 2 with a surface matched with the temporal bone 3, and filling an ultrasonic coupling agent in the packaging body 2; the packaging body 2 is made of a material consistent with the acoustic window material of the low-frequency ultrasonic phased array probe 1, for example, a silica gel material is adopted to enhance the sound transmission efficiency of ultrasonic passing through the temporal window.
(1.2) Coupling the low-frequency ultrasonic phased array probe 1 into a temporal window acoustic lens device, so that one side of an acoustic lens 4 is an acoustic window of the low-frequency ultrasonic phased array probe 1, and the other side is a surface of the encapsulation body 2, which is matched with the temporal bone 3; the low-frequency ultrasonic phased array probe 1 transmits ultrasonic waves through a temporal window, receives and collects ultrasonic echo signals, and obtains brain tissue-blood vessel original radio frequency data. Wherein the center frequency of the low-frequency ultrasonic phased array probe 1 is 2.0-3.0 MHz.
Step two, designing a microbubble mother wavelet attenuated by the skull, and carrying out microbubble wavelet decorrelation full-frequency domain detection and echo statistical parameter imaging on brain tissue-blood vessel original radio frequency data obtained by a phased array probe to obtain a brain blood vessel three-dimensional (3D) echo statistical parameter contrast image sequence matrix.
(2.1) Constructing microbubble mother wavelets in cerebral blood vessels driven by the sound field according to the distribution of the acoustic attenuation sound field of the skull, and carrying out multi-scale microbubble wavelet decorrelation full-frequency domain detection (Ultrasound Med Biol,37 (2011), pp.1292-1305) on the obtained cerebral tissue-blood vessel original radio frequency data by scanning lines, thereby inhibiting cerebral tissue echo and improving cerebral blood vessel echo contrast. The full frequency domain is determined by the bandwidth of the ultrasonic probe, that is, the full receiving frequency range of the low-frequency ultrasonic phased array probe 1.
And (2.2) changing the window size (the window size is odd) by taking 3 times of the wavelength as the reference window length of the rectangular window, taking a single pixel as the moving step length of the rectangular window with the corresponding window size, moving from window to window, estimating the brain blood vessel microbubble radiography Nakagami echo statistical parameter in the scale window, and returning to the window center pixel (which is equivalent to smoothing the image).
And (2.3) coding and imaging the brain blood vessel microbubble contrast echo statistical parameter, recovering the brain blood vessel echo signal which is shielded and attenuated by the skull, and obtaining a brain blood vessel 3D echo statistical parameter contrast image sequence matrix.
Reconstructing a cerebral blood vessel 3D echo statistical parameter contrast image sequence matrix along a time sequence, performing machine learning component analysis and characterization separation, correlating the obtained time change curves of a plurality of perfusion components with the perfusion characteristics of the middle cerebral, small and micro blood vessels, and separating the middle cerebral, small and micro blood vessel distribution.
(3.1) Carrying out two-dimensional linear transformation on the obtained brain blood vessel 3D echo statistical parameter contrast image sequence matrix according to a time sequence (I= [ … It … ]), and reconstructing the obtained brain blood vessel 3D echo statistical parameter contrast image sequence matrix into a two-dimensional (2D) matrix (omega= [ … Dt … ]), wherein the specific transformation process is as follows:
dt＝[(x1)T…(xm)T]T,1≤t≤T
T, M and N are the number of frames, the number of columns of pixels of a single frame image and the number of rows of pixels of a single frame image of a cerebrovascular 3D echo statistical parameter contrast image sequence matrix, respectively. The superscript T denotes a transpose. It denotes a contrast image at time T in the T period. xm represents the mth column vector of the contrast image at time t. dt denotes the numerical matrix of each pixel of the contrast image at time t. xNM represents the pixel value of the pixel point of the last row and the last column of the contrast image.
(3.2) Performing principal component analysis on the obtained two-dimensional matrix omega to obtain a plurality of perfusion components, wherein the principal component analysis process is as follows:
In the model assumption, the values of the elements in the two-dimensional (2D) matrix (Ω) are a linear combination of the principal component vector matrix U and the spatial weight matrix V:
ΩK×T＝UK×fVf×T+ε
K=m×n, representing all pixels in the matrix Ω; f represents the order of the principal component, and epsilon may be omitted (epsilon means residual) when f is 4 or more. Therefore, the method of calculating the principal component vector matrix U is as follows:
according to the above formula, after the calculation of U, the principal component (specific experimental determination) corresponding to the perfusion parameter, that is, the perfusion component, is determined.
And (3.3) respectively calculating time change curves for different perfusion components, and then respectively establishing association (experimentally established association) with perfusion areas of the middle, small and micro blood vessels of the brain according to the difference of the time change curves of the perfusion components.
And (3.4) separating to obtain the distribution of the middle, small and micro blood vessels of the brain according to the association result of the perfusion components.
And fourthly, layering the perfusion areas of the middle, small and micro blood vessels of the cranium, respectively extracting the perfusion time intensity curves of the middle, small and micro blood vessels of the cranium, and calculating to obtain respective time, intensity and ratio parameters.
(4.1) Layering treatment (layering treatment is to treat the perfusion areas of the middle, small and micro blood vessels according to the distribution of the middle, small and micro blood vessels) of the brain, wherein each layering treatment uses 3X 3 pixels as window sizes, uses single pixels as window moving step length, and extracts the perfusion time intensity curves of the middle, small and micro blood vessels from window to window respectively.
(4.2) Calculating to obtain respective time, intensity and ratio parameters according to the obtained perfusion time and intensity curves of the middle cerebral, small cerebral and microvasculature; wherein the time class parameters include: time of flushing in, time of flushing out, time of reaching peak; the intensity class parameters include: peak value, area under curve; the ratio class parameters include: the punching rate and the punching rate. The time parameters are used for realizing the hemodynamic analysis of the craniocerebral arteries and veins respectively.
And fifthly, carrying out hemodynamic analysis and coded imaging on craniocerebral arteriovenous.
(5.1) According to the perfusion characteristic difference of 'arterial vessel first-to-first-off and venous then-to-last-off' existing in the blood circulation system of the contrast agent and the characterization of time parameters on the contrast agent flushing time, the peak reaching time and the flushing time, the middle cerebral, small arterial pulse and venous can be separated according to the numerical ranges of different time parameters.
(5.2) Performing pseudo-color coding imaging according to the values of various parameters; and (3) adopting an intensity double-threshold method, and performing artery and vein separation on the middle cerebral, small and micro-blood vessels by adjusting an upper threshold value and a lower threshold value of the pseudo-color coded image to obtain artery/vein multi-parameter perfusion parameter imaging of the middle cerebral, small and micro-blood vessels layering.
The invention has the following advantages:
1. The brain tissue-blood vessel signal obtained by the phased array probe can be subjected to microbubble wavelet decorrelation detection in the full frequency domain by utilizing the microbubble mother wavelet attenuated by the skull, so that the sensitivity and contrast of microbubble detection are improved.
2. The echo statistical parameter coding and imaging are carried out on the contrast micro-bubble signals in the cerebral vessels, so that series of problems of underestimation of the contrast micro-bubble signals, poor imaging contrast and the like caused by the effects of skull sound shielding and attenuation can be recovered.
3. Provides time change curves of various perfusion components, and obtains perfusion parameter information of middle, small and micro blood vessels of brain through separation.
4. The perfusion parameters according to time class can be used for carrying out hemodynamic analysis on craniocerebral artery and vein and respectively carrying out coded imaging.