Disclosure of Invention
In view of the above, the present invention provides a 3D vessel imaging algorithm based on an inverse principal component analysis method to alleviate the technical problems of low vessel imaging quality and serious noise in the prior art.
The embodiment of the invention provides a 3D blood vessel imaging algorithm based on a reverse principal component analysis method, which comprises the following steps:
a data acquisition step: adopting an electrocardio gate control method to collect electrocardio signals, outputting gate control signals according to the electrocardio signals and transmitting the gate control signals to a spectrometer, generating structural image data according to the output gate control signals, and registering the structural image data;
an image reconstruction step: constructing a complex signal statistical model according to the structural image data, wherein the complex signal statistical model is a linear statistical model comprising tissue reflection signals of non-blood flow tissue components, red blood cell reflection signals of blood flow components and Gaussian white noise; specifically, the complex signal statistical model is I ═ Ic+Ib+ N, where I is the tissue complex signal intensity matrix, IcIs a matrix of intensity of reflected signals from non-blood-flowing tissue, IbThe matrix is the intensity matrix of the blood vessel red cell reflection signal in the blood flow tissue, and N is a Gaussian white noise component;
extracting vascular erythrocyte reflected signals in the complex signal statistical model by adopting a reverse principal component analysis method, and calculating the characteristic value and the characteristic vector of the vascular erythrocyte reflected signals, wherein the extracted vascular erythrocyte reflected signals are IbGenerating a blood flow image, wherein h (w) is a filter signal response function constructed according to eigenvalues and eigenvectors of erythrocyte reflection signals;
3D image registration step: contour point cluster F for constructing any two blood flow imagesiThe contour point cluster is Fi=[xi,yi,zi]TI ═ 1,2,3,. and n), contour point clusters F of the two blood flow imagesoiAnd FtiAre respectively represented as Foi=xoie1+yoie2+zoie3And Fti=xtie1+ytie2+ztie3(ii) a Wherein,FoiContour point clusters representing the blood flow image at the initial time, FtiRepresenting a contour point cluster of the blood flow image at the next moment;
obtaining a contour point cluster F by adopting an independent component analysis modeliThe independent component analysis model is an optimization algorithm model based on an objective function;
and calculating the translation amount and the rotation amount of the independent element axis by using a differential search algorithm, and registering the blood flow image according to the translation amount and the rotation amount.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the electrocardiographic gating method specifically includes: and initializing the gate control module, acquiring an electrocardiosignal at the same time, comparing the electrocardiosignal serving as input data with a gate control value preset in the gate control module, judging whether the electrocardiosignal is higher than a threshold value, and if so, outputting the gate control signal corresponding to the current input electrocardiosignal.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the acquiring of the electrocardiographic data by using the electrocardiographic gating method specifically further includes: and adding a sensitive displacement compensation model in the acquisition process, and calculating the sensitive motion displacement parameters of the sensitive displacement compensation model by using a frequency domain filtering method and a multi-mode searching method.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, after extracting the eigenvalue and eigenvector of the vascular red blood cell reflection signal in the complex signal statistical model by using the inverse principal component analysis method, the method further includes: and (3) refining the blood flow red blood cell reflection signals by adopting a superposition mean phase elimination method.
Further, in the 3D vessel imaging algorithm based on the inverse principal component analysis provided in the embodiment of the present invention, the registration of the structural image specifically uses a feature-based registration method to register the structural image, and the feature-based registration method includes: carrying out orthogonal transformation on the pseudo-difference matrix Q; in particular, Q ═ Ii-Ii+1Wherein, Ii、Ii+1Respectively pixel matrixes of two adjacent frame images.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the optimization algorithm is any one of an artificial bee colony algorithm, an ant colony algorithm, a differential evolution algorithm, a bat algorithm, and a cuckoo algorithm.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the objective function in the optimization algorithm model based on the objective function specifically is:
establishing F ═ AfSfDe-mixing matrix WfSatisfy Yf=WfF=WfAfSfWherein S isfIs a set of points, AfIs a point set matrix, F is a point set, YfIs FiEstimation of the separated independent components, WfI.e. the objective function.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the calculation of the rotation amount and the translation amount corresponding to the independent axes specifically includes:
the rotation amount is:
translation quantity Δ C ═ C
o-C
tWherein l is
o-first,l
o-secondFirst and second axes, l, respectively, of the first blood flow structure image
t-first,l
t-secondA first and a second component axis of a second image of the structural flow, respectively, wherein,
and respectively representing the centroids of the contour point models of the two blood flow structure images.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided by the embodiment of the present invention, the sensitive displacement compensation model is si(n)=d(n)+ri(n)+ci(n)+hi(n) wherein si(n) is a sensitive motion displacement parameter, d (n) is a respiratory motion component, ri(n) is a cardiac motion component, ci(n) is a translationComponent of motion, hiAnd (n) is a noise component, and each component in the sensitive displacement compensation model is a function of the displacement along with the change of time.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the expression of the superposition mean phase elimination method is as follows:
wherein, I
b_clearTo refine the intensity matrix of the blood flow information, I
b_iIntensity matrix of single frame blood flow information for the ith frame, I
b_meanThe matrix is the multi-frame average blood flow information intensity from the ith frame to the Mth frame, and M is the total number of B-scans.
The embodiment of the invention has the following beneficial effects: according to the 3D blood vessel imaging algorithm based on the reverse principal component analysis method, firstly, an electrocardio gating method is adopted for acquiring electrocardio signals, gating signals are output according to the electrocardio signals and transmitted to a spectrometer, structural image data are generated according to the output gating signals, and the structural image data are registered. Secondly, a complex signal statistical model is constructed according to the structural image data, wherein the complex signal statistical model is a linear statistical model comprising tissue reflection signals of non-blood flow tissue components, red blood cell reflection signals of blood flow components and Gaussian white noise. And extracting the vascular erythrocyte reflection signals in the complex signal statistical model by adopting a reverse principal component analysis method, calculating the characteristic values and the characteristic vectors of the vascular erythrocyte reflection signals, extracting the vascular erythrocyte reflection signals, and generating a blood flow image. Finally, constructing contour point clusters of any two blood flow images, and obtaining independent element axes of the contour point clusters by adopting an independent component analysis model, wherein the independent component analysis model is an optimization algorithm model based on a target function; and calculating the translation amount and the rotation amount of the independent element axis by using a differential search algorithm, and registering the blood flow image according to the translation amount and the rotation amount. According to the technical scheme, an electrocardio-gating method is adopted for data acquisition, a complex signal statistical model and a reverse principal component analysis method are combined to generate a blood flow image, and finally 3D image registration is carried out on the blood flow image, so that three-dimensional angiography of tissue blood vessels is realized, the signal to noise ratio of imaging is improved, disordered background information generated by biological tissue reflection is reduced, the influence of biological jitter is weakened, and the quality of the imaging image is improved, so that the technical problems of low blood vessel imaging quality and serious noise in the prior art are solved.
In order to make the aforementioned and other objects, features and advantages of the present invention comprehensible, preferred embodiments accompanied with figures are described in detail below.
Detailed Description
To make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings, and it is apparent that the described embodiments are some, but not all embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
At present, in the living body imaging process of an OCTA system, the imaging signal-to-noise ratio is reduced and the image quality is damaged due to inevitable biological jitter such as heartbeat, respiration and the like; in the OCTA data acquisition process, the imaging quality is reduced due to related noise caused by the position offset of adjacent light sequence scanning; in the reconstruction process of the OCTA image, the position deviation of two adjacent frames of images causes the jitter noise in the images. The blood flow imaging algorithm widely used at present has low sensitivity and cannot extract images of tiny blood vessels; based on the fact that the existing blood flow image reconstruction algorithm cannot efficiently remove the disordered background information caused by biological tissue reflection, the 3D blood vessel imaging algorithm based on the reverse principal component analysis method provided by the embodiment of the invention can improve the signal-to-noise ratio of three-dimensional blood vessel imaging, reduce the disordered background information generated by biological tissue reflection and improve the imaging image quality.
Referring to fig. 1, a flowchart of a 3D blood vessel imaging algorithm based on an inverse principal component analysis method according to an embodiment of the present invention is provided. The 3D blood vessel imaging algorithm based on the reverse principal component analysis method provided by the embodiment of the invention comprises the following steps:
a data acquisition step S100: the method comprises the steps of adopting an electrocardio-gating method to collect electrocardiosignals, outputting gating signals according to the electrocardiosignals and transmitting the gating signals to a spectrometer, generating structural image data according to the output gating signals, and registering the structural image data. In the data acquisition process, because the imaging quality is influenced by the heartbeat and the artifact caused by respiratory motion, the electrocardio signal is acquired while the sample data is acquired by adopting an electrocardio gate control mode in the data acquisition process, so that the purpose of removing the imaging artifact caused by the heartbeat and respiratory motion is achieved.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the electrocardiographic gating method specifically includes: the gating module is initialized, the electrocardiosignal acquisition module acquires electrocardiosignals at the same time, the electrocardiosignal comparison module compares the electrocardiosignals serving as input data with a gating value preset in the gating module, whether the electrocardiosignals are higher than a threshold value or not is judged, and if yes, gating signals corresponding to the currently input electrocardiosignals are output. Referring to fig. 2 and fig. 3, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, comparison graphs of the effects before and after the cardiac gating method are used.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the acquiring of the electrocardiographic data by using the electrocardiographic gating method specifically further includes: and adding a sensitive displacement compensation model in the acquisition process, and calculating the sensitive motion displacement parameters of the sensitive displacement compensation model by using a frequency domain filtering method and a multi-mode searching method. In special cases, for example, some patients have some diseases on their hearts, which results in the heart beating not being periodic, or some non-periodic sensitive movements caused by the patients being young children, the imaging quality is seriously affected, and at this time, the compensation for the non-periodic sensitive displacement can be realized within a certain range by adding a sensitive movement displacement model in the data acquisition process.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided by the embodiment of the present invention, the sensitive displacement compensation model is si(n)=d(n)+ri(n)+ci(n)+hi(n) wherein si(n) is a sensitive motion displacement parameter, d (n) is a respiratory motion component, ri(n) is a cardiac motion component, ci(n) is the translational motion component, hiAnd (n) is a noise component, and each component in the sensitive displacement compensation model is a function of the displacement of the corresponding parameter of the sensitive displacement compensation model along with the change of time.
Referring to fig. 4, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided by the embodiment of the present invention, a schematic diagram of a registration effect of a structural image is shown. Furthermore, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, in the data acquisition step, the feature-based registration method is specifically adopted to register the structured image, so as to compensate the position offset between two adjacent B-scans, and remove the artifact, i.e., the noise part, caused by the jitter.The feature-based registration method comprises: carrying out orthogonal transformation on the pseudo-difference matrix Q; in particular, Q ═ Ii-Ii+1Wherein, Ii、Ii+1Respectively pixel matrixes of two adjacent frame structure images. Furthermore, after the step of registering the structural image, a zero padding technology can be added, so that the aim of increasing the registration precision of the structural image is fulfilled.
The 3D blood vessel imaging algorithm based on the reverse principal component analysis method provided by the embodiment of the invention further comprises an image reconstruction step S200, wherein the step S210: constructing a complex signal statistical model of a certain point in a spatial domain according to the structural image data, wherein the complex signal statistical model can ensure the integrity of signals, and the complex signal statistical model is a linear statistical model comprising tissue reflection signals of non-blood flow tissue components, red blood cell reflection signals of blood flow components and additional Gaussian white noise; specifically, the complex signal statistical model is I ═ Ic+Ib+ N, where I is the tissue complex signal intensity matrix, IcIs a matrix of intensity of reflected signals from non-blood-flowing tissue, IbThe intensity matrix of the blood vessel red cell reflection signal in the blood flow tissue, and N is a Gaussian white noise component.
Step S220: extracting a vascular erythrocyte reflection signal in a complex signal statistical model by adopting a reverse principal component analysis method, wherein the vascular erythrocyte reflection signal represents the reflection degree of blood vessels in tissues, the characteristic value and the characteristic vector of the vascular erythrocyte reflection signal are calculated as principal components in the complex signal statistical model, and the extracted vascular erythrocyte reflection signal is IbAnd designing a filtering function according to the characteristic value and the characteristic vector of the blood vessel erythrocyte reflection signal, further designing a PCA inverse filter, filtering non-blood flow tissue components, reserving blood flow tissue components, and generating a blood flow image, wherein H (w) is a filtering signal response function constructed according to the characteristic value and the characteristic vector of the erythrocyte reflection signal.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, after extracting the eigenvalue and eigenvector of the vascular red blood cell reflection signal in the complex signal statistical model by using the inverse principal component analysis method, the method further includes: and (3) refining the blood flow red blood cell reflection signals by adopting a superposition mean phase elimination method.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the expression of the superposition mean phase elimination method is as follows:
wherein, I
b_clearTo refine the intensity matrix of the blood flow information, I
b_iIntensity matrix of single frame blood flow information for the ith frame, I
b_meanThe matrix is the multi-frame average blood flow information intensity from the ith frame to the Mth frame, and M is the total number of B-scans.
The 3D blood vessel imaging algorithm based on the reverse principal component analysis method further comprises a 3D image registration step S300, the 3D image registration technology can effectively calculate the offset and the rotation amount of the acquired images at different time, and reference is provided for doctors to detect the preoperative and postoperative changes of the lesion position, the state of illness and other related aspects. Wherein, in step S310: contour point cluster F for constructing any two blood flow imagesiThe contour point cluster is Fi=[xi,yi,zi]TN in the formula is a point set of contour points, the contour point cluster data represents position information of the contour points, and contour point clusters F of the two blood flow imagesoiAnd FtiAre respectively represented as Foi=xoie1+yoie2+zoie3And Fti=xtie1+ytie2+ztie3(ii) a Wherein, FoiContour point clusters representing the blood flow image at the initial time, FtiRepresenting a contour point cluster of the blood flow image at the next moment; wherein e is1、e2、e3Respectively, unit direction vectors.
Step S320: contour point cluster F is obtained by adopting Independent Component Analysis (ICA) model in unsupervised learningiThe independent component analysis model is an optimization algorithm model based on an objective function. Further, the embodiment of the invention provides a method based on reverse principal componentIn the 3D blood vessel imaging algorithm of the analysis method, the optimization algorithm is any one of an artificial bee colony algorithm, an ant colony algorithm, a differential evolution algorithm, a bat algorithm and a cuckoo algorithm.
Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the objective function in the optimization algorithm model based on the objective function specifically is: establishing F ═ AfSfSpecifically, according to the ICA algorithm principle, there is a set of points SfAnd a point set matrix AfSatisfy F ═ AfSfThe goal of the ICA algorithm is to find a demixing matrix WfSatisfy Yf=WfF=WfAfSfWherein S isfIs a set of points, AfIs a point set matrix, F is a point set, YfIs FiEstimation of the separated independent components, WfNamely the objective function in the optimization algorithm model based on the objective function.
Step S330: and calculating the translation amount and the rotation amount of the independent element axis in the maximum statistical correlation direction by using a differential search algorithm, and registering the blood flow image according to the translation amount and the rotation amount. Further, in the 3D blood vessel imaging algorithm based on the inverse principal component analysis method provided in the embodiment of the present invention, the calculation of the rotation amount and the translation amount corresponding to the independent axes specifically includes:
the rotation amount is:
translation quantity Δ C ═ C
o-C
tWherein l is
o-first,l
o-secondFirst and second axes, l, respectively, of the first blood flow structure image
t-first,l
t-secondA first and a second component axis of a second image of the structural flow, respectively, wherein,
and respectively representing the centroids of the contour point models of the two blood flow structure images.
According to the 3D blood vessel imaging algorithm based on the reverse principal component analysis method, firstly, an electrocardio gating method is adopted for acquiring electrocardio signals, gating signals are output according to the electrocardio signals and transmitted to a spectrometer, structural image data are generated according to the output gating signals, and the structural image data are registered. Secondly, a complex signal statistical model is constructed according to the structural image data, wherein the complex signal statistical model is a linear statistical model comprising tissue reflection signals of non-blood flow tissue components, red blood cell reflection signals of blood flow components and Gaussian white noise. And extracting the vascular erythrocyte reflection signals in the complex signal statistical model by adopting a reverse principal component analysis method, calculating the characteristic values and the characteristic vectors of the vascular erythrocyte reflection signals, extracting the vascular erythrocyte reflection signals, and generating a blood flow image. Finally, constructing contour point clusters of any two blood flow images, and obtaining independent element axes of the contour point clusters by adopting an independent component analysis model, wherein the independent component analysis model is an optimization algorithm model based on a target function; and calculating the translation amount and the rotation amount of the independent element axis by using a differential search algorithm, and registering the blood flow image according to the translation amount and the rotation amount. According to the technical scheme, an electrocardio-gating method is adopted for data acquisition, a complex signal statistical model and a reverse principal component analysis method are combined to generate a blood flow image, and finally 3D image registration is carried out on the blood flow image, so that three-dimensional angiography of tissue blood vessels is realized, the signal to noise ratio of imaging is improved, disordered background information generated by biological tissue reflection is reduced, the influence of biological jitter is weakened, and the quality of the imaging image is improved, so that the technical problems of low blood vessel imaging quality and serious noise in the prior art are solved.
The above-mentioned embodiments are only specific embodiments of the present invention, which are used for illustrating the technical solutions of the present invention and not for limiting the same, and the protection scope of the present invention is not limited thereto, although the present invention is described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: any person skilled in the art can modify or easily conceive the technical solutions described in the foregoing embodiments or equivalent substitutes for some technical features within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the embodiments of the present invention, and they should be construed as being included therein. Therefore, the protection scope of the present invention shall be subject to the protection scope of the appended claims.