CN114663592A - Model making method and device based on medical image three-dimensional reconstruction - Google Patents

Model making method and device based on medical image three-dimensional reconstruction Download PDF

Info

Publication number
CN114663592A
CN114663592A CN202210301311.6A CN202210301311A CN114663592A CN 114663592 A CN114663592 A CN 114663592A CN 202210301311 A CN202210301311 A CN 202210301311A CN 114663592 A CN114663592 A CN 114663592A
Authority
CN
China
Prior art keywords
function
contour points
contour
tissue
gaussian
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202210301311.6A
Other languages
Chinese (zh)
Inventor
赵振峰
王彩兰
汪志宇
王立轩
李超
高瑜
骆健
诸葛晓旻
郭建明
张龙
董璇
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hebei Caiju Technology Co ltd
Original Assignee
Hebei Caiju Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hebei Caiju Technology Co ltd filed Critical Hebei Caiju Technology Co ltd
Priority to CN202210301311.6A priority Critical patent/CN114663592A/en
Publication of CN114663592A publication Critical patent/CN114663592A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T17/00—Three dimensional [3D] modelling, e.g. data description of 3D objects
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T5/00—Image enhancement or restoration
    • G06T5/70—Denoising; Smoothing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/10—Segmentation; Edge detection
    • G06T7/12—Edge-based segmentation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/60—Analysis of geometric attributes
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30004—Biomedical image processing
    • G06T2207/30008—Bone

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Geometry (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computer Graphics (AREA)
  • Software Systems (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)

Abstract

The invention relates to the technical field of image processing, and provides a model making method and a device based on medical image three-dimensional reconstruction, wherein the model making method based on the medical image three-dimensional reconstruction comprises the steps of obtaining coordinates of M multiplied by N contour points on M tissue slices; constructing a Gaussian low-pass filter, and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points; and establishing a three-dimensional model of the sample according to the filtered contour points. Through the technical scheme, the problem that the three-dimensional reconstruction model based on the medical image is poor in precision in the prior art is solved.

Description

Model making method and device based on medical image three-dimensional reconstruction
Technical Field
The invention relates to the technical field of image processing, in particular to a model making method based on medical image three-dimensional reconstruction.
Background
The three-dimensional reconstruction technology is to obtain medical image volume data by using computed tomography, magnetic resonance imaging and other computed imaging technologies, and select a suitable three-dimensional reconstruction algorithm as required to obtain a three-dimensional projection image which can be observed in all directions at three hundred and sixty degrees. Therefore, doctors can clearly and conveniently observe the structures of internal tissues or organs of the human bodies and obtain correct diagnosis results. The application of the three-dimensional reconstruction technology is beneficial to doctors to observe the three-dimensional structure of internal organs of a human body more intuitively and quantitatively, and can also strengthen certain original details in medical images according to the requirements of different disease diagnoses, so that the doctors are helped to make correct medical diagnosis, and therefore, the improvement of the three-dimensional reconstruction precision has important academic research significance and application value.
Disclosure of Invention
The invention provides a model making method and a device based on medical image three-dimensional reconstruction, which solve the problem of poor precision of a three-dimensional reconstruction model based on a medical image in the related technology.
The technical scheme of the invention is as follows:
in a first aspect, a method for modeling based on three-dimensional reconstruction of medical images includes:
obtaining coordinates of M multiplied by N contour points on M tissue slices, wherein the M tissue slices are obtained by sequentially carrying out image acquisition along the Z direction of the sample; obtaining coordinates of N contour points on any tissue slice, wherein each coordinate comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence;
constructing a Gaussian low-pass filter, and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points;
establishing a three-dimensional model of the sample according to the filtered contour points;
wherein, for any tissue slice, the coordinates of N contour points on the tissue slice are obtained, and the method specifically comprises the following steps:
obtaining N contour points on the edge line of the tissue slice at the same angle interval from the polar axis, and determining the polar diameter and the polar angle of each contour point; the angular spacing is the same on all tissue section edge lines.
In a second aspect, a modeling apparatus based on three-dimensional reconstruction of medical images includes:
a first obtaining unit configured to obtain coordinates of M × N contour points on M tissue slices, the M tissue slices being obtained by sequentially performing image acquisition along a Z direction of the sample; obtaining coordinates of N contour points on any tissue slice, wherein each coordinate comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence;
the first calculation unit is used for constructing a Gaussian low-pass filter and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points;
the first processing unit is used for establishing a three-dimensional model of the sample according to the filtered contour points;
wherein, for any tissue slice, the coordinates of N contour points on the tissue slice are obtained, and the method specifically comprises the following steps:
obtaining N contour points on the edge line of the tissue slice at the same angle interval from the polar axis, and determining the polar diameter and the polar angle of each contour point; the angular spacing is the same on all tissue section edge lines.
In a third aspect, a computer-readable storage medium has stored therein a computer program, which when executed by a processor implements the steps of the modeling method described above.
The working principle and the beneficial effects of the invention are as follows:
the M multiplied by N contour points on M sample tissue slices are obtained, the M tissue slices are distributed along the Z direction of the sample, the M multiplied by N contour points form a three-dimensional data set of the sample, the M multiplied by N contour points are filtered by designing a Gaussian low-pass filter, points with large fluctuation are filtered, the filtered contour points are adopted to carry out three-dimensional model reconstruction, and the method is favorable for obtaining a three-dimensional model showing smoothness and improving the reconstruction precision of the three-dimensional model.
Drawings
The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a schematic diagram of the contour point data structure of the present invention;
FIG. 3 is a profile point distribution plot of any one of the tissue slices of the present invention;
FIG. 4 is a schematic view showing the distribution of the slices in the Z direction according to the present invention;
FIG. 5 is a three-dimensional model of the auricular malleus obtained without filtering the contour points;
FIG. 6 is a three-dimensional model of the auricular malleus obtained from the filtered contour points;
FIG. 7 is a schematic diagram of approximate processing error analysis in accordance with the present invention;
FIG. 8 is a schematic view of the apparatus of the present invention;
in the figure: 1 polar axis, 2 contour points, 3 tissue sections.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any inventive step, are intended to be within the scope of the present invention.
Example one
As shown in fig. 1, a flowchart of a model making method according to this embodiment includes:
s100: obtaining coordinates of M multiplied by N contour points on M tissue slices, wherein the M tissue slices are obtained by sequentially carrying out image acquisition along the Z direction of the sample; obtaining coordinates of N contour points on any tissue slice, wherein each coordinate comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence;
taking the auricular malleus as an example, in this embodiment, an image of the auricular malleus is acquired through CT scanning and input into EAR-Lab image analysis software to obtain data of M × N contour points (contour points), and the data of the M × N contour points are displayed in a list form as shown in fig. 2. List0 corresponds to N contour points on the first tissue slice, List1 corresponds to N contour points on the second tissue slice, … and so on, List (M-1) corresponds to N contour points on the mth tissue slice.
As shown in fig. 3, a polar coordinate system is set on one tissue slice, the polar coordinate is (Xc, Yc), the polar axis direction is the X axis direction, the circumference of the tissue slice is divided equally according to the angle N from the polar axis, so as to obtain N contour points, and the coordinates of each contour point include the one-to-one corresponding X coordinate, Y coordinate, Z coordinate, polar diameter and polar angle. The contour points on the same tissue slice have the same pole coordinates and Z coordinates, the polar diameter of any contour point is the distance from the contour point to several points, and the polar angle of any contour point is as follows: the angle between the straight line from the pole to the contour point and the polar axis.
As shown in fig. 4, the distribution of M tissue slices in the Z direction.
S200: constructing a Gaussian low-pass filter, and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points;
due to dirt on the surface of the auricular malleus or shadow generated in the image acquisition process, abnormal contour points exist in the M × N contour points, and if the data containing the abnormal contour points are used for constructing the three-dimensional model of the auricular malleus, the fitted three-dimensional model has obvious burrs, which interferes with subsequent analysis, as shown in fig. 5. In this embodiment, a gaussian low-pass filter is designed to filter M × N contour points, and the filtered points are used for three-dimensional modeling of the auricular malleus, so that the burr problem can be effectively avoided, and the design process of the gaussian low-pass filter is as follows:
firstly, performing gaussian filtering on N contour points on each tissue slice on an XY plane, specifically including:
s210: executing the operation of establishing a first function on each tissue slice to obtain M first functions;
specifically, for any tissue slice, numbering N contour points on the slice in sequence; establishing a functional relation r between the polar diameter of any contour point on the tissue section and the number of the contour pointn＝fs(n), the functional relationship rn ═ fs (n) is the first function fs(n); n is any contour point on the slice, and N is 1,2nThe polar diameter of the nth contour point.
The angular separation between adjacent contour points is marked as Δ θ, then the polar angle θ of any contour point is:
θ＝θ0+nΔθ
where N is 1,2, … N, theta0＝0。
Polar diameter r corresponding to N contour pointsnA discrete signal, which can be considered as one-dimensional, is represented by the function fs (n) as:
rn＝fs(n)＝rn＝fs(θ0+nΔθ)。
fs (n) may be a functional expression obtained by fitting a plurality of contour points, or may be stored in the form of a data matrix.
S220: for each first function, determining a corresponding first Gaussian filter to obtain M first Gaussian filters; the calculation process of the first Gaussian filter corresponding to any first function comprises the following steps:
s221: performing discrete Fourier transform on the first function to obtain a corresponding first frequency domain function;
s222: determining a cut-off frequency σ of the first frequency-domain filter based on the spectral distribution of the first frequency-domain functionf；
After the spectrum distribution of the first frequency domain function is obtained, according to the principle of normal distribution 3 sigma: 99.7% of the energy is distributed in the interval (mu-3 sigma, mu +3 sigma) to obtain the cut-off frequency sigma of the first frequency-domain filterfAnd will not be described herein.
S223: according to the cut-off frequency sigma of the first frequency-domain filterfCalculating the cut-off frequency sigma of the first time-domain filtersThe method specifically comprises the following steps:
Figure BDA0003562982600000041
in this example σfAnd σsThe relation of (2) is deduced by the inventor, and the specific process is as follows:
given a continuous function f (x), x is initially x0Dividing the value range of x into N equal parts, and performing discrete processing on f (x) to obtain a discrete numerical value sequence { f (x)0),f(x0+Δx),f(x0+2Δx),...,f(x0+ (N-1) Δ x), the sequence of discrete values can be represented as:
f(x)＝f(x0+nΔx)
where N is 0,1,2, … N-1, and thus the sequence of discrete values may be expressed as { f (0), f (1), f (2),. f (N-1) }, performing a discrete fourier transform on the function f (x) yields:
Figure BDA0003562982600000042
wherein, u is 0,1,2, …, N-1.
Figure BDA0003562982600000043
Wherein x is 0,1,2, … N-1.
When u is 0,1,2, …, N-1,
Figure BDA0003562982600000051
the sequences of (a) are:
Figure BDA0003562982600000052
thereby obtaining:
Figure BDA0003562982600000053
as is known to those skilled in the art, σfExpressing the cut-off frequency of the function F (u), from which the cut-off frequency σ is obtainedfAnd a sampling frequency σsThe relation of (c):
σs＝σfΔu (2)
from the above equations (1) and (2):
Figure BDA0003562982600000054
s224: according to the cut-off frequency σ of the first time-domain filtersObtaining a formula of the first gaussian filter, specifically:
Figure BDA0003562982600000055
at the cut-off frequency σ of the first frequency domain filterfAfterwards, it is conventional practice to work according to σfA first frequency domain gaussian filter is obtained and then inverse fourier transformed to obtain the first gaussian filter, but this can be very computationally intensive. In this embodiment, the cutoff frequency σ of the first frequency domain filter is obtainedfThen, first according to σfAnd σsThe operational relationship of (a) yields σsAccording to σsAnd a first Gaussian filter is constructed, so that the operation amount is greatly reduced.
Secondly, filtering the contour points in the Z direction, specifically including:
s230: numbering the M tissue slices in sequence, and executing the operation of establishing a second function on N contour points of any tissue slice to obtain N second functions;
the operation of establishing the second function is executed on any contour point of any tissue slice, and the operation specifically comprises the following steps:
obtaining M contour points with the same polar angle as the contour point, and establishing a functional relation r between the polar diameter of any contour point in the M contour points and the slice numberm＝fz(m); wherein M is the section number of any tissue section, and M is 1,2mIs the polar diameter of any one of the M contour points.
The distance interval between two adjacent tissue slices is marked as Δ Z, and then in the Z direction, the height Z of any contour point is:
z＝z0+mΔz
wherein m is 1,2, … N, z0Is the initial height of the first picture.
Polar diameter r corresponding to M contour points with same polar anglemA discrete signal, which can be considered as one-dimensional, is represented by the function fz (m) as:
rm＝fz(m)＝rn＝fz(z0+mΔz)。
fs (n) may be a functional expression obtained by fitting a plurality of contour points, or may be stored in the form of a data matrix.
S240: and determining a corresponding second Gaussian filter for each second function by adopting the same steps as the steps S221 to S224 to obtain N second Gaussian filters.
S300: and establishing a three-dimensional model of the sample according to the filtered contour points.
Because the M multiplied by N contour points cover the image of the auricular malleus in the three-dimensional space, the filtered contour points are interpolated to obtain the three-dimensional model of the auricular malleus.
As shown in fig. 6, the filtered contour points are used to obtain a three-dimensional model of the auricular malleus, which is smoother in fig. 6 than in fig. 5.
The process of interpolating the filtered contour points in this embodiment is described as follows:
s310: calculating an error of the approximation process;
the error of the approximation processing includes a slice division error E generated at the slice division stepsegmentThe contour searching error E generated in the step of selecting contour pointscontourfindAnd filtering errors generated in the filtering step, wherein the calculation process of each error is as follows:
(1) error of slice division Esegment
Firstly, calculating the segmentation error of each tissue slice to obtain the segmentation errors of M tissue slices, wherein the segmentation error E of any tissue slicem,segmentThe calculation process comprises the following steps:
Figure BDA0003562982600000061
then, the segmentation errors of each tissue slice are accumulated to obtain a slice segmentation error Esegment：
Figure BDA0003562982600000062
Wherein M represents any tissue section, M-1, 2, …, M, ROIm,1And ROIm,2Two regions, S, on the tissue section m, respectivelymThe similarity of the two regions is characterized.
(2) Contour finding error Econtourfind
Firstly, calculating the contour error of each tissue slice to obtain the contour errors of M tissue slices, wherein the contour error E of any tissue slicem,contourfindThe calculation process comprises the following steps:
Figure BDA0003562982600000071
then accumulating the contour error of each tissue slice to obtain a contour searching error Econtourfind：
Figure BDA0003562982600000072
Wherein M represents any tissue section, M-1, 2, …, M, ROIm,segmentFinding a Pre-image, ROI, for the contourm,contourfindFor the contour found image, Sm,contourfindThe similarity of the two images is characterized.
(3) Slice filtering error Efiltering
Slice filtering error EfilteringThere are two methods for calculating (c).
The first method comprises the following steps:
firstly, calculating the filtering error of each tissue slice to obtain the filtering error of M tissue slices, wherein the filtering error E of any tissue slicem,filtering1The calculation process comprises the following steps:
Figure BDA0003562982600000073
then accumulating the filtering error of each tissue slice to obtain a slice filtering error Efiltering1：
Figure BDA0003562982600000074
Wherein M represents any tissue section, M-1, 2, …, M, ROIm,contourfindFor pre-filtered images, ROIm,filteringFor the filtered image, Sm,filtering1The similarity of the two images is characterized.
The second method comprises the following steps:
firstly, calculating the polar diameter error of each contour point to obtain the polar diameter errors of M contour points, wherein the polar diameter error E of any contour pointm,n,filtering2The calculation process comprises the following steps:
Figure BDA0003562982600000081
accumulating the polar diameter error of each contour point to obtain slice filtering error Efiltering2：
Figure BDA0003562982600000082
Wherein M represents any tissue slice, M is 1,2, …, M, r "m,nFor any contour point filtered polar diameter, rm,nAnd (4) the polar diameter before filtering any contour point.
The three errors generated by the approximation processing are introduced to obtain a slice segmentation error EsegmentContour finding error EcontourfindAnd after filtering the error, carrying out error combination on the three errors to obtain an error E of approximate processingapproxAs shown in fig. 7.
Three kinds of errors can be comprehensively calculated to determine the error E of the approximation processingapproxAs shown in the following formula:
Figure BDA0003562982600000083
the error E of the approximation process can also be determined by selecting the largest error as the dominant errorapproxAs shown in the following formula:
Eapprox＝max{Esegment,Econtourfind,Efiltering}。
as shown in fig. 7, when slice images are sliced, manual intervention is required, and in order to reduce errors caused by human intervention, the present embodiment performs three-time slice segmentation on the same set of slice images, the first two times are performed by the same person, the third time is performed by another person, and the errors E of the three-time slice segmentation are calculated respectivelysegment1,Esegment2,Esegment3Then take the average of the three errorsValue as the final slice segmentation error Esegment。
S320: according to the error of the approximate processing, interpolating the filtered contour points in the Z direction by adopting a self-adaptive interpolation algorithm to obtain an interpolated image; the interpolated image is used to construct a three-dimensional model of the sample.
Any contour point comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence, any M contour points with the same polar angle theta form an interpolation point set corresponding to the polar angle theta, and M multiplied by N contour points form N interpolation point sets. And respectively carrying out self-adaptive interpolation on the N interpolation point sets to obtain an interpolated image. The interpolation process of any polar angle theta region is as follows:
(1) and selecting a plurality of points from the interpolation point set corresponding to the polar angle theta at an initial sampling interval to be used as first-order interpolation points, and performing piecewise monotonic interpolation to obtain a first-order interpolation function.
In this embodiment, the M contour points in each interpolation point set are numbered in sequence, and the initial sampling interval is set such that one point is taken for each 16 contour points at intervals, that is, the contour point with the sequence number of 0 is taken as a first-order interpolation point, the contour point with the sequence number of 16 is taken as a second-order interpolation point, the contour point with the sequence number of 32 is taken as a third-order interpolation point, and so on.
The region between two adjacent first order interpolation points is called a first order interval, and all the first order intervals are numbered in sequence.
(2) And selecting contour points which are twice as many as the first-order interpolation points from the interpolation point set corresponding to the polar angle theta as second-order interpolation points, and performing piecewise monotonic interpolation again to obtain a second-order interpolation function.
Two times of second-order interpolation points can be realized by halving the initial sampling interval, that is, the contour point with the sequence number of 0 is taken as the first second-order interpolation point, the contour point with the sequence number of 8 is taken as the second-order interpolation point, the contour point with the sequence number of 16 is taken as the third second-order interpolation point, and so on.
The region between two adjacent second-order interpolation points is called a second-order interval, and all the second-order intervals are numbered in sequence.
(3) And calculating errors of the first-order interpolation function and the second-order interpolation function.
In this embodiment, the error of the first-order interpolation function and the second-order interpolation function at each second-order interval is calculated respectively, and whether the error of each second-order interval is smaller than the error set value is determined. Taking any two-order interval as an example, whether the following relation is true is judged:
Figure BDA0003562982600000091
wherein, TkRepresenting the number of the contour points in the second-order interval, t representing the serial number of the contour points in the second-order interval, and k representing the sequence number of the second-order interval; beta is coefficient, 0 < beta < 1, in this example, beta is 0.5; interpolant2(z) represents a second order interpolation function, Interpolant1(z) represents a first order interpolation function, Interpolant2(z) and Interpolant1(z) each represents a polar diameter corresponding to the polar angle θ at the height z; eapprox(z) represents the approximation processing error for all contour points in this second order interval.
z＝zk,0+tΔz
Wherein z isk,0Represents a smaller end point (second-order interpolation point) in the second-order interval, and Δ z represents a height difference between two adjacent contour points in the interpolation point set corresponding to the polar angle θ.
(4) If the above relation is true, stopping interpolation and outputting a second-order interpolation function Interpolant2(Z) as an interpolation function (Z-direction interpolation function) corresponding to the polar angle θ; otherwise, selecting contour points which are twice as many as second-order interpolation points from the interpolation point set corresponding to the polar angle theta as third-order interpolation points, and performing piecewise monotonic interpolation again to obtain a third-order interpolation function; judging whether the error between the third-order interpolation function and the second-order interpolation function is smaller than the error set value by adopting the same method as the step (3), if so, stopping interpolation, and outputting the third-order interpolation function as the interpolation function corresponding to the polar angle thetaAnd (3) counting (a Z-direction interpolation function), otherwise, selecting contour points which are twice as many as the third-order interpolation points from the interpolation point set corresponding to the polar angle theta, taking the contour points as fourth-order interpolation points, performing piecewise monotonic interpolation again to obtain a fourth-order interpolation function, and repeating the steps until the relation is satisfied, and outputting the interpolation function at the moment as the interpolation function corresponding to the polar angle theta.
For convenience of calculation, the initial sampling interval is set to be n-th power of 2, for example, the initial sampling interval is set to be 16-2 in this embodiment4Thus, the subsequent sampling intervals may be set to 8, 4, 2, 1 in sequence.
In this embodiment, the error of the approximate processing is used as an input variable of the adaptive interpolation algorithm, the number of contour points used for the adaptive interpolation is adjusted according to the error of the approximate processing, when the error of the approximate processing is large, the error allowed by the adaptive interpolation is large, and the interpolation can be performed by using fewer contour points to reduce the operation amount; when the error of the approximation processing is small, the error allowed by the adaptive interpolation is small, and more contour points are needed to be used for interpolation, so that the interpolation precision is improved. In this embodiment, adaptive interpolation is performed according to the error of the approximation processing, and the calculation amount can be reduced as much as possible while the accuracy requirement is satisfied.
Example two
As shown in fig. 7, a schematic structural diagram of the modeling apparatus of this embodiment includes:
a first obtaining unit configured to obtain coordinates of M × N contour points on M tissue slices obtained by sequentially performing image acquisition along a Z direction of the sample; obtaining coordinates of N contour points on any tissue slice, wherein each coordinate comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence;
the first calculation unit is used for constructing a Gaussian low-pass filter and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points;
the first processing unit is used for establishing a three-dimensional model of the sample according to the filtered contour points;
wherein, for any tissue slice, the coordinates of N contour points on the tissue slice are obtained, and the method specifically comprises the following steps:
obtaining N contour points on the edge line of the tissue slice at the same angle interval from the polar axis, and determining the polar diameter and the polar angle of each contour point; the angular spacing is the same on all tissue section edge lines.
Further, still include:
the second calculation unit is used for executing the operation of establishing the first function on each tissue slice to obtain M first functions;
the third calculation unit is used for determining a corresponding first Gaussian filter for each first function to obtain M first Gaussian filters; the calculation process of the first Gaussian filter corresponding to any first function comprises the following steps:
performing Fourier transform on the first function to obtain a corresponding first frequency domain function;
determining a cut-off frequency σ of the first frequency-domain filter based on the spectral distribution of the first frequency-domain functionf；
According to the cut-off frequency sigma of the first frequency-domain filterfCalculating the cut-off frequency sigma of the first time-domain filtersThe method specifically comprises the following steps:
Figure BDA0003562982600000111
according to the cut-off frequency sigma of the first time-domain filtersObtaining a formula of the first gaussian filter, specifically:
Figure BDA0003562982600000112
the method for establishing the first function for any tissue slice specifically comprises the following steps:
numbering the N contour points on the tissue slice in sequence; establishing a functional relation r between the polar diameter of any contour point on the tissue section and the number of the contour pointn＝fs(n) the functional relation rn＝fs(n)As a first function; n is any contour point on the slice, and N is 1,2nThe polar diameter of the nth contour point.
Further, still include:
the third calculation unit is used for numbering the M tissue slices in sequence and executing the operation of establishing a second function on the N contour points of any tissue slice to obtain N second functions;
the fourth calculation unit is used for determining a corresponding second Gaussian filter for each second function to obtain N second Gaussian filters; the calculation process of a second Gaussian filter corresponding to any second function is the same as the calculation process of a first Gaussian filter corresponding to any first function;
the operation of establishing the second function is executed on any contour point of any tissue slice, and specifically includes:
obtaining M contour points with the same polar angle as the contour point, and establishing a functional relation r between the polar diameter of any contour point in the M contour points and the slice numberm＝fz(m); wherein M is the section number of any tissue section, and M is 1,2mIs the polar diameter of any one of the M contour points.
Various modifications and specific examples of the modeling method in the first embodiment are also applicable to the modeling apparatus in the present embodiment, and a person skilled in the art can clearly know the implementation method of the modeling system in the present embodiment through the detailed description of the modeling method, so that the detailed description is omitted here for brevity of the description.
EXAMPLE III
The present embodiment also proposes a computer-readable storage medium, wherein a computer program is stored in the computer-readable storage medium, and when being executed by a processor, the computer program implements the steps of the modeling method described above.
The present invention is not limited to the above preferred embodiments, and any modifications, equivalent substitutions, improvements, etc. within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims (7)

1. A model making method based on medical image three-dimensional reconstruction is used for constructing a three-dimensional model of a sample according to a tissue slice of the sample, and is characterized by comprising the following steps:
obtaining coordinates of M multiplied by N contour points on M tissue slices, wherein the M tissue slices are obtained by sequentially carrying out image acquisition along the Z direction of the sample; obtaining coordinates of N contour points on any tissue slice, wherein each coordinate comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence;
constructing a Gaussian low-pass filter, and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points;
establishing a three-dimensional model of the sample according to the filtered contour points;
wherein, for any tissue slice, the coordinates of N contour points on the tissue slice are obtained, and the method specifically comprises the following steps:
obtaining N contour points on the edge line of the tissue slice at the same angle interval from the polar axis, and determining the polar diameter and the polar angle of each contour point; the angular spacing is the same on all tissue section edge lines.
2. The method for modeling based on three-dimensional reconstruction of medical image according to claim 1, wherein the gaussian low-pass filter comprises a first gaussian filter, and the constructing of the gaussian low-pass filter specifically comprises:
executing the operation of establishing a first function on each tissue slice to obtain M first functions;
for each first function, determining a corresponding first Gaussian filter to obtain M first Gaussian filters; the calculation process of the first Gaussian filter corresponding to any first function comprises the following steps:
performing Fourier transform on the first function to obtain a corresponding first frequency domain function;
determining a cut-off frequency sigma of the first frequency-domain filter based on the spectral distribution of the first frequency-domain functionf；
According to the cut-off frequency σ of the first frequency-domain filterfCalculating the cut-off frequency sigma of the first time-domain filtersThe method specifically comprises the following steps:
Figure FDA0003562982590000011
according to the cut-off frequency sigma of the first time-domain filtersObtaining a formula of the first gaussian filter, specifically:
Figure FDA0003562982590000012
the method for establishing the first function for any tissue slice specifically comprises the following steps:
numbering N contour points on the tissue slice in sequence; establishing a functional relation r between the polar diameter of any contour point on the tissue section and the number of the contour pointn＝fs(n) the functional relation rnFs (n) as a first function; n is any contour point on the slice, and N is 1,2nThe polar diameter of the nth contour point.
3. The method of claim 2, wherein the gaussian low-pass filter further comprises a second gaussian filter, and the constructing the gaussian low-pass filter further comprises:
numbering the M tissue slices in sequence, and executing the operation of establishing a second function on N contour points of any tissue slice to obtain N second functions;
for each second function, determining a corresponding second Gaussian filter to obtain N second Gaussian filters; the calculation process of a second Gaussian filter corresponding to any second function is the same as the calculation process of a first Gaussian filter corresponding to any first function;
the operation of establishing the second function is executed on any contour point of any tissue slice, and specifically includes:
obtaining M contour points with the same polar angle as the contour point, and establishing a functional relation r between the polar diameter of any contour point in the M contour points and the slice numberm＝fz(m); wherein M is the section number of any tissue section, and M is 1,2mIs the polar diameter of any one of the M contour points.
4. A model making device based on medical image three-dimensional reconstruction is characterized by comprising:
a first obtaining unit configured to obtain coordinates of M × N contour points on M tissue slices obtained by sequentially performing image acquisition along a Z direction of the sample; obtaining coordinates of N contour points on any tissue slice, wherein each coordinate comprises an X coordinate, a Y coordinate, a Z coordinate, a polar diameter and a polar angle which are in one-to-one correspondence;
the first calculation unit is used for constructing a Gaussian low-pass filter and filtering the coordinates of the M multiplied by N contour points to obtain filtered contour points;
the first processing unit is used for establishing a three-dimensional model of the sample according to the filtered contour points;
wherein, for any tissue slice, the coordinates of N contour points on the tissue slice are obtained, and the method specifically comprises the following steps:
obtaining N contour points on the edge line of the tissue slice at the same angle interval from the polar axis, and determining the polar diameter and the polar angle of each contour point; the angular spacing is the same on all tissue section edge lines.
5. The apparatus for modeling based on medical image three-dimensional reconstruction according to claim 4, further comprising:
the second calculation unit is used for executing the operation of establishing the first function on each tissue slice to obtain M first functions;
the third calculation unit is used for determining a corresponding first Gaussian filter for each first function to obtain M first Gaussian filters; the calculation process of the first Gaussian filter corresponding to any first function comprises the following steps:
performing Fourier transform on the first function to obtain a corresponding first frequency domain function;
determining a cut-off frequency σ of the first frequency-domain filter based on the spectral distribution of the first frequency-domain functionf；
According to the cut-off frequency σ of the first frequency-domain filterfCalculating the cut-off frequency sigma of the first time-domain filtersThe method specifically comprises the following steps:
Figure FDA0003562982590000031
according to the cut-off frequency sigma of the first time-domain filtersObtaining a formula of the first gaussian filter, specifically:
Figure FDA0003562982590000032
the method for establishing the first function for any tissue slice specifically comprises the following steps:
numbering N contour points on the tissue slice in sequence; establishing a functional relation r between the polar diameter of any contour point on the tissue section and the number of the contour pointn＝fs(n) the functional relation rnFs (n) as a first function; n is any contour point on the slice, and N is 1,2nThe polar diameter of the nth contour point.
6. The apparatus for modeling based on medical image three-dimensional reconstruction according to claim 5, further comprising:
the third calculation unit is used for numbering the M tissue slices in sequence and executing the operation of establishing a second function on the N contour points of any tissue slice to obtain N second functions;
the fourth calculation unit is used for determining a corresponding second Gaussian filter for each second function to obtain N second Gaussian filters; the calculation process of a second Gaussian filter corresponding to any second function is the same as the calculation process of a first Gaussian filter corresponding to any first function;
the operation of establishing the second function is executed on any contour point of any tissue slice, and the operation specifically comprises the following steps:
obtaining M contour points with the same polar angle as the contour point, and establishing a functional relation r between the polar diameter of any contour point in the M contour points and the slice numberm＝fz(m); wherein M is the section number of any tissue section, and M is 1,2mIs the polar diameter of any one of the M contour points.
7. A computer-readable storage medium, in which a computer program is stored which, when being executed by a processor, carries out the steps of the modeling method according to any one of claims 1-3.
CN202210301311.6A 2022-03-24 2022-03-24 Model making method and device based on medical image three-dimensional reconstruction Pending CN114663592A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210301311.6A CN114663592A (en) 2022-03-24 2022-03-24 Model making method and device based on medical image three-dimensional reconstruction

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210301311.6A CN114663592A (en) 2022-03-24 2022-03-24 Model making method and device based on medical image three-dimensional reconstruction

Publications (1)

Publication Number Publication Date
CN114663592A true CN114663592A (en) 2022-06-24

Family

ID=82031006

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210301311.6A Pending CN114663592A (en) 2022-03-24 2022-03-24 Model making method and device based on medical image three-dimensional reconstruction

Country Status (1)

Country Link
CN (1) CN114663592A (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017028516A1 (en) * 2015-08-18 2017-02-23 青岛海信医疗设备股份有限公司 Three-dimensional image calibration method, apparatus and system
CN106934785A (en) * 2015-12-28 2017-07-07 哈尔滨工业大学 The medical image cutting method of hepatic model in a kind of training system for Robot Virtual

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017028516A1 (en) * 2015-08-18 2017-02-23 青岛海信医疗设备股份有限公司 Three-dimensional image calibration method, apparatus and system
CN106934785A (en) * 2015-12-28 2017-07-07 哈尔滨工业大学 The medical image cutting method of hepatic model in a kind of training system for Robot Virtual

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
赵振峰: "Application of Image Segmentation and Adaptive Interpolation Techniques to 3D Reconstruction of the Human Temporal Bones", 《SAINT MARY"S UNIVERSITY》 *
陈文文等: "基于面绘制的医学序列图像三维重建", 《计算机工程与应用》 *

Similar Documents

Publication Publication Date Title
CN109978778B (en) Convolutional neural network medical CT image denoising method based on residual learning
CN107527359B (en) PET image reconstruction method and PET imaging equipment
WO2007002406A2 (en) Interactive diagnostic display system
CN109523584A (en) Image processing method, device, multi-mode imaging system, storage medium and equipment
EP1209622B1 (en) Method and device for registering images
US20120032676A1 (en) Spatial intensity correction for RF shading non-uniformities in MRI
DE102012216327A1 (en) A method for detecting movement of a patient during a medical imaging examination
CN110070612B (en) CT image interlayer interpolation method based on generation countermeasure network
CN109069122B (en) System and method for determining dimensions of structures in medical images
CN103026379A (en) Method for estimating image noise level
WO2007115859A1 (en) Method and apparatus for providing tomographic image data for an object
CN110279429B (en) Four-dimensional ultrasonic reconstruction method and device
CN108596995B (en) PET-MRI maximum posterior joint reconstruction method
JP7300811B2 (en) Medical information processing apparatus, medical information processing method, and program
CN114663592A (en) Model making method and device based on medical image three-dimensional reconstruction
DE112022000826T5 (en) CREATING AN ANATOMIC MODEL OF AN ANATOMIC CAVITY
US20210333347A1 (en) Magnetic resonance imaging apparatus and image processing method
CN117893682A (en) Model making method and device based on medical image three-dimensional reconstruction
WO2014053256A1 (en) Isotropic reconstruction of image data
CN114663447A (en) AI medical image processing method and device based on computer vision
EP3384302B1 (en) A method for metabolite signal quantitation for magnetic resonance spectroscopy data
JP2021099794A (en) System and method for reconstructing medical images using deep neural network and recursive decimation of measurement data
CN113269798B (en) Medical image segmentation method and device, computer equipment and storage medium thereof
CN113936069B (en) Array element virtual interpolation method for photoacoustic tomography
Moraes et al. 3D Lanczos interpolation for medical volumes

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
RJ01 Rejection of invention patent application after publication

Application publication date: 20220624

RJ01 Rejection of invention patent application after publication