Specific embodiment
The invention will be further described with the following Examples.
Referring to Fig. 1, the tumour Precise Position System provided in this embodiment for medical treatment includes DDR tumor image acquisition mould
Block 1, DR tumor image acquisition module 2, image registration module 3, tumor-localizing module 4;The DDR tumor image acquisition module 1
For acquiring the DDR image in the patient tumors region of telltale mark point for containing positioning patient tumors;The DR tumor image is adopted
Collection module 2 is used to acquire the DR image in patient tumors region;Described image registration module 3 is used for the DDR image, DR image
Image registration is carried out, tumor-localizing parameter is obtained;The tumor-localizing module 4 is used for according to the tumor-localizing parameter to tumour
It relocates.
Preferably, the DDR image in the patient tumors region of telltale mark point of the acquisition containing positioning patient tumors, packet
It includes: acquiring the CT image of patient, DDR image is generated according to the CT image reconstruction of patient.
Preferably, the DR image in the acquisition patient tumors region, comprising: patient tumors region is shot from different perspectives,
Obtain multiple DR images.
The above embodiment of the present invention uses image registration techniques, realizes the accurate positioning of tumour.
Preferably, referring to fig. 2, the acquisition contains the patient tumors region of the telltale mark point of positioning patient tumors
DDR image, further includes: using customized screening function from multiple preferable CT images of CT optical sieving mass, for weight
It builds and generates DDR image: the wherein customized screening function are as follows:
Q={ Qi,Qi> 0, i=1 ..., m }
Wherein
In formula, Q is the target image set after screening, ZiFor the average gray value of i-th image in multiple images, m is to adopt
The quantity of the image of collection, WiFor the edge sharpness of i-th image in multiple images, W is that the edge that is set according to actual conditions is sharp
Threshold value is spent, whenWhen,When,
CT image after screening is used to rebuild generation DDR image by this preferred embodiment, can be improved the DDR figure of generation
The quality degree of picture lays the foundation to realize that tumour accurately positions.
Preferably, image registration is carried out to the DDR image, DR image, comprising:
(1) DDR image is chosen as reference picture S0, DR image calculates separately reference picture S as image S subject to registration0
With the Arimoto entropy of the entropy diagram picture of image S subject to registrationVS, define the calculation formula of Arimoto entropy are as follows:
In formula, VS(x,y)Indicate the Arimoto entropy of the entropy diagram picture of image S (x, y), U1、U2For the adjustment parameter of setting, and U1
>0,U1≠ 1, c (i, j) are the image block centered on pixel (x, y), having a size of n × n, and wherein n is odd number,J [c (i, j)] indicates the gray level of image block c (i, j), nkIt is
The frequency that k-th of gray level occurs, A are total pixel of image block c (i, j);
(2) it is based on differomorphism Demons algorithm, regards the registration of image as a gas diffusion process, gives iteration
The initial value Ψ of displacement field0, displacement field is updated by following iterative formula:
In formula, GδFor Gaussian filter, δ indicates the mean square deviation of Gaussian filter kernel function, and * indicates convolution operation, ΨkTable
Show displacement field when kth walks iteration, Ψk-1Indicate displacement field when -1 step iteration of kth, PSIndicate the gray scale of image S subject to registration
Value,Indicate the gray value of reference picture,Indicate the gradient of reference picture;
(3) constantly iteration updates displacement field, if meeting the stop condition of the objective function of differomorphism Demons algorithm, jumps
Circulation obtains final mean annual increment movement field Ψ out, otherwise continues to update displacement field, until reaching maximum number of iterations;
(4) using final mean annual increment movement field Ψ as the optimal transformation between image subject to registration, reference picture S is completed0With figure subject to registration
The registration of picture.
In this preferred embodiment, carries out stating DDR image, DR image registration using aforesaid way, reduce gray scale between image
Difference is influenced caused by registration result, improves the precision of image registration, to be advantageously implemented high-precision tumor-localizing.
Preferably, to realize more optimized image registration effect, the objective function of differomorphism Demons algorithm is carried out
Optimization introduces regularization term and gradient distribution distance terms, the objective function after definition optimization in objective function are as follows:
The stop condition of objective function are as follows:
In formula,For the regularization term of introducing, B1、B2For weight factor, ξ (Ψk) it is displacement field Ψk
Jacobian, M indicates the number of pixels of lap between reference picture and image subject to registration,It indicates to use
Displacement field ΨkDeformation is carried out to the entropy diagram picture of image subject to registration;For introducing gradient distribution away from
From item, α is the sample point in image gradient,Indicate the gradient distribution of image S subject to registration,Indicate reference picture
Gradient distribution.
In this preferred embodiment, it is contemplated that the rough problem in spatial information and registration between pixel, it is same to differential
The objective function of embryo Demons algorithm optimizes, and regularization term and gradient distribution distance terms is introduced in objective function, then
Optimal solution is sought using the objective function of the differomorphism Demons algorithm after optimization, relative to traditional differomorphism Demons
Algorithm can obtain higher registration accuracy, so as to obtain the tumor-localizing effect of degree of precision.
Finally it should be noted that the above embodiments are merely illustrative of the technical solutions of the present invention, rather than the present invention is protected
The limitation of range is protected, although explaining in detail referring to preferred embodiment to the present invention, those skilled in the art are answered
Work as understanding, it can be with modification or equivalent replacement of the technical solution of the present invention are made, without departing from the reality of technical solution of the present invention
Matter and range.