CN102542598A - Local characteristic reinforcing volume rendering method oriented to medical volume data - Google Patents

Abstract

The invention discloses a local characteristic reinforcing volume rendering method oriented to medical volume data. The method comprises the following steps of: reading medical image data; performing moving least square method smoothness on sampling rays along sampling rays; searching for a minimum value point along sampling rays, judging whether the minimum value point is a characteristic demarcation point according to a gradient threshold value provided by a user, and recording parameters; judging whether a current sampling point is positioned inside a characteristic tissue according to a characteristic analysis result in first light ray projection, if so, judging whether the characteristic function threshold value f (Sa, Ga) of a current characteristic fragment is larger than a characteristic threshold value Tfeature provided by the user, if so, executing a reinforced accumulation method, and otherwise, executing a direct volume rendering accumulation method; and when light ray accumulation is completed, performing Tone attenuation operation, and quickly and clearly rendering tissue characteristics in medical volume data. The method has the advantages of simple interaction, high instantaneity, good rendering effect and display of more local characteristics.

Description

A kind of local feature intensive aspect method for drafting towards medical volume data

Technical field

The present invention relates to a kind of local feature intensive aspect method for drafting towards medical data.

Background technology

Along with clinical medical development, the medical visualization technology obtains application more and more widely at medical domain.On behalf of this medical visualization technology, the new therapy such as the appearance of remote diagnosis, remote operation etc. also become the indispensable important technology of field of medical applications.The medical visualization technology appears at the seventies in last century, mainly in computed tomography and nuclear magnetic resonance equipment, diagnoses as the graph technology medical assistance that shows two dimension slicing.Along with the development of computer hardware technology and the raising of medical demand, the medical visualization technology of three-dimensional real-time rendering has obtained developing fast and using, like maximum intensity value projection (MIP) technology, direct volume drawing technological (DVR) etc.

A kind of rendering technique of characteristic fast in the maximum intensity value shadow casting technique is used for drawing fast the maximum intensity value in positron emission computed tomography (PET) data by the foreign study person at first.Original M IP lacks the three-dimensional information of volume data; Development through decades; The MIP technology has been introduced three-dimensional spatial informations such as the depth information, gradient information of volume data; The bigger zone of intensity level in can high-quality demonstration tissue data is such as the bone in the CT data, blood vessel and the aneurysm in the MRI brain data etc. in the angiographic data.Yet,,, lack space three-dimensional information, and the characteristic that can draw is less though there is fast, the mutual simple advantage of speed in MIP with the direct volume drawing compared with techniques.The direct volume drawing technology can be optical properties with the data map that samples in the volume data, and finally be presented in the two dimensional image through the light integration.Along with developing rapidly of high performance graphics hardware, the researcher has overcome the shortcoming that direct volume drawing technology to drawing speed is slow, lack real-time, interactive, can access the real-time mutual drafting of high-quality.This year, domestic and international researcher introduced senior illumination model on the basis of direct volume drawing, made the effect of the realistic drafting of drafting effect of direct volume drawing.

The drawing result of direct volume drawing technology depends on the design and the adjusting of transition function, and simple transition function often can not come out separate tissue, and complicated transition function requires a great deal of time and energy carries out manual adjustment.And regulating transition function needs certain domain knowledge, in adjustment process, can not be fed back intuitively, can only obtain desirable transition function through making repeated attempts.The design of transition function has become the bottleneck of direct volume drawing technology, and many in the last few years researchers have obtained some achievements in the research of semi-automatic and automatic transition function, do not regulate a complicated difficult problem but solve transition function fully.Both at home and abroad some scholars have proposed some volume datas and have peeled off the method for drafting on the basis that the graphics degree of depth is peeled off, and have realized the delamination of volume data is drawn, and have reasonable effect aspect the drafting in that medical volume data is peeled off.For example nuclear magnetic resonance brain data are peeled off drafting, can cerebral cortex and adipose tissue be peelled off, draw out the gully of brain clearly.Also there is the researcher to combine the advantage of the method for maximum intensity value projection and direct volume drawing; Design the method for maximum intensity tissue in a kind of quick display body data; And be applied in biological neurology research field; But this method can only the display body data in the zone that changes of maximum intensity value, ignored the demonstration of local feature.Therefore, although the researcher has proposed the sorting technique of many volume datas, how fast the user's tissue of interest characteristic in the display of medical volume data remains focus of medical visualization research field.

Summary of the invention

For the mutual complicacy that overcomes existing direct volume drawing technology towards medical volume data, draw effect relatively poor, show the less deficiency of local feature, the present invention provide a kind of mutual simple, draw respond well, as to show more local feature local feature intensive aspect method for drafting towards medical volume data.

The technical solution adopted for the present invention to solve the technical problems is:

A kind of local feature intensive aspect method for drafting towards medical volume data may further comprise the steps:

A) read medical image data, and import to the video card internal memory;

B) carry out ray cast for the first time, it is level and smooth along sampling light sampling light to be moved least square method;

C), and judge according to the Grads threshold that the user provides whether minimum point is the characteristic separation along sampled light line search minimum point, and recording feature separation position P _i, and the average scalar value S in calculated characteristics zone _a, average gradient mould value G _a, and according to formula

$f(S_a,G_a)=(1-\mathrm{\λ})S_a+\frac{\mathrm{\λ}}{n}{G}_a---\left(1\right)$

Calculate the characteristic threshold value functional value of each characteristic area, and be recorded in structure { P _s, P _e, f (S _a, G _a) in, λ is user-defined parameter;

D) ray cast for the second time, the result according to signature analysis in the first time ray cast judges whether current sampling point is in feature organization inside, if then judge the fundamental function threshold value f (S of current characteristic fragment _a, G _a) the characteristic threshold value T that whether provides greater than the user _Feature, if, then carry out and strengthen accumulation method, otherwise, the direct volume drawing accumulation method carried out;

Said reinforcement accumulation method is: introduce characteristic and strengthen rendering parameter β, the accumulation formula is as follows:

C _acc←C _acc×β _i+(1-α _acc×β _i)×C _i

(4)

α _acc←α _acc×β _i+(1-α _acc×β _i)×α _i

Wherein, C _AccBe accumulation color value, α _AccBe accumulation opacity value, C _iAnd α _iBe respectively current color value and opacity value, β _iFor characteristic is strengthened rendering parameter; In sampling process, if current sampling point belongs to characteristic area, and the characteristic threshold value function of current characteristic area satisfies condition, then β _i=1-(s _i-s _I-1), otherwise β _i=1.0;

E) when the light accumulation finishes, carry out the luminance balance of final image, carry out the Tone attenuation operations, shown in formula (3)

$C_{\mathrm{new}}=C_{\mathrm{cur}}\×\frac{\ln (C_{\mathrm{cur}}+\mathrm{\λ})}{\ln (C_{\max }+\mathrm{\λ})}---\left(3\right)$

C wherein _NewFor carrying out the color value after Tone decays, C _CurBe the accumulation color value of current sampling light, C _MaxBe the maximum brightness value that equipment can show, λ is user-defined brightness decay coefficient, is used to control the brightness of final image color;

Tissue signature in the medical volume data is drawn fast and clearly.

Further again, in the said step d), carry out illumination calculation, adopt the Phong illumination model, with the gradient of current sampling point surface normal, utilize formula as illumination

C _sh＝(k _a+k _d(N·L))×C+k _s(N·H) ⁿ (2)

Calculate new color value and participate in color accumulation, wherein k _a, k _d, and k _sBe respectively surround lighting, scattered light and Gao Guang influence parameter, and N represents the direction of the normal vector of current sampling point; Normalization gradient vector by current sampling point representes that L represents the direction of incident ray, is confirmed by the position of current sampling point and the position of light source; H represents the half-angle between reflected light and the normal vector, is used for calculating Gao Guang, and n is high spectrum number; Be used for controlling scope and the brightness of Gao Guang, C represents the color value of current sampling point, in direct volume drawing, is specified by transition function.

Technical conceive of the present invention is: utilize to strengthen the accumulation method for drafting, the local feature on the sampling light is strengthened accumulation draw, thereby the maximum intensity value tag in the medical volume data and local feature are plotted in the middle of the final image; Utilize the Phong illumination model to strengthen and draw the sense of effect space multistory, and use the Tone decay technique to realize the luminance balance of final graphics, thereby the tissue signature in the medical volume data is drawn fast and clearly.

Through accumulation method is strengthened in local feature analysis on the sampling light and characteristic, the feature organization in the medical volume data is plotted in the two dimensional image, and provides a kind of characteristic to draw controlling mechanism, be used for the simple mutual characteristic of drawing.This method is compared with direct volume drawing, has mutual simple, the advantage of local feature in the display body data fast; Compare with the maximum intensity value projection algorithm, have draw effective, have spatial information and the strong advantage of three-dimensional stereopsis, and can show more local feature.Solved the problem of characteristic in the medical science volume drawing being strengthened drafting to a certain extent, experiment shows that our method can obtain gratifying characteristic to medical image datas such as PET/CT data, MRI data and draw effect.

Beneficial effect of the present invention mainly shows: can realize the characteristic demonstration of three-dimensional tissue fast of medical image data, and not need the labor time to regulate transition function; When medical image data obtains, can offer user's visual information intuitively timely.

Description of drawings

Fig. 1 is the level and smooth synoptic diagram of mobile least square on the sampling optical fiber.

Fig. 2 is system module figure.

Fig. 3 is the signature analysis process flow diagram.

Fig. 4 is that characteristic is strengthened accumulation drawing process figure.

Embodiment

Below in conjunction with accompanying drawing the present invention is further described.

With reference to Fig. 1～Fig. 4, a kind of local feature intensive aspect method for drafting towards medical volume data may further comprise the steps:

A) read medical image data, and import to the video card internal memory;

B) carry out ray cast for the first time; It is level and smooth along sampling light sampling light to be moved least square method (MLS), and level and smooth effect is shown in figure (1), and wherein (a) is the scalar value sampling curve on a certain the light of sampling of MRI brain data; (b) for this through moving the sampling curve of least square method after level and smooth; Can find out that by figure mobile least square method has been removed the most of high frequency noise data in the raw data, and has kept data characteristics.

C), and judge according to the Grads threshold that the user provides whether minimum point is the characteristic separation along sampled light line search minimum point, and recording feature separation position P _i, and the average scalar value S in calculated characteristics zone _a, average gradient mould value G _a, and according to formula

$f(S_a,G_a)=(1-\mathrm{\λ})S_a+\frac{\mathrm{\λ}}{n}{G}_a---\left(1\right)$

Calculate the characteristic threshold value functional value of each characteristic area, and be recorded in structure { P _s, P _e, f (S _a, G _a) in, to offer next step as the information that characteristic is drawn, wherein λ is user-defined parameter, can control scalar and the Grad weight in function.

D) ray cast for the second time, the result according to signature analysis in the first time ray cast judges whether current sampling point is in feature organization inside, if then judge the fundamental function threshold value f (S of current characteristic fragment _a, G _a) the characteristic threshold value T that whether provides greater than the user _Feature, if, then carry out and strengthen accumulation method, otherwise, the direct volume drawing accumulation method carried out; This process is carried out illumination calculation simultaneously, adopts the Phong illumination model, with the gradient of the current sampling point surface normal as illumination, utilizes formula

C _sh＝(k _a+k _d(N·L))×C+k _s(N·H) ⁿ (2)

Calculate new color value and participate in color accumulation, wherein k _a, k _d, and k _sBe respectively surround lighting, scattered light and Gao Guang influence parameter, and N represents the direction of the normal vector of current sampling point; Normalization gradient vector by current sampling point representes that L represents the direction of incident ray, is confirmed by the position of current sampling point and the position of light source; H represents the half-angle between reflected light and the normal vector, is used for calculating Gao Guang, and n is high spectrum number; Be used for controlling scope and the brightness of Gao Guang, C represents the color value of current sampling point, in direct volume drawing, is specified by transition function usually.

E) when the light accumulation finishes, carry out the luminance balance of final image, carry out the Tone attenuation operations, shown in formula (3)

$C_{\mathrm{new}}=C_{\mathrm{cur}}\×\frac{\ln (C_{\mathrm{cur}}+\mathrm{\λ})}{\ln (C_{\max }+\mathrm{\λ})}---\left(3\right)$

C wherein _NewFor carrying out the color value after Tone decays, C _CurBe the accumulation color value of current sampling light, C _MaxBe the maximum brightness value that equipment can show, λ is user-defined brightness decay coefficient, is used to control the brightness of final image color.

As shown in Figure 2, comprise the data importing module, characteristics analysis module and characteristic are strengthened drafting module; Wherein data importing module realizes that through the C++ program characteristics analysis module and characteristic are strengthened drafting module and realized through the GLSL program.The parallel acceleration carried out because the GLSL language is a graphic hardware, therefore can reach the drafting efficient of real-time, interactive.

Characteristics analysis module of the present invention is shown in figure (3); Before carrying out signature analysis; It is level and smooth at first need to move least square method to the data on the sampling light; Data after level and smooth are shown in Fig. 1 (b), and data have smooth effect preferably, help to seek more efficiently, more accurately the characteristic separation.Need search out the minimum point on the scalar value sampling curve in the process of signature analysis, and judge through user-defined Grads threshold whether current minimum point is the characteristic separation; In the process of seeking the characteristic separation, program is carried out the calculating of scalar value average and gradient-norm value average simultaneously, when finding a characteristic fragment, carries out the calculating of characteristic threshold value function, and the record functional value is as the controlled variable of characteristic drafting.

Characteristic is strengthened the program circuit of drafting module shown in figure (4), and this process is the basis with the light projecting algorithm framework of standard.At first need introduce characteristic and strengthen rendering parameter β, feature organization strengthened drawing to be implemented in the light accumulation.The accumulation formula is as follows:

C _acc←C _acc×β _i+(1-α _acc×β _i)×C _i

(4)

α _acc←α _acc×β _i+(1-α _acc×β _i)×α _i

Wherein, C _AccBe accumulation color value, α _AccBe accumulation opacity value, C _iAnd α _iBe respectively current color value and opacity value, β _iFor characteristic is strengthened rendering parameter.In sampling process, if current sampling point belongs to characteristic area, and the characteristic threshold value function of current characteristic area satisfies condition, then β _i=1-(s _i-s _I-1), otherwise β _i=1.0.By this method, can realize regulating dynamically the accumulation opacity in the accumulation, some feature organization that possibly be blocked is plotted in the final image.This module has comprised illumination calculation simultaneously and the Tone decay is calculated; Strengthened the three-dimensional stereopsis of final drafting effect; And strengthened the contrast of feature organization to a certain extent, made the final drafting effect display organization characteristic information more clearly of medical image data.

Claims (2)

1. local feature intensive aspect method for drafting towards medical volume data is characterized in that: may further comprise the steps:

A) read medical image data, and import to the video card internal memory;

B) carry out ray cast for the first time, it is level and smooth along sampling light sampling light to be moved least square method;

C), and judge according to the Grads threshold that the user provides whether minimum point is the characteristic separation along sampled light line search minimum point, and recording feature separation position P _i, and calculate the special average scalar value S that detects the zone _a, average gradient mould value G _a, and according to formula

$f(S_a,G_a)=(1-\mathrm{\λ})S_a+\frac{\mathrm{\λ}}{n}{G}_a---\left(1\right)$

Calculate the characteristic threshold value functional value of each characteristic area, and be recorded in structure { P _s, P _e, f (S _a, G _a) in, λ is user-defined parameter;

D) ray cast for the second time, the result according to signature analysis in the first time ray cast judges whether current sampling point is in feature organization inside, if then judge the fundamental function threshold value f (S of current characteristic fragment _a, G _a) the characteristic threshold value T that whether provides greater than the user _Feature, if, then carry out and strengthen accumulation method, otherwise, the direct volume drawing accumulation method carried out;

Said reinforcement accumulation method is: introduce characteristic and strengthen rendering parameter β, the accumulation formula is as follows:

C _acc←C _acc×β _i+(1-α _acc×β _i)×C _i

(4)

α _acc←α _acc×β _i+(1-α _acc×β _i)×α _i

Wherein, C _AccBe accumulation color value, α _AccBe accumulation opacity value, C _iAnd α _iBe respectively current color value and opacity value, β _iFor characteristic is strengthened rendering parameter; In sampling process, if current sampling point belongs to characteristic area, and the characteristic threshold value function of current characteristic area satisfies condition, then β _i=1-(s _i-s _I-1), otherwise β _i=1.0;

E) when the light accumulation finishes, carry out the luminance balance of final image, carry out the Tone attenuation operations, shown in formula (3)

$C_{\mathrm{new}}=C_{\mathrm{cur}}\×\frac{\ln (C_{\mathrm{cur}}+\mathrm{\λ})}{\ln (C_{\max }+\mathrm{\λ})}---\left(3\right)$

C wherein _NewFor carrying out the color value after Tone decays, C _CurBe the accumulation color value of current sampling light, C _MaxBe the maximum brightness value that equipment can show, λ is user-defined brightness decay coefficient, is used to control the brightness of final image color;

Tissue signature in the medical volume data is drawn fast and clearly.

2. the local feature intensive aspect method for drafting towards medical volume data as claimed in claim 1 is characterized in that: in the said step d), carry out illumination calculation, adopt the Phong illumination model, with the gradient of the current sampling point surface normal as illumination, utilize formula

C _sh＝(k _a+k _d(N·L))×C+k _s(N·H) ⁿ (2)

Calculate new color value and participate in color accumulation, wherein k _a, k _d, and k _sBe respectively surround lighting, scattered light and Gao Guang influence parameter, and N represents the direction of the normal vector of current sampling point; Normalization gradient vector by current sampling point representes that L represents the direction of incident ray, is confirmed by the position of current sampling point and the position of light source; H represents the half-angle between reflected light and the normal vector, is used for calculating Gao Guang, and n is high spectrum number; Be used for controlling scope and the brightness of Gao Guang, C represents the color value of current sampling point, in direct volume drawing, is specified by transition function.

Images