CN1142511C - Medical infrared heat image analysis method based on autoregressive slip mean spectrum analysis - Google Patents

Abstract

The present invention belongs to the field of medical infrared chart computer-assisted predication. Based on autoregressive moving average spectrum analysis, the present invention carries out the modern spectrum analysis by multiple thermal charts in the process of temperature change, a response process of temperature change for a pathologic change region is represented in a spectrum curve form by spectrum parameters, and the temperature response characteristics of the pathologic change region is quantitatively represented by a series of wave peaks and wave troughs of wavelets. The representation method of a characteristic curve deeply discloses inherent characteristics of diseases, the inherent characteristics are displayed in a direct mode so as to be helpful to determine and distinguish different disease characteristic parameters, the diagnose fuzzyness is reduced, and the diagnose reliability is increased.

Description

Medical infrared chart analytical approach based on the autoregressive moving average analysis of spectrum

Technical field the invention belongs to medical infrared chart computer-aided diagnosis technical field, particularly the autoregressive moving average spectral analysis method is applied to the analysis of medical infrared chart.

The background technology autoregressive moving-average model is the novel spectral analysis method based on traditional Fourier transform and filtering theory.From modern autoregressive moving average spectrum produce two during the last ten years, in numerous field, all obtained good application, as in fields such as engineering, medical science, meteorology, agricultures.In medical domain, use the autoregressive moving average spectrum cardiogram, electroencephalogram, electromyogram are carried out signal analysis; In the signal communication sphere, be used for processing to the identification of voice and voice with synthetic; Aspect earthquake prediction, be used as the modeling of seismic signal; In the fault diagnosis of mechanical system, be used for carrying out the parameter recognition of mechanical system; Aspect military input, autoregressive moving average spectrum method of estimation has also obtained ripe application, as to detection of radar and sonar signal or the like.

Autoregressive moving-average model is a kind of effective time series analysis model.Viewpoint from analysis of spectrum, the autoregressive moving average analysis of spectrum the record time series data more in short-term, can produce high-resolution spectrum estimates, method itself is a data adaptive, its special case autoregressive model and moving average model have their own characteristics each in the systematic parameter identification of carrying out time series analysis.With the system that AR Model Identification is come out, be the system of a full limit, the crest of the power Spectral Estimation of carrying out is accurate on this basis; And the system that the moving average model identification is come out is the system at a full zero point, and the trough of the power spectrum that estimates on this basis is accurate; Autoregressive moving-average model has then been concentrated the strong point of autoregressive model and moving average model, and the performance of the each side of the estimated power spectrum that comes out is all relatively good.

Time series is meant one group of data of arranging in chronological order, mainly is meant one group of orderly random data.They are with producing them, can be described as that the objective things of system are associated.Therefore, sequence is just reflecting the dynamic changing process of this system sometime, is exactly the relevant output or the response of this system, and the mode that connects each other with the external world of this system.

Medical infrared chart is the reflection in body surface temperature field, is to utilize infrared video camera, according to the Planck principle: ${\mathrm{\ω}}_{\mathrm{\λ}}=\frac{C_1}{{\mathrm{\λ}}^5}\frac{1}{\exp (c_2/\mathrm{\λT})-1}$ (ω _λ---corresponding to the emittance of wavelength X, λ---radiation energy wavelength, T---the absolute temperature of energy emission body, C ₁C ₂---constant) calculate from the human infrared radiation energy meter that measures.According to medico physics, the metabolism of the human body overwhelming majority all is heat release, and major part radiate by skin, human body pathology occurs or when unusual, at first can be on the body surface temperature field embodying to some extent, is exactly to analyze the process that the human body temperature field changes based on the medical diagnosis of medical infrared chart.

Existing medical infrared chart analytic system is mainly obtained part (infrared video camera) by thermal map and thermal imagine analysis software two parts are formed.Thermal map obtains part, and what adopt mostly is the infrared video camera of optical mechaical scanning formula, although sweep velocity is slower, its spatial resolution and temperature resolution are all higher, can satisfy the requirement of medical diagnosis.The thermal imagine analysis part, the analytical approach that is provided, comparatively simple mostly, only be confined to analyze the single width infrared chart, carry out some simple temperature statistics (as the temperature histogram of appointed area), can not analyze the pathological information that is comprised in the medical infrared chart fully in general, the reliability of diagnosis is not high.Such problem why occurring is that difference is bigger because the body surface feature varies with each individual.The performance of focus on thermal map also has very big difference, often can not analyze its thermal map feature simply with the single width thermal map for some recessiveness, potential disease, thereby can not provide strong analysis result and pathology information to auxiliary diagnosis.

Summary of the invention the objective of the invention is to overcome the weak point of prior art, the application of a kind of modern spectrum analysis method aspect medical infrared chart analysis is provided, obtain the response process of human body surface to be measured zone by several infrared charts to temperature variation, the time domain temperature data that representative is changed response process is done the estimation of modern spectrum parameter then, thereby variation with shell temperature, especially the variation with the dynamic process shell temperature is converted into the spectrum signature parameter, these parameters and then can be provided for the data of clinical diagnosis.

The present invention proposes a kind of medical infrared chart analytical approach based on the autoregressive moving average analysis of spectrum, may further comprise the steps:

1) calculates the temperature variation time domain data of the medical infrared chart of several diseased regions be transfused to, generate the canonical equation of autoregression (n) model and running mean (m) model, wherein n and m are respectively the order of autoregressive model and moving average model, and said order n and m determine according to model applicability test criterion;

2) find the solution the canonical equation of autoregression (n) model and represent the limit parameter of the temperature variation response process of this diseased region institute measuring point, in the canonical equation with described parameter substitution running mean (m) model, find the solution the parameter at zero point of the temperature variation response process of representing this diseased region institute measuring point, zero, the limit parameter that estimate with this model have promptly characterized the described response characteristic to thermal stimulus on the described diseased region;

3) (therefrom autoregressive moving average (n, m) power spectrum curve of the temperature variation response process of this diseased region institute measuring point represented in calculating for n, the m) canonical equation of model to generate autoregressive moving average by autoregression (n) model and running mean (m) model;

4) parameter of autoregression (n) model that calculates is done cross-correlation coefficient with autoregression (n) model parameter in the standard pathology template and calculate, will characterize the relevant degree of this temperature changing process as auxiliary diagnosis index one to the possibility of this disease generation;

5) parameter of running mean (m) model that calculates is done cross-correlation coefficient with running mean (m) model parameter in the standard pathology template and calculate, will characterize the relevant degree of this temperature changing process as auxiliary diagnosis index two to the possibility of this disease generation;

6) thus index one and index two taken all factors into consideration for clinical diagnosis provide the shutoff data.

(n, m) the order n of model and m all can be 10 for the order of the order of said autoregression (n) model, running mean (m) model and autoregressive moving average.

(n, m) computing method of the described parameter of the computing method of power spectrum curve and diseased region and described power spectrum curve are identical for autoregression (n) model parameter of above-mentioned standard pathology template and running mean (m) model parameter and autoregressive moving average.

The theoretical foundation of the inventive method is:

As mentioned above, modern spectrum analysis method especially autoregressive moving average Zymography is comprising that many frequency domains that medical information is handled have obtained application, the advantage of this method is to utilize limited time series data to come analytic target is set up model, and estimate the parameter of model, utilize parameter to disclose the inward nature of measurand.Therefore, the present invention is incorporated into the autoregressive moving average spectral analysis method analysis and the diagnostic field of medical infrared chart.

Any a part of human body all is a system, and the systematic parameter of oneself is arranged, and the systematic parameter that the zone of identical pathology takes place necessarily has its similar composition.The process of diagnosis is analyzed the process that the standard pathology systematic parameter in regional pathological parameter to be measured and the standard pathology template compares exactly.And for the calculating of systematic parameter, the way that comparative maturity has been arranged on engineering, can be to diseased region environmental stimuli in addition, analyze its response data, utilize existing time series analysis method, estimate its systematic parameter (being generally the systematic parameter of zero/limit shape), and show its change procedure in the mode of curve with the form of modern autoregressive moving average spectrum.

The relevant basic theory of the applied autoregressive moving average spectrum of the present invention is described as follows: linear regression model (LRM):

A stochastic variable y _tDepend on a plurality of stochastic variable x _1t, x _2t, x _3t... x _Nt:

y _t=β ₁x _1t+ β ₂x _2t+ ... β _nx _Nt+ ε _tε _t～NID (0, σ ²) β wherein _tBe the parameter of model, ε _tIt is the residual error of model.Utilize least square method to estimate β _t: $S=\underset{t=1}{\overset{N}{\mathrm{\Σ}}}{\widehat{\mathrm{\ϵ}}}_1^2$ $=\underset{t=1}{\overset{N}{\mathrm{\Σ}}}\[y_t-({\mathrm{\β}}_1^{^}{x}_{1t}+{\mathrm{\β}}_2^{^}{x}_{2t}+...{\mathrm{\β}}_n^{^}{x}_{\mathrm{nt}}){\]}^2$ To β _tDifferentiate, making it is zero: $\frac{\∂S}{\∂{\mathrm{\β}}_t}=0---t=\mathrm{0,1,2,3}...N.$ The solving equation group obtains the system linearity regression model: $\left[\begin{array}{c}{y}_1\\ y_2\\ M\\ y_n\end{array}\right]=\left[\begin{array}{c}{\mathrm{\β}}_1\\ {\mathrm{\β}}_2\\ M\\ {\mathrm{\β}}_n\end{array}\right]\left[\begin{array}{cccc}{x}_{11}& x_{21}& \mathrm{\Λ}& x_{n1}\\ x_{21}& x_{22}& \mathrm{\Λ}& x_{n2}\\ \mathrm{\Λ}& \mathrm{\Λ}& \mathrm{\Λ}& \mathrm{\Λ}\\ x_{1n}& x_{2n}& \mathrm{\Λ}& x_{\mathrm{nn}}\end{array}\right]+\left[\begin{array}{c}{\mathrm{\ϵ}}_1\\ {\mathrm{\ϵ}}_2\\ M\\ {\mathrm{\ϵ}}_n\end{array}\right]$

y＝βx+ε

Yt is made up of two parts, and a part is the determinacy part, depends on the linear combination of each independent variable fully

It is the mathematical expectation of Yt; Another part is a random partial, is totally independent of each x _1t, by white noise ε _tDecision is to cause Yt to be called the reason of stochastic variable.Autoregressive moving average (n, m) model:

According to the thought of multiple linear regression, (n, m) model can be expressed as an autoregressive moving average: x _t= ₁x _T-1+ ₂x _T-2+ ...+ _nx _T-n-θ ₁α _T-1-θ ₂α _T-2-...-θ _mα _T-m+ α _t ${\mathrm{\α}}_t~\mathrm{NID}(0,{\mathrm{\σ}}_{\mathrm{\α}}^2)$ N, m represent the order of autoregression part and running mean part respectively; _i(i=1,2 ..., n), θ _i(i=1,2 ..., m) be the model parameter of each several part; x _i(i=t-1, t-2 ... t-n) discrete-time series for sampling; α _t(i=t, t-1 ... t-m) for having zero-mean and variances sigma _α ²White noise.N, whether the order of m is applicable to model, need verify.Concrete checkout procedure is as follows: (1) detects α _tWhether be white noise, detect α _tWhether with α _T-1, α _T-2... irrelevant.Calculate α _tCoefficient of autocorrelation: ${\mathrm{\ρ}}_{\mathrm{\α},n}=\frac{\underset{t=2}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_t{\mathrm{\α}}_{t-n}}{\underset{t=2}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_t^2}$ If ρ _{α, n}-＞0, think that then order is applicable to model.(2) check α _tWhether with x _T-2, x _T-3... irrelevant.Calculate α _tWith x _tCross-correlation coefficient: ${\mathrm{\ρ}}_{\mathrm{\αx},n+m}=\frac{\underset{t=3}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_t{x}_{t-3}}{\sqrt{\left(\underset{t=n+2}{\overset{N}{\mathrm{\Σ}}}{\mathrm{\α}}_t^2\right)\left(\underset{t=n+2}{\overset{N}{\mathrm{\Σ}}}{x}_{t-n-1}^2\right)}}$ If ρ _{α x, n+m}-＞0, think that then order is suitable for.

After model order is determined, autoregressive moving average (n, m) model also can be expressed as: $x\left(t\right)=-\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{a}_{k}x(t-k)+\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{b}_{k}u(t-k)$ A wherein _k=- _k(k=1,2 ... n), b _k=-β _k(k=1,2 ... m).This model is made up of two parts, and one is n rank autoregressive models $x\left(t\right)=-\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{a}_{k}x(t-k),$ Another is m rank moving average models $x\left(t\right)=\underset{k=1}{\overset{m}{\mathrm{\Σ}}}b\left(k\right)u(t-k).$ Autoregressive model: the basal expression formula of n rank autoregressive model is: $x\left(t\right)=-\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{a}_{k}x(t-k)+u\left(t\right),$ With multiply by x (t+m), averaged in the equation both sides, obtain: $r_x\left(m\right)=E\left\{x\right(t\left)x\right(t+m\left)\right\}=E\left\{\right[-\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{a}_{k}x(t+m-k)+u(t+m)\left]x\right(t\left)\right\}$ Promptly $r_x\left(m\right)=-\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{a}_k{r}_t(t-k)+r_{\mathrm{xt}}\left(m\right)$

U (t) is that variance is σ ²White noise, the unit sample respo of establishing this system is h (t), so: $r_{\mathrm{xu}}\left(m\right)=E\left\{u\right(t+m\left)x\right(t\left)\right\}=E\left\{u\right(t+m\left)\underset{k=0}{\overset{\∞}{\mathrm{\Σ}}}h\right(k\left)u\right(t-k\left)\right\}={\mathrm{\σ}}^2\underset{k=0}{\overset{\∞}{\mathrm{\Σ}}}h\left(k\right)\mathrm{\δ}(m+k)={\mathrm{\σ}}^{2}h(-m)$ According to the transform definition, ${\underset{z->\∞}{\mathrm{Lim}}}^{H\left(z\right)=h\left(0\right)},z->\∞$ The time h (0)=1, so its matrix equation of deriving is: $\left[\begin{array}{ccccc}{r}_x^{\·}\left(0\right)& r_x\left(1\right)& r_{xx}\left(2\right)& ...& r_x\left(n\right)\\ r_x\left(1\right)& r_x\left(0\right)& r_x\left(1\right)& ...& r_x(n-1)\\ r_x\left(2\right)& r_x\left(1\right)& r_x\left(0\right)& ...& r_x(n-2)\\ ...& ...& ...& ...& ...\\ r_x\left(n\right)& r_x(n-1)& r_x(n-2)& ...& r_x\left(0\right)\end{array}\right]\left[\begin{array}{c}1\\ a_1\\ a_2\\ ...\\ a_n\end{array}\right]=\begin{array}{c}\left[\begin{array}{c}{\mathrm{\σ}}^2\\ 0\\ 0\\ ...\\ 0\end{array}\right]\end{array}$ Calculate this equation, just obtain the model parameter of n rank autoregressive model.Pass through the power spectrum of following formula sequence computing time again: $p_x\left(e^{\mathrm{jw}}\right)=\frac{{\mathrm{\σ}}^2}{{|1+\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{\mathrm{\α}}_k{e}^{-\mathrm{jkw}}|}^2}$ Moving average model:

The basic representation of m rank moving average model: $x\left(t\right)=u\left(t\right)+\underset{k=1}{\overset{m}{\mathrm{\Σ}}}b\left(k\right)u(t-k)$ $H\left(z\right)=1+\underset{k=1}{\overset{m}{\mathrm{\Σ}}}b\left(k\right)z^{-k}$ $r_x\left(t\right)=E\left\{\right[u(t+n)+\underset{k=t}{\overset{m}{\mathrm{\Σ}}}b\left(k\right)u(t+n)\left]x\right(n\left)\right\}=\underset{k=0}{\overset{m}{\mathrm{\Σ}}}b\left(k\right)r_{\mathrm{xt}}(t-k)$

b(0)＝1。Because:

r _x(t-k)＝E{x(n)u(n+t-k)}＝σ ²h(k-t)

H (i)=b (i), i=0,1,2 ..., m just can obtain the canonical equation of moving average model:

Calculate this equation, can obtain the model parameter of m rank moving average model.And calculate its power spectrum by following formula: $p_x\left(e^{\mathrm{jw}}\right)={\mathrm{\σ}}^2\underset{k=-m}{\overset{m}{\mathrm{\Σ}}}{r}_x\left(k\right)e^{-\mathrm{jwk}}$

The algorithm of integrated use autoregressive model and moving average model, obtain autoregressive moving average (n, m) canonical equation of model is: By finding the solution the following formula equation, can obtain the n rank autoregressive model parameter and the m rank moving average model parameter of model, and then the power spectrum that can calculate this sequence is: $P_x\left(e^{\mathrm{jw}}\right)=\frac{{\mathrm{\σ}}^2{|1+\underset{k=1}{\overset{m}{\mathrm{\Σ}}}{b}_k{e}^{-\mathrm{jwk}}|}^2}{{|1+\underset{k=1}{\overset{n}{\mathrm{\Σ}}}{a}_k{e}^{-\mathrm{jwk}}|}^2}$

Outstanding contributions of the present invention are:

At first the autoregressive moving average spectral analysis method is incorporated into the analysis and the diagnostic field of medical infrared chart, the application of a kind of modern spectrum analysis method aspect medical infrared chart analysis is provided, obtain the response process of human body surface to be measured zone by several infrared charts to temperature variation, the time domain temperature data that representative is changed response process is done the estimation of modern spectrum parameter then, thereby variation with shell temperature, especially the variation with the dynamic process shell temperature is converted into the spectrum signature parameter, these parameters and then can be provided for the data of clinical diagnosis.

Existing thermal imagine analysis method all is based on the analysis to the single width thermal map, utilizes the asymmetry and the temperature difference of the Temperature Distribution of lesion region and normal region to carry out medical diagnosis on disease.The temperature difference data of certain illness diagnosis of this single conduct since the difference of patient's physiological condition and the difference of pathology happening part often be difficult to determine, still there is not unified standard, therefore normal clinical experience by means of the doctor, thus subjectivity and the inexactness diagnosed introduced more.

Opposite with prior art, method of the present invention is carried out modern spectrum analysis by several thermal maps to temperature changing process, the measured body table section is shown as the form of spectral curve to the response process of temperature variation with the spectrum parameter, the temperature-responsive feature of lesion region crest and the trough with a series of wavelets showed quantitatively.The method for expressing of this characteristic curve more comprehensively, has more profoundly disclosed the inherent inherent feature of disease, and show in more intuitive mode, help determining and differentiation to the various disease characteristic parameter, also help to reduce the ambiguity that adopts single temperature difference standard to diagnose, thereby improved the reliability of diagnosis.

Characteristics of the present invention are:

(1) adopt autoregressive moving-average model to carry out the temperature variation characteristic parameter that the system features parameter estimation is obtained body surface.

(2) utilize limited time series data to come analytic target is set up model, and estimate the parameter of model, utilize parameter to disclose the inward nature of measurand.

(3) adopt modern autoregressive moving average spectrum parameter estimation method to analyze infrared chart.The variation of shell temperature parameter can be converted into the power spectrum characteristic curve, provide pathological data more accurately thereby can be diagnosis.

(4) adopt modern autoregressive moving average spectrum parameter estimation method can disclose response process and the inherent pathological information thereof of lesion region more fully, thereby might provide the data of usefulness the diagnosis of some latent disease to temperature variation.

Brief Description Of Drawings:

Fig. 1 is that the thermal map autoregressive moving average spectrum of a kind of embodiment of the present invention is estimated process flow diagram.

Embodiment

Describe method of the present invention and characteristics in detail below in conjunction with embodiment and accompanying drawing.

In an embodiment of method of the present invention, adopt the time series analysis means, introduce environmental stimuli--thermal stimulus, and thermal stimulus asked for its response, and response data is done the estimation of modern power spectrum parameters, thereby with the variation of shell temperature, especially the variation of dynamic process shell temperature is converted into the spectrum signature parameter.These characteristic parameters and then the pathological parameter template that can be used to standard (healthy people) compare, thereby the clinical diagnosis that can be the medical personnel provides necessary pathological analysis data.

For obtaining temperature analysis data required for the present invention, take following step to absorb the human body infrared chart:

(1) patient sat quietly 5 minutes in room temperature is 25 degrees centigrade room.

(2) select suspicious lesion region,, make this zone local heating with 45 degrees centigrade hot-water bottle hot compresses two minutes.

(3) remove hot-water bottle, with medical infrared thermal imaging system (model: DH98) take 64 thermal maps continuously, write down temperature damping's process in this suspicious lesions zone.

(4) gather the same point in suspicious lesions district among 64 figure automatically, form the time domain data of analyzing the temperature damping's curve that needs.

Fig. 1 is that the thermal map autoregressive moving average spectrum of one embodiment of the invention is estimated process flow diagram, and the autoregressive moving-average model that is adopted is one 10 rank autoregressions, 10 rank moving average models.At first calculate the cross correlation function of the described time domain data that is transfused to, generate the canonical equation of autoregression (10) and running mean (10).Find the solution the limit parameter of the canonical equation and the diseased region of autoregression (10).In the canonical equation with described parameter substitution running mean (10), find the solution the parameter at zero point of diseased region.According to the zero limit parameter of the diseased region of obtaining, calculate autoregressive moving average (10, the 10) power spectrum curve of diseased region.The parameter of autoregression (10) model that calculates is done cross-correlation coefficient with autoregression (10) parameter in the standard pathology template calculate, obtain auxiliary diagnosis index one.The parameter of running mean (10) model that calculates is done the calculating of cross-correlation coefficient with running mean (10) parameter in the standard pathology template, obtain auxiliary diagnosis index two.Thereby index one and index two taken all factors into consideration for clinical diagnosis provides relevant data.Wherein the computing method of the described parameter of the computing method of the autoregression of standard pathology template (10) parameter and running mean (10) parameter and autoregressive moving average (10,10) power spectrum curve and diseased region and described power spectrum curve are identical.

Claims (2)

1, a kind of medical infrared chart analytical approach based on the autoregressive moving average analysis of spectrum comprises: thermal stimulus is carried out in human body surface to be measured zone; Obtain several infrared charts of this warm change process in body surface zone; Temperature data to these several infrared charts carries out analysis of spectrum to obtain the parameter that characterizes this body surface zone temperature profile, is used as the clinical diagnosis data; It is characterized in that,

From several infrared charts, obtain the time domain temperature data of human body surface area relative dynamic process shell temperature to be measured;

Utilize this limited time series temperature data to come analytic target is set up model, this time domain temperature data is made the modern spectrum parameter estimation;

The variation of this dynamic process shell temperature is converted into the spectrum signature parameter, obtains the frequency response process of lesion region temperature variation;

The spectrum signature parameter of variation of temperature is done correlation computations with the spectrum signature parameter of standard pathology template, and its result is provided for the data of clinical diagnosis;

Described modern spectrum parameter estimation is an autoregressive moving average spectrum parameter estimation, may further comprise the steps:

1) calculates the temperature variation time domain data of the medical infrared chart of several diseased regions be transfused to, generate the canonical equation of autoregressive model and moving average model, determine the order of autoregressive model and moving average model according to model applicability test criterion;

2) find the solution the canonical equation of autoregressive model and represent the limit parameter of the temperature variation response process of this diseased region institute measuring point, in the canonical equation with described parameter substitution moving average model, find the solution the parameter at zero point of the temperature variation response process of representing this diseased region institute measuring point, zero, the limit parameter that estimate with this model have promptly characterized the described response characteristic to thermal stimulus on the described diseased region;

3), therefrom calculate the autoregressive moving average power spectrum curve of the temperature variation response process of representing this diseased region institute measuring point by the canonical equation of autoregressive model and moving average model generation autoregressive moving-average model;

4) parameter of the autoregressive model that calculates is done cross-correlation coefficient with the autoregressive model parameter in the standard pathology template and calculate, degree that this is relevant is as the auxiliary diagnosis index one to the possibility of this disease generation;

5) parameter of the moving average model that calculates is done cross-correlation coefficient with the moving average model parameter in the standard pathology template and calculate, degree that this is relevant is as the auxiliary diagnosis index two to the possibility of this disease generation;

6) comprehensively as the related coefficient of auxiliary diagnosis index one with as the related coefficient of auxiliary diagnosis index two, obtain total weighted correlation coefficient, the size of this coefficient has promptly been represented the size of the possibility that this disease produces.

2, the medical infrared chart analytical approach based on the autoregressive moving average analysis of spectrum as claimed in claim 1 is characterized in that the order of the order of said autoregressive model, the order of moving average model and autoregressive moving-average model is 10.