CN104574385A - Myocardial fatty acid metabolism quantitative detection method based on dynamic PET image - Google Patents

Abstract

The invention provides a myocardial fatty acid metabolism quantitative detection method based on a dynamic PET (positron emission tomography) image, and aims to solve the problem that quantitative detection is inaccurate due to noise of a PET scanning dynamic image and the influence of overflow and partial volume effects when quantitative detection is carried out on myocardial fatty acid metabolism by adopting the PET scanning dynamic image. A noninvasive technology for acquiring an input function from the image is adopted; dynamic image data acquired by PET is preprocessed by data processing measures such as interpolation and filtration to improve the fitting precision of model data; correction coefficients of the partial volume effect and the overflow effect are added into the model, and parameter estimation is carried out on the input, output and correction coefficients during data fitting, so that quantitative detection on myocardial fatty acid metabolism is completed; the method has a noninvasive, quick and accurate detection effect.

Description

A kind of myocardial fatty acid metabolic quantified detection method based on dynamic PET image

Technical field

The invention belongs to dynamic modeling and the quantitative analysis tech field of positron emission tomography, relate to a kind of myocardial fatty acid metabolic quantified detection method based on dynamic PET image.

Background technology

PET (positron emission tomography) is a kind of medicine imaging technique, can non-wound ground by the structure of the in-vivo tissue originally do not observed, organ or function intuitively visual present.The foundation of the radiopharmaceutical dynamic model of PET be exactly by radiopharmaceutical (or being called radioactive indicator) from being injected in vivo the process having participated in a series of dynamic biochemical and reacted, portray out quantitatively by the mode of mathematical model.Glucose and fatty acid are the energy sources of myocardial cell, by PET quantitative analysis and detection, the quantitative target that glucose metabolic rate, fatty acid metabolism rate and oxygenation efficiency etc. have biological meaning can be calculated, the physiological function of organ is tested and assessed accurately, the early diagnosis of the cardiovascular disease such as tumor, coronary heart disease, therapeutic scheme are formulated, observation of curative effect and efficacy of new drug detection etc. have important clinical effect, can be used as the analytical tool of a kind of high sensitivity, pinpoint accuracy.

Due to the restriction of the restriction and PET self image-forming principle that are subject to hardware, the picture quality obtained after PET image reconstruction is poor, spatial resolution is lower, causes the radioelement activity obtained from image to there is PV (partial volume) effect and SP effect (spilling).For dynamic pet imaging, in order to catch the change of complete nuclide concentration, the change in tissue of incipient stage nuclide concentration is very fast, and it is very short that the sampling interval is chosen, and can cause like this and meets number of photons deficiency thus bring the reconstruction image scanning early stage to have very large noise; Simultaneously because heart Repiration can produce the transposition of partial of scanogram cardiac, the data fluctuations that making samples obtains is large, there is a lot of dummy values point, produces a very large impact accurately weighing metabolic coefficients.This problem, especially particularly serious in the experiment of laboratory research small animal position emission tomography (PET), will precision and the confidence level of experimental data be had a strong impact on.

Present stage, the radiopharmaceutical dynamic model many employings kinetic parameter model based on PET is studied.Be detected as example with FDG (glucose) metabolism, employing be that the kinetics compartment model of three Room four parameters carries out quantum chemical method, main approaches has:

1, PET dynamic acquisition sequential images is utilized to obtain the data of FDG change in local organization, again by multipoint acquisition arterial blood, obtain the FDG delta data in blood plasma, FDG dynamic data in tissue and blood plasma is substituted in kinetic model, with nonlinear least square method or the every metabolic rate of generalized linear least square fitting.But because needs tremulous pulse is taken a blood sample continuously, to patient, there is traumatic and certain risk; In laboratory small animal experiment, arterial blood drawing also tool acquires a certain degree of difficulty.

2, by selecting the method for area-of-interest to obtain input and output data from high-resolution PET image, being carried out curve fitting by weighting nonlinear least squares method, obtaining four parameters and the weight coefficient of kinetic model.The method does not need continuous blood sampling, but does not adopt the method corrected to eliminate the impact of PV, SP effect, and acquired results accuracy is not high.

FA (fatty acid) metabolic process is more more complicated than FDG metabolism.The tracer used for reflection FDG metabolic process is clinically ¹⁸f-FDG (fluorodeoxyglucose), FA metabolism then selects tracer ¹¹c-Palmitate (cetylate), due to the radioelement of backbone ¹¹c has ratio ¹⁸the positron emission scope that F is larger, the spatial resolution that PET is scanned is deteriorated, and cause, from blood plasma to seriously polluted at the SP of preliminary sweep point the chamber of tissue, therefore needs the impact considering emphatically SP effect in FA metabolism detects.FA with FDG in vivo metabolic process is different: FDG enters histiocyte from blood, by hexokinase phosphorylation in cell, finally by cell capture, therefore adopts the kinetic model of three chamber 4 parameters in research; FA enters after histiocyte from blood, metabolism comprises reversible esterification and irreversible beta oxidation two parts, therefore adopts the kinetics compartment model of four chamber 5 parameters to study.Present stage does not also retrieve the relevant method scanning the myocardial fatty acid metabolic quantification detection of dynamic image based on PET.

Summary of the invention

The object of the invention is for solve adopt PET scan dynamic image carry out myocardial fatty acid metabolic quantification detect time, cause quantizing to detect inaccurate problem due to the noise of PET scanogram and the impact of spilling and partial volume effect, propose a kind of four Room kinetic models according to fatty acid metabolism, for PET dynamic scan data, carried out the data fitting of multiparameter, dual output by the method such as interpolation, filtering, thus the method for detection by quantitative is carried out to myocardial fatty acid metabolic indices.

The present invention is achieved through the following technical solutions:

Based on a myocardial fatty acid metabolic quantified detection method for dynamic PET image, comprise the following steps:

Step one, use the method drawing area-of-interest from the PET image that dynamic scan obtains, obtain respectively putting the radioactive concentration sequence in blood and cardiac muscular tissue each sweep time, thus draw in blood time m-activity curve and cardiac muscular tissue in time m-activity curve, as detect fatty acid metabolism time four Room kinetic models input and output;

Step 2, according to the time m-activity curve in the blood obtained in step one and cardiac muscular tissue, adopt the method for low-pass digital filter, dummy values point during removal in m-activity curve also reduces high-frequency noise;

Step 3, rise and decline very fast due to the nuclide concentration gathering incipient stage blood and tissue, according to the time m-activity curve obtained after step 2 filtering, interpolation arithmetic is carried out near the peak value of curve, increase the resolution in the very fast part of data variation speed, m-activity curve when obtaining the blood after interpolation and organize;

Step 4, adopt four Room kinetic models of myocardial fatty acid metabolic, carry out after Laplace transformation solves to the differential equation group of four Room kinetic models, then carry out corresponding inverse transformation, obtain by model metabolic coefficients k ₁-k ₅cardiac muscular tissue's fatty acid metabolism concentration with input function Using Convolution, exports in this, as model;

Step 5, use the function of image sources as input-output function replace actual plasma and tissue time m-activity curve time, consider, between the two due to the difference that the partial volume effect of spills-over effects and blood vessel radioactivity causes, to introduce parameter S _mb, S _bm, r _m, r _b, the input and output function that step 4 obtains is corrected, wherein r _m, r _bthe recovery coefficient to PV effect, S _mb, S _bmthat obtain the plasma concentration after correcting as mode input function, the tissue concentration after correction is as model output function by blood plasma to the spilling coefficient of cardiac muscular tissue and cardiac muscular tissue to the spilling coefficient of blood plasma respectively;

Model input-output function after step 6, the blood obtained according to step 3 and correction that when organizing, m-activity curve and step 5 obtain, adopt sum of square of deviations minimum principle, actual measurement data and pattern function are carried out to the curve fitting of multiparameter simultaneously, obtain the metabolic coefficients k of Myocardial Fatty Acids four Room kinetic model ₁-k ₅, and calculate cardiac muscular tissue to the overall utilization (MFAU) of fatty acid and fatty acid esterification (MFAE), oxidation (MFAO) parameter, complete and the quantification of fatty acid metabolism is detected.

Beneficial effect of the present invention:

The present invention contrasts prior art, according to the metabolic characteristic of Myocardial Fatty Acids, can adopt the hurtless measure noninvasive technology from Image Acquisition input function; By the data processing means such as interpolation, filtering, pretreatment is carried out to the dynamic image data that PET obtains and improve model data fitting precision; Add the correction coefficient to partial volume effect and spills-over effects in a model, and when data fitting, parameter estimation is carried out to input, output and correction coefficient simultaneously, complete and the quantification of myocardial fatty acid metabolic is detected, there is hurtless measure, fast and accurately Detection results.

Accompanying drawing explanation

Fig. 1 is the overview flow chart of detection algorithm of the present invention.

Fig. 2 is four compartmental kinetics model structure figure of fatty acid metabolism.

Detailed description of the invention

Elaborate below in conjunction with the embodiment of accompanying drawing to the inventive method.

As shown in Figure 1, a kind of myocardial fatty acid metabolic quantified detection method based on dynamic PET image of the present invention, its concrete steps comprise:

Step one, obtain the tissue of live body heart and the time m-activity curve of blood plasma from dynamic PET images.In live body injection can reflecting myocardium fatty acid metabolism process tracer ( ¹¹c-Palmitate), Positron Emission Computed Tomography equipment is used to obtain PET dynamic image through scanning after a while; By selecting the area-of-interest of blood and tissue from image, obtain the radioactive concentration sequence in point sequence sweep time, each time point blood plasma and the radioactive concentration sequence in tissue, thus draw the time m-activity curve of blood and tissue, as the input and output of four compartment models.

Step 2, adopt digital filtering remove blood plasma and tissue time m-activity curve in dummy values point and reduce noise.Determine the minimum sampling interval according to PET image acquisition protocols, design meets the low pass filter of maximum sample frequency, the denoising of m-activity curve and smoothing processing when to complete pair.

Step 3, carry out interpolation at the data variation vicinity fast that rises, improve data fitting precision.Determine the peak value of sequence, near peak value, cubic spline interpolation is carried out to sequence, obtains new blood plasma and tissue radioactivity concentration sequence:

$\begin{array}{c}{\mathrm{PET}}_{\mathrm{IDIF},i}=[{\mathrm{PET}}_{\mathrm{IDIF},1},...,{\mathrm{PET}}_{\mathrm{IDIF},\max -1},{\mathrm{PET}}_{\mathrm{IDIF},k_1},...,{\mathrm{PET}}_{\mathrm{IDIF},k_n},{\mathrm{PET}}_{\mathrm{IDIF},\max },{\mathrm{PET}}_{\mathrm{IDIF},k_{n+1}},...,\\ {\mathrm{PET}}_{\mathrm{IDIF},k_{2n}},{\mathrm{PET}}_{\mathrm{IDIF},\max +1},...]\end{array};$

$\begin{array}{c}{\mathrm{PET}}_{\mathrm{myo},i}=[{\mathrm{PET}}_{\mathrm{myo},1},...,{\mathrm{PET}}_{\mathrm{myo},\max -1},{\mathrm{PET}}_{\mathrm{myo},k_1},...,{\mathrm{PET}}_{\mathrm{myo},k_n},{\mathrm{PET}}_{\mathrm{myo},\max },{\mathrm{PET}}_{\mathrm{myo},k_{n+1}},...,\\ {\mathrm{PET}}_{\mathrm{myo},k_{2n}},{\mathrm{PET}}_{\mathrm{myo},\max +1},...]\end{array}.$

Step 4, four Room kinetics compartment models are adopted to carry out the research of myocardial fatty acid metabolic.According to the four Room kinetics compartment models of Fig. 2, obtaining its differential equation group is:

$\left\{\begin{array}{c}\frac{{\∂q}_1}{\∂t}=F[\mathrm{Ca}\left(t\right)-\frac{q_1}{V}]+k_2{q}_2-k_1{q}_1\\ \frac{{\∂q}_2}{\∂t}=k_1{q}_1+k_4{q}_3-(k_2+k_3+k_5)q_2\\ \frac{{\∂q}_3}{\∂t}=k_3{q}_2-k_4{q}_3\\ \frac{{\∂q}_4}{\∂t}=k_5{q}_2-\frac{F}{V}*q_4\end{array}\right.---\left(1\right)$

Wherein, q ₁~ q ₄represent the concentration of FA in each chamber respectively, kinetic parameter represents the velocity coefficient of mass exchange between chamber.

Total tracer concentration C in cardiac muscular tissue _tfor each chamber concentration sum:

C _T＝q ₁(t)+q ₂(t)+q ₃(t)+q ₄(t) (2)

Step 5, solve differential equation group, obtain exporting the corresponding relation with input.Adopt Laplace transform to solve to differential equation group, obtained the convolution relation that in time domain, model exports and inputs by anti-change:

$C_T\left(t\right)=F\·M\⊗C_a\left(t\right)---\left(3\right)$

Wherein,

$\begin{array}{c}M=\left[\frac{(A_4-B_1)((A_1-B_1)(A_2-B_1)+k_1(k_4-B_1)+k_1{k}_3)+k_1{k}_2(k_4-B_1)}{(A_4-B_1)(B_2-B_1)(B_3-B_1)}\right]e^{-B_{1}t}\\ +\left[\frac{(A_4-B_2)((A_1-B_2)(A_2-B_2)+k_1(k_4-B_2)+k_1{k}_3)+k_1{k}_2(k_4-B_2)}{(A_4-B_2)(B_1-B_2)(B_3-B_2)}\right]e^{-B_{2}t}\\ +\left[\frac{(A_4-B_3)((A_1-B_3)(A_2-B_3)+k_1(k_4-B_3)+k_1{k}_3)+k_1{k}_2(k_4-B_3)}{(A_4-B_3)(B_1-B_3)(B_2-B_3)}\right]e^{-B_{3}t}\\ +\frac{k_1{k}_2(k_4-A_4)}{(B_1-A_4)(B_2-A_4)(B_3-A_4)}{e}^{-A_{4}t}\end{array}$

A ₁~ A ₄, B ₁~ B ₃, by parameter k ₁~ k ₅and F, V composition.

Step 6, correction input-output function.During activity actual time using the data of image sources to replace in blood plasma and tissue, the difference that the spills-over effects (SP) between consideration blood plasma and tissue and partial volume (PV) effect of blood vessel radioactivity cause, introduces parameter S _mb, S _bm, r _m, r _b, input and output data are corrected, obtain the input and output function in realistic model:

Model _IDIF,i＝S _mbC _T(t)+r _bC _a(t) (4)

Model _myo,i＝r _mC _T(t)+S _bmC _a(t) (5)

Wherein, r _m, r _bthe recovery coefficient to PV effect, S _mb, S _bmto the spilling coefficient of cardiac muscular tissue and cardiac muscular tissue to the spilling coefficient of blood plasma respectively by blood plasma.

Step 7, the initial value setting parameter to be optimized and value bound, data fitting is carried out by sum of square of deviations O (p) minimum principle, tried to achieve the final value of each parameter when O (p) obtains minima by iteration based on internal maps Newton method, the account form of O (p) is as follows:

$O\left(p\right)=\underset{i=1}{\overset{n}{\mathrm{\Σ}}}[{({\mathrm{Model}}_{\mathrm{IDIF},i}-{\mathrm{PET}}_{\mathrm{IDIF},i})}^2+{({\mathrm{Model}}_{\mathrm{myo},i}-{\mathrm{PET}}_{\mathrm{myo},i})}^2]---\left(6\right)$

Step 8, according to optimizing the four Room kinetic parameter k obtained in step 7 ₁~ k ₅, every metabolic index of calculating myocardium fatty acid, thus obtain cardiac muscular tissue to the overall utilization (MFAU) of fatty acid and fatty acid esterification (MFAE), oxidation (MFAO) parameter:

$\mathrm{MFAE}=k_1{k}_3\left[\frac{{\mathrm{FC}}_a}{(k_2+k_5)(\frac{F}{V}+k_1)-k_1{k}_2}\right]---\left(7\right)$

$\mathrm{MFAO}=k_1{k}_5\left[\frac{{\mathrm{FC}}_a}{(k_2+k_5)(\frac{F}{V}+k_1)-k_1{k}_2}\right]---\left(8\right)$

MFAU＝MFAO+MFAE (9)

Thus achieve the quantification of myocardial fatty acid metabolic is detected.

Although describe embodiments of the present invention by reference to the accompanying drawings, to those skilled in the art, under the premise without departing from the principles of the invention, can also make some improvement, these also should be considered as belonging to protection scope of the present invention.

Claims (1)

1., based on a myocardial fatty acid metabolic quantified detection method for dynamic PET image, it is characterized in that, comprise the following steps:

Step one, use the method drawing area-of-interest from the PET image that dynamic scan obtains, obtain respectively putting the radioactive concentration sequence in blood and cardiac muscular tissue each sweep time, thus draw in blood time m-activity curve and cardiac muscular tissue in time m-activity curve, as detect fatty acid metabolism time four Room kinetic models input and output;

Step 2, according to the time m-activity curve in the blood obtained in step one and cardiac muscular tissue, adopt the method for low-pass digital filter, dummy values point during removal in m-activity curve also reduces high-frequency noise;

Step 3, rise and decline very fast due to the nuclide concentration gathering incipient stage blood and tissue, according to the time m-activity curve obtained after step 2 filtering, interpolation arithmetic is carried out near the peak value of curve, increase the resolution in the very fast part of data variation speed, m-activity curve when obtaining the blood after interpolation and organize;

Step 4, adopt four Room kinetic models of myocardial fatty acid metabolic, carry out after Laplace transformation solves to the differential equation group of four Room kinetic models, then carry out corresponding inverse transformation, obtain by model metabolic coefficients k ₁-k ₅cardiac muscular tissue's fatty acid metabolism concentration with input function Using Convolution, exports in this, as model;

Step 5, use the function of image sources as input-output function replace actual plasma and tissue time m-activity curve time, consider, between the two due to the difference that the partial volume effect of spills-over effects and blood vessel radioactivity causes, to introduce parameter S _mb, S _bm, r _m, r _bthe input and output function that step 4 obtains is corrected, wherein r _m, r _bthe recovery coefficient to PV effect, S _mb, S _bmthat obtain the plasma concentration after correcting as mode input function, the tissue concentration after correction is as model output function by blood plasma to the spilling coefficient of cardiac muscular tissue and cardiac muscular tissue to the spilling coefficient of blood plasma respectively;

Model input-output function after step 6, the blood obtained according to step 3 and correction that when organizing, m-activity curve and step 5 obtain, adopt sum of square of deviations minimum principle, actual measurement data and pattern function are carried out to the curve fitting of multiparameter simultaneously, obtain the metabolic coefficients k of Myocardial Fatty Acids four Room kinetic model ₁-k ₅, and calculate cardiac muscular tissue to the overall utilization (MFAU) of fatty acid and fatty acid esterification (MFAE), oxidation (MFAO) parameter, complete and the quantification of fatty acid metabolism is detected.