Error adaptive functional imaging

FIELD: information technology.

SUBSTANCE: method for use in functional medical imaging involves adaptively partitioning functional imaging data as a function of a spatially varying error model. The functional image data are partitioned according to an optimisation strategy. The data may be visualised or used to plan a course of treatment. In one version, the image data are partitioned so as to vary their spatial resolution. In another version, the number of clusters is varied based on the error model.

EFFECT: high efficiency of functional medical imaging owing to accounting for noise effects and other uncertainties in functional imaging.

40 cl, 6 dwg

The present invention relates primarily to a functional medical imaging. It finds particular application in situations where it is desirable to identify and take into account the uncertainty arising from noise and other sources of errors in information functional imaging.

Imaging technologies have become essential tools in the diagnosis and treatment of diseases. One aspect of medical imaging, which attracted considerable attention, became functional imaging, which provides information on functional parameters, important for the diagnosis and/or treatment.

Functional imaging, in General, entails the formation of a parametric maps of interest parameter. One example of this is the use of images FMISO-PET in cancer research, where the functional associated with hypoxia parameter is used to distinguish between more and less aggressive part of the tumor. Another example includes studies of cardiac activity based on PET or SPECT, where parametric maps in functional associated with the perfusion parameters are used to identify hypoperfusion region of the myocardium. Other examples are functional magnetic resonance imaging (fMRI), to ora particularly useful for providing information, important for brain function, and molecular imaging, which provides information about the molecular marker or agent.

However, one complicating factor was the effect of noise and other uncertainties. Although based on the voxels of the parameter estimates provide a relatively high spatial resolution, these estimates are particularly susceptible to the effects of noise in the base image data. If they are not taken into account, the effects of such noise can lead to suboptimal diagnosis or treatment.

Used techniques of smoothing in order to reduce these fluctuations of parameters, arising from such factors as noise visualization, noise modeling and data sparseness. One popular method of smoothing is clustering, where the area of interest or the volume is divided into contiguous and non-contiguous sub-regions having similar parameter values. Set the desired number of clusters, and the clustering algorithm is used to divide the area by the specified number of clusters. The so-called area, in turn, are used for diagnosis (for example, determine the size of the tumor or ischemic region in the myocardium) and/or therapy (for example, determination of radiation doses from radiotherapy treatment planning).

However, one of the drawbacks is the above method is the number and/or distribution of clusters may be inadequate, leading to sub-optimal clustering. Suboptimal clustering, in turn, may lead to suboptimal diagnosis and/or treatment.

Accordingly, it is desirable to provide improved methods of accounting for the effects of noise and other uncertainties in functional imaging.

Aspects of the present invention solve these and other problems.

According to the first aspect of the present invention, the method includes receiving data functional image indicating the area of interest of the object, and data of the functional images include the value of the spatially varying functional parameter and model spatially varying functional errors. The method also includes partitioning the region of interest depending on the model spatially varying errors and generate image indicating a partitioned area of interest.

According to the first aspect of the present invention, the device includes a means of receiving data functional image indicating the area of interest of the object, and data of the functional images include the value of the spatially varying functional parameter of the model spatially varying functional errors. The method also includes a means of partitioning the region of interest depending on the values of the spatial variable errors and the means of formation of the image, indicating a partitioned area of interest.

According to another aspect of the invention a machine-readable storage medium contains instructions that, when enforced by a computer, instruct the computer to perform a method that includes receiving data functional image indicating the area of interest of the object, and data clustering functional images into multiple clusters. Data functional images include functional values and the model of functional errors. The number of clusters is a function of the model of functional errors in the function values included in the cluster.

According to another aspect of the invention a machine-readable storage medium contains instructions that, when enforced by a computer, instruct the computer to perform a method that includes receiving data functional image indicating the area of interest of the object, using models of spatially varying functional errors in order to spatially vary the spatial is the information and communications technology data functional images, and generate image data indicating the functional images.

According to another aspect of the invention, the method includes receiving data functional image indicating the area of interest, the output of the functional images to a human-perceivable form in the GUI and interactive partitioning the region of interest based on the model of functional errors and evaluation of human data to display functional images.

Experts in the art should recognize still other aspects of the present invention with the understanding the accompanying drawings and description.

The present invention is illustrated as an example, and not limitation, in the figures of the attached drawings, in which similar references indicate similar elements, of which:

Figure 1 illustrates the steps of partitioning the image volume;

Figa, 2B, 2C and 2D illustrate the amount of images at different stages of the two-element partitioning;

Figure 3 illustrates the different clusters.

Parametric maps and functional information extracted from a practical imaging procedures such as PET, SPECT, or fMRI, can have a significant error or uncertainty due to noise in the base image data. The noise value of the functional parameter in General, associated with the size of the area in which it is evaluated. Based on the voxels estimates show higher levels of spatial resolution, but, in General, have greater uncertainty or confidence intervals in the displayed values. Although reducing the spatial resolution seeks to reduce uncertainty and, consequently, narrow confidence intervals, spatial detail may be lost.

These effects can be improved by adaptively varying the spatial resolution estimates of parameters in the volume image, or other areas of interest depending on the error. Spatial resolution can vary according to the required criteria optimization errors, for example, to achieve the desired error or desired distribution of the errors in the region of interest. Doing this in many cases can improve the visualization of data or provide other information that is useful for diagnosis and treatment.

On clustering techniques can also be affected by errors or uncertainties underlying functional data, especially when the number of clusters is set a priori. For example, prior specification of the number of clusters can lead to the clustering results, for which the average value of the functional parameters is as in two or more clusters may lead to statistically insignificant deviations. Therefore, the separation between two or more of the clusters may not be physiologically and/or statistically significant. On the other hand, the characteristics of the data may be such that there might be additional significant clusters.

These effects can be improved by the inclusion of the issue as part of the clustering procedure, and use this index in order to adaptively adjust the number of clusters as a function of error. For example, the error may be used to evaluate the significance of the clustering operation and to adjust the number of clusters accordingly. Doing this in many cases can improve the separation of the clusters, thereby improving visualization of the data or providing other information that is useful for diagnosis and treatment.

Approximate iterative procedure partition volume or other areas of interest depending on the error shown in figure 1.

Evidence from functional imaging, such as parametric map, taken on the stage 102. Evidence from functional imaging typically include the value of f(x, y, z) spatial varying functional parameter and associative associated model e(x, y, z) spatial varying errors, which vary according to the volume image is available.

When the area of interest is a subset of the image, an optional step of selecting the region of interest is performed at step 104. The area of interest can be selected by the operator, for example, by displaying the volume perceived by the human form and requesting that the user has specified the desired area. The area of interest can also be determined by segmenting or other methods of processing images either alone or in connection with the operator input.

The initial partitioning P₀the region of interest is selected at step 106. More specifically, the area of interest is divided into one or more subareas. In one implementation of the sub-regions are spatial, for example, by a two-element partition of the data set. In a different implementation of subfields are set using clustering techniques or building areas.

At step 108, the current partition P_nmodified according to based on the error partitioning strategies to form a new partition P_n+1.

The new section is evaluated according to based on the error criterion of approval on the stage 110. In one implementation, the partitioning is performed so that the estimation errors of the different subareas minimized or otherwise were men who e the desired threshold value. In another embodiment, the partitioning is performed so that different sub-regions had the desired uniformity errors, for example, so that the difference between the error values in different subareas minimized or otherwise was less than the required threshold (or to put it another way, so that the homogeneity of error maximinerals).

The optimization strategy and the criteria for approval may also consider a functional parameters when performing partition accordingly. As one example, the criteria for approval may also be considered as a variation of the values of the voxels, and the error estimates of functional parameters in a specific sub-region or multiple regions.

At step 112, if the approval criteria are not satisfied, the processing returns to step 108, where the volume of the region of interest re-partitioning. If the criterion for approval is satisfied, the partitioning is approved.

The proposed partitioning may not necessarily be presented to the operator for approval. Depending on the implementation, the operator may reject the proposed partitioning, and in this case the previous partitioning approved. Alternatively, the operator may choose to approve a new partition, even if the criteria for approval are not satisfied. The operator also can the t to be able to manually modify the partitioning.

The proposed partitioning can be presented to the operator in various ways. For example, the subregion, which should be the steps that can be highlighted by color-coding, by flashing, alternating visualizations or other graphic techniques using graphical user interface (GUI). Alternative or additionally, information such as histograms, digital display of parameter values and/or errors and text messages can also be presented.

Approved partitioning is rendered at step 114, for example, by displaying parametric maps in a human-readable form on a computer display or monitor. Alternative or additionally, the data can be used as input to the automated diagnostic program or procedure planning therapy to help in the diagnosis or treatment planning.

The following describes an exemplary partitioning, in which the spatial resolution of the region of interest varies as a function of error. P is a complete partitioning of the volume or region of interest, i.e. the set of sets S on p of voxels:

Equation 1

P = {S1, ..., S_P},

Equation 2

Si = {(x1, y1, z1), ..., (x_Ni, y_Ni, z_Ni)}, i = 1, .., P.

Each set of voxels Si specifies the area consisting of Ni of voxels for which the functional parameterf(Si) and its errore(Si) can be calculated:

Equation 3

{f(S),e(S)} = F(S)

where F represents the procedure of parameter estimation using the estimated value of the functional parameterfand related associative models of functional errorsegenerated by the model functional imaging.

Error in parameter values can be reduced by averaging over a larger area. The smallest possible area is one voxel, whereas the largest possible area is the entire volume. The volume is partitioned so that the error for the parameter values satisfy the required criteria for approval.

As described above, allowed different optimization strategy and criteria for approval. For example, the volume may be partitioned according to the strategy of minimizing the functional of the error so that a functional error in each sub-region was smaller than the maximum error threshold:

Equation 4

e(S_i) ≤ e_max, i = 1, ..., P.

If the error is small enough, the spatial resolution is not reduced further. Another approach is the minimum error:

Equation 5

e(S_i) → min, i = 1, ..., P.

Protrans the governmental resolution is always consistent with the minimization of errors.

Another approach focuses on the homogeneity of errors among subareas, so that the variation of the functional errors among two or more subareas minimized. For example, the volume may be partitioned so that the variation of error was less than the threshold value:

Equation 6

,

where e_thresh- this is the maximum variation. The variation of the functional errors can also be minimized:

Equation 7

.

The latter technique will agree spatial resolution uniformity errors.

Note that the above explanation has focused on the optimization of the entire volume. Optimization may also be limited by region or other information enclosed volume. Various criteria for approval can also be set for different parts of the volume or area. Thresholds can also be set as a function of values of the functional parameters of thef.

The following describes an exemplary strategy two-element partitioning with reference to figure 2. The initial volume 202 shown in figa. Volume 202 includes a set of voxels 204, each of which is set functional parameters of thefand the model of functional errorse. Figv and 2C illustrate the two-element Razlog is the second and third volume level on the sub region for areas with relatively more approximate spatial resolution, respectively. When the objective is to vary the resolution as a function of the error distribution, the values of the functional parameters are preferably calculated at each level of resolution. Although the third level decomposition is shown for simplicity of illustration, may also be implemented more (for example, four or more) or fewer levels.

Select the most approximate resolution (i.e. the resolution figs in this example). This section can be tested on the subject of the approval criteria to determine whether he is already satisfied. Section modified by decomposition of each sub-region that is not already at the highest resolution level (i.e. the level figa in this example)to a higher resolution. The proposed modification is tested to determine whether it is moving in the direction of the criterion for approval. After the approval criteria is satisfied for all nested scopes (or if additional sections not suggested), the current section is approved, and the process is terminated.

The approximate final decomposition shown in fig.2D. As illustrated, the lower portion 206 of the volume has a relatively low resolution, the Central portion 208 of the volume has a relatively high resolution, and the remainder of the upper part 10 has an intermediate resolution. As you can see, the volume demonstrates the spatial range spatial resolution.

As described above, the partitioning can be performed interactively through the GUI or otherwise, so that the user had the ability to approve, reject, or otherwise modify the proposed section. It may also be useful to keep a history (or a partial history of the process of partitioning. The story can then act as rapid changes section in interactive scenarios. History can also be used to provide partitioning based on previous decisions, for example, for use in adaptive treatment radiation therapy (i.e., recurrent visualization). Although the above example explains the strategy of two-element partitioning can be used strategy of clustering, capacity areas or nevokletocny strategy.

Approximate partitioning strategy, which uses clustering to vary the number of clusters as a function of the error modeleand statistical models ofs, further described in connection with figure 3. Below the error functional of the parameterxin the voxelireferred to ase_i. The functional value of the parameterxin the voxelithus, equal to(x_{i ±ei).}

Partitioning can be performed using K-Means or other known clustering algorithms. The K-Means algorithm can be represented as follows:

To begin initializing:nNC,μ₁, μ₂, ..., μ_c

Runclassificationnsamples according to nearestμ_i

recalculateμ_i

Untilthere are no changes inμ_i

Returnμ₁, μ₂, ..., μ_c

End

wherenis the sample size, NC is a pre - defined number of clusters, andμ_care the values of the cluster centers. Other clustering methods, such as the well-known algorithm K-Harmonic Means (which is relatively less sensitive to initial conditions) or a known Genetic algorithm Clustering (which sets the number of clusters NC based on the values of the parametersf), can also be used.

Classification samplex_iaccording nearest the center of the clusterc_jis performed using an appropriate measure of distancedwherelis vector data components:

Equation 8

,

whereρis the Exhibitor weighting in order to adjust the distance value, for example,ρ = 1: city block (L1),ρ= 2 : Euclid (L2). Weightw_kin addition, you can specify the index and may reflect the duration of a frame in case of dynamic data.

As shown in figure 3, the approximate clustering the region of interest 302 may provide the result of the first 304 and second 306 and 308 third clusters. The value of the functional parameters (calculated as average, average, another function of the voxels in the cluster) of the cluster centers can be represented as 304_c, 306_cand 308_cand the error functional values - ase304_c, e306_cande308_c. If, for example, the value of the parameters of the second cluster 306 falls within the limits of error of the third cluster 308:

Equation 9

308_c-e308_c< 306_c< 308_c+e308_c,

then, as a rule, it is undesirable to distinguish 306 and second 308 third clusters for the purposes of diagnosis and treatment, and fewer clusters should be provided (i.e., the clusters should be merged). On the other hand, if the statistical and physiological characteristics of the data such that additional allocation of one or more clusters should provide additional information which is relevant for diagnosis and treatment, should be allocated a larger number of clusters.

Statistical errorsvalues parametrov> xin the clusterkcan be expressed as the standard deviation of the values of the voxelsxbelonging to the clusterk:

Equation 10

,

where s_kis the standard deviation, c_kis the center of the cluster, and N_kis the number of voxels or elements in the clusterk. Note that the distribution ofxcan also be expressed as the mean, median or other function.

In this example, the model of functional errorseand the statistical errorsboth are used to access the separation of the clusters and to adjust the number of clusters, and the strategy of combining/splitting is used to vary the number of clusters. The pseudocode sample strategy combining/splitting of the upper level can be expressed as follows:

FOR j = 1 TO NC

FOR i = 1 TO NC

IF i != j And functional value (μ_i-e_i) <μ_i< (μ_i+e_i), THEN merge the cluster i and j

IFs_j> α • e_j,THEN divide the clusterj

UNTIL there are no more changes in the allocation of clusters

whereαis an empirically determined constant. As mentioned above, the operations of merging and splitting can be performed automatically or confirmed group is a rotary user interaction.

Further more describes the approximate merging clusters. If, for example, the functional value ofμ_jclusterjfall within the error bars (μ_i-e_i) clusterithe algorithm proposes merging clustersiandj.One implementation is as follows:

To begin initializing:nNC,μ₁, μ₂, ..., μ_c

Runk-means or another clustering, resulting in the centers of clustersμ'₁, μ'₂, ..., μ'_c0

If for any two clustersiandj μ'_i-e'_i<μ'_j<μ'_i+e'_i,then offer to merge clustersiandj

If the user approves, then ask NC-> NC-1and calculate the average clusteriandj

Re-clustering

Untilthe centers of the clusters will not be explicit in the model errorse

Returnthe estimated cluster centers

The end.

Alternative criteria, including statistical distribution ofsalso assumed.

In the example in figure 3, the clustering procedure is performed with three clusters 304, 306, 308. If the functional values in the first 304 and second 306 cannot be distinguished in relation to models of functional errors e, two clusters are merged.

Further more describes the approximate separation of the clusters. If the statistical error s_kcluster k more functional errors e_k, the algorithm has to offer to split the cluster k:

To begin initializing:n, NC, μ₁, μ₂that ...,μ_NC

Runk-means or another clustering, resulting in the centers of clustersμ₁', μ'₂, ..., μ'_c0

If for any clusterk:e_k<c * s_kandN_k> Θ, then offer to split clustersk

If the user approves, then set NC ->NC+1 andμ_k= μ'_k- ε,μ_k+1= μ'_k+ ε

To re-run the clustering algorithm k-means

Untilno additional separation

Returnthe estimated cluster centers

The end.

Above, c and Θ is empirically determined constants, and ε is a vector of small perturbations. Note that the merging and splitting of clusters can be combined if necessary.

In the example in figure 3, if the statistical error of the first cluster 304 is large in comparison with the model of functional errorsethe number of clusters increases, the clustering procedure is upuskaetsya again at the first cluster 304, divided into two clusters.

As described above, the merging and splitting of clusters can be presented to the operator for approval. Thus, clusters can be distinguished by color-coding, alternating visualization clusters, flashing, text messages etc. Statistics, such as histograms, functional values ofμ_iand/or functional errorse_ifor one or more clusters, can also be represented. The user can then decide to approve or reject the proposed new section, to perform the clustering algorithm with the adjusted number of clusters or approve the result and complete the process.

The techniques described above are ideally suited for data generated by PET, SPECT, fMRI, functional CT or other scanners that can provide functional information. Methods are also optimally suited for use with data generated by molecular imaging, which provides information on other functional characteristics, such as the consumption of glucose (for example, FDG PET), growth of cells (for example, FLT PET), apoptosis (for example, Annexin-V) and the density of receptors (in the brain or anywhere else).

In radiation therapy and other types of applications where anatomies what I localization is important, the patient may also be scanned using MRI, computed tomography (CT), ultrasound (US), x-rays or other scanner, and functional and anatomical data jointly registered. In some cases (for example, in the case of MR and fMRI) functional and anatomical information can be obtained in a single scan procedure. It also assumes the use of hybrid scanners, such as hybrid PET/MR, PET/CT, SPECT/CT or other hybrid modality.

Data from functional imaging modalities are modeled using appropriate physical model to generate one or more parametric maps, and data is processed as described above. The resulting data can be used by the physician or user in connection with the diagnosis or treatment planning. The resulting data can also be used as input into the treatment planning system. In an exemplary case pack radiotherapy treatment planning (RTP) data are used to plan radiation dose, for example, by providing a relatively higher exposure doses for relatively more resistant to radiation parts of the tumor.

Embodiments of the invention described above may be materially implemented in the computer about the Ramm, stored in a machine-readable recording media. The computer program contains instructions that, when read and enforced by a computer, instruct the computer to perform the steps necessary to execute steps or elements of the present invention. Exemplary machine-readable recording media include, but not limited to, fixed hard disks, optical disks, magnetic tape, semiconductor memory devices, such as permanent memory (ROM), random access memory (RAM) and programmable (PROM). A storage device containing computer-readable code that is used by the execution of the code directly from the storage device or by copying the code from one storage device to another storage device, or by transmitting the code over a network for remote execution.

It is obvious that modifications and changes should come to mind after reading and understanding the above description. The invention should be construed as including all such alterations and modifications insofar as they fall within the scope of the attached claims of the invention or its equivalents.

1. The method of functional health provide the purpose, containing phases in which:
accept these functional images indicating the area of interest of the object, and data of the functional images include the value (f) spatially varying functional parameter and model (e) spatially varying functional errors;
partition the region of interest depending on the model spatially varying errors;
form the image, indicating a partitioned area of interest.

2. The method according to claim 1, wherein partitioning includes a stage on which partition the region of interest into subdomains with different spatial resolution(206, 208, 210).

3. The method according to claim 2, in which the partitioning includes a stage on which partition the region of interest according to the criterion of homogeneity model of functional errors.

4. The method according to claim 2, in which the partitioning includes a stage on which partition the region of interest according to the criterion of minimization of the functional model errors.

5. The method according to claim 1, wherein partitioning includes a stage on which partition the region of interest into multiple clusters (304, 306, 308), with similar values of functional parameters, with the number of clusters is a function of model errors and statistics the standard error.

6. The method according to claim 1, wherein partitioning includes the steps are:
partition the region of interest on the first number of clusters (304, 306, 308);
use the model errors in order to evaluate the separation of the clusters;
partition the region of interest on the second number of clusters.

7. The method according to claim 6, in which the use of the error model includes a stage on which to use the model error and the statistical error in order to evaluate the separation of the cluster.

8. The method according to claim 7, in which partitioning the region of interest on the second number of clusters includes a stage on which to perform at least one of the separate cluster and merge the first and second cluster.

9. The method according to claim 6, comprising the steps are:
show the proposed partitioning in a human-perceivable form;
provide the user an opportunity to reject the proposed partitioning.

10. The method according to claim 9, comprising a stage on which distinguish the proposed partitioning using the GUI.

11. The method according to claim 1, wherein the functional parameter specifies hypoxia.

12. The method according to claim 1, in which the method includes a stage on which use a partitioned area of interest in order to compute the distribution of the doses.

13. The method according to claim 1, in which fu is clonally parameter specifies perfusion.

14. The method according to claim 1, wherein the image indicates a molecular agent rendering.

15. The method according to claim 1, wherein the functional parameter specifies the consumption of glucose, growth, apoptosis or density of receptors.

16. Device for functional medical imaging, comprising:
means for receiving data of the functional images, indicating the area of interest of the object, and data of the functional images include the value (f) spatially varying functional parameter and model (e) spatially varying functional errors;
a means of partitioning the region of interest depending on the values of the spatial variable errors;
the means of formation of the image, indicating a partitioned area of interest.

17. The device according to clause 16, which includes means forming data of the functional images.

18. The device according to clause 16, in which the partitioning tool includes a GUI for interactive partitioning the region of interest depending on the model spatially varying errors and input man.

19. The computer-readable recording medium for functional medical imaging, containing instructions, which, when enforced by a computer, will structureat the computer to perform the method, containing phases in which:
accept these functional images indicating the area of interest of the object, and data of the functional images include functional values (f) and model errors;
clusterized data functional images into multiple clusters, with the number of clusters depends on the model of functional errors in the functional values included in the cluster.

20. The computer-readable recording medium according to claim 19, in which the number of clusters depends on the statistical error (s).

21. The computer-readable recording medium according to claim 19, in which the number of clusters depends on the model of functional errors (e).

22. The computer-readable recording medium according to claim 19, in which the method includes the steps are:
clustershot functional data on the first number of clusters;
use the model errors in order to evaluate the clusters;
adjust the number of clusters according to the evaluation result;
clustershot functional data on the second number of clusters.

23. The computer-readable recording medium according to item 22, in which the cluster includes the statistical error and the model of functional errors and in which the method includes the steps are:
compare the measure of statistical error and indicator models of functional errors;
if the measure of the distribution of parameter which exceeds the indicator of the physiological distribution of the errors, increase the number of clusters.

24. The computer-readable recording medium according to item 23, in which the method includes a stage on which to share the cluster.

25. The computer-readable recording medium according to item 23, in which the indicator of statistical error represents the standard deviation.

26. The computer-readable recording medium according to claim 19, in which the first cluster includes the value of the parameter, and the second cluster includes the error distribution, and in which the method includes the steps are:
determine whether the parameter value within the distribution of the errors;
if the parameter value is within the distribution of the errors, reduce the number of clusters.

27. The computer-readable recording medium on p, in which the method includes a stage on which unite the first and second clusters.

28. The computer-readable recording medium according to claim 19, in which the method includes the steps are:
show the proposed clustering data in a human-perceivable form;
provide an opportunity for the user to approve or reject the proposed clustering.

29. The computer-readable recording medium according to claim 19, in which the method includes a stage on which the use of clustered data in order to calculate the radiation dose.

30. The computer-readable recording medium according to claim 19, in which the MD area of interest is the brain, and clusters indicate the function of the brain.

31. The computer-readable recording medium for functional medical imaging, containing instructions, which, when enforced by a computer, instruct the computer to perform a method containing the steps are:
accept these functional images indicating the area of interest of the object, and data of the functional images include the value (f) spatially varying functional parameter and model (e) spatially varying functional errors;
use the model spatially varying functional errors in order to spatially vary the spatial resolution of the functional images;
form the image, data indicating the functional images.

32. The computer-readable recording medium on p, in which the method includes a stage on which to vary the spatial resolution according to one criterion of homogeneity of the functional model errors and the criterion of minimization of the functional model errors.

33. The computer-readable recording medium on p, in which the area of interest includes the myocardium and functional parameter indicates ischemia.

34. The computer-readable recording medium on p, in which the image in the cancel into itself, at least two sub-regions with different spatial resolutions based on the model of functional errors.

35. The method of functional medical imaging, comprising stages, which are:
accept these functional images indicating the area of interest of the object, and data of the functional images include the value (f) of the functional parameter and model (e) functional errors;
display data of the functional images to a human-perceivable form in the GUI;
interactively partition the region of interest based on the model of functional errors and evaluation of human data to display functional images.

36. The method according to p, including the stage at which repeat steps display and interactive partitioning the region of interest many times.

37. The method according to p, including the stage at which accept the input of the person approving the proposed partitioning interest amount.

38. The method according to p, including the stage at which accept the input of the person offering the partitioning of interest of the volume.

39. The method according to p in which interactive partitioning a volume of interest includes a stage on which to interactively partition interest amount into many subareas, while num is subareas based on the model of functional errors and the assessment of the person.

40. The method according to p, in which the functional parameter specifies hypoxia or perfusion.