System and method for estimating a quantity of interest of a dynamic artery/tissue/vein system

Abstract

The invention relates to a system and method for estimating a quantity of interest from perfusion data resulting from the acquisition of a plurality of a patient volumes or stations. Such a method may include a step for triggering the output of said quantity of interest in the form of a consolidated map by means of an adapted man-machine interface.

Claims

1. A method for producing an estimate of a quantity of interest of an artery/tissue/vein dynamic system of an elementary volumecalled a voxelof an organ, said method being performed by a processor of a perfusion imaging analysis system, and comprising: obtaining perfusion data linked to a first station, the first station corresponding to a volume of the organ defined by an acquisition field of a magnetic resonance imaging device, moving the organ, relative to the magnetic resonance imaging device, to a location associated with a second station at which the acquisition field of a magnetic resonance imaging device defines a different respective volume of the organ, obtaining perfusion data linked to the second station, determining an arterial input function that is specific to at least one station among the plurality of said stations, building a joint arterial input function from said specific arterial input function, and estimating the quantity of interest by applying the joint arterial input function built from said specific arterial input function to data obtained from each of the plurality of stations.

2. The method according to claim 1, wherein the perfusion imaging analysis system comprises output means for a user of said system, said output means cooperating with the processor, said method comprising a subsequent step to cause an output to the joint arterial input function by said output means.

3. The method according to claim 1, comprising: a preliminary step for preprocessing the perfusion data, said step being arranged to correct said perfusion data.

4. A method for producing an estimate of a quantity of interest of an artery/tissue/vein dynamic system of a region of interest, said region comprising at least one voxel, said quantity of interest being estimated by voxel using the method according to claim 1.

5. The method according to claim 4, in which said region of interest extends over several stations among the plurality of stations.

6. The method according to claim 4, the perfusion imaging analysis system having output means for a user of said system, said output means cooperating with the processor, said method comprising a subsequent step to trigger an output of said estimated quantity of interest for the voxels of the region of interest by said output means.

7. A processing unit having communication means with the outside world and processing means cooperating with storage means, wherein: the communication means are adapted to receive from the outside world perfusion data linked to a plurality of stations, each station being a volume corresponding to an acquisition field defined by a magnetic resonance imaging device; and the storage means include instructions executable or interpretable by the processing means whose interpretation or execution of said instructions causes the implementation of a method according to claim 4.

8. A processing unit according to claim 7, in which the communication means deliver an estimated quantity of interest in a format suitable for output means adapted to output it to a user.

9. A perfusion imaging analysis system comprising a processing unit, according to claim 7, and output means configured to output to a user an estimated quantity.

10. A non-transitory computer readable medium encoded with a program comprising one or more instructions interpretable or executable by the processing means of a processing unit, said processing unit comprising storage means or cooperating with storage means, said program being loadable into said storage means, wherein the interpretation or execution of said instructions by said processing means triggers the implementation of a method according to claim 4.

11. The method according to claim 1, wherein said joint arterial input function is built from perfusion data linked to at least the first and second stations.

12. The method according to claim 1, wherein the joint arterial input function is built from arterial input functions that are respectively specific to each of at least two stations among said plurality of stations.

Description

(1) Other features and advantages will become clearer after reading the following description and examining the accompanying figures in which:

(2) FIGS. 1 and 2, previously described, show two alternative embodiments of a medical imaging analysis system;

(3) FIG. 3, previously described, shows images of perfusion, obtained by an imaging device by Nuclear Magnetic Resonance in connection with several imaged stations of a human being after injection of a contrast agent;

(4) FIG. 4, previously described, shows examples of arterial input functions determined for four stations, in the form of concentration curves C(t) of a contrast agent flowing within a voxel of an artery as a function of time;

(5) FIGS. 5a and 5b show simplified flow diagrams of methods in accordance with the invention;

(6) FIG. 6 shows an example of an arterial input function built from the perfusion data of four stations, in accordance with the invention;

(7) FIG. 7 shows five maps relative to the estimated blood flow from a plurality of stations according to the invention;

(8) FIG. 8 shows a consolidated map relating to said estimated blood flow.

(9) FIG. 5a shows an example of a simplified algorithm of method 100according to the inventionfor producing an estimate of a quantity of interest of an artery/tissue/vein dynamic system of an organ elementary volumecalled a voxel. Such a method 100 may be performed by a processing unit of a perfusion imaging analysis system, such as the system described in connection with FIG. 1 or 2 and adapted accordingly.

(10) The method 100 according to the invention mainly comprises a step 130 for estimating a quantity of interest, by way of non-limiting examples, a hemodynamic parameter or the residue function, from perfusion data linked to a plurality of stations. Within the meaning of the invention, the term station designates a volume corresponding to the acquisition field of a medical imaging device, such as the device 1 of the perfusion imaging analysis system described in connection with FIG. 1 or 2. Thus, each station is linked to or associated with a volume corresponding to an acquisition field defined by a medical imaging device. Said acquisition field may or not be, depending on the variants, associated with a particular position of the movable table of the medical imaging device on which the patient is generally lying. Within the meaning of the invention, the term perfusion data designates all the images or signals acquired at several times by a medical imaging device in order to study the evolution of a previously injected contrast agent in the body of a patient. Such data may advantageously consist of image sequences, such as the image sequences 12 described in connection with FIG. 1 or 2. As mentioned above in connection with FIGS. 1 and 2, the perfusion data may advantageously be stored in a server at the time of their acquisition so that they can be processed later by a user of an imaging analysis system, preferably a practitioner. Perfusion data are advantageously linked to or associated with a particular station.

(11) According to the invention, prior to the estimation of hemodynamic parameters 130, in order to overcome the temporal sampling of the arterial input function, such a method 100 comprises a step 120 for building a joint arterial input function from perfusion data linked to at least one station among said plurality of stations. However, building such arterial input function in accordance with step 120 may not be limited to the perfusion data in connection with a single station: in fact, in order, for example, for fitting a model function of arterial input closer to reality and refine the construction of said joint arterial input function, perfusion data can be linked or associated with two or more stations, namely as much as stations available. By way of non-limiting example, as described in connection with FIG. 3, the perfusion data can come from the five stations I to V to construct a joint arterial input function.

(12) FIG. 6 shows an example of an arterial input function built from four stations I to IV of the five acquired, said stations being defined by FIG. 3 and acquired according to the experimental protocol of acquisition of Prof. Rahmouni and Luciani.

(13) Generally, the built joint arterial input function is advantageously unique for all stations: such arterial input function can be described by an analytic function given as C.sub.a(t,.sub.a) where .sub.a is the set of the parameters of the arterial input function. Such parameters .sub.a are estimated in 120 from the perfusion data of different stations and are the same for all voxels of all stations. Such an estimate of parameters .sub.a can be achieved, by way of non-limiting example, by adjusting the model defined C.sub.a(t,.sub.a) on concentration curves of the contrast agent from arteries selected from all stations. Said arteries can be selected manually or automatically according to the chosen protocol. Alternatively or additionally, parameters .sub.a may also be adjusted in conjunction with hemodynamic parameters .sub.hem of the voxels. According to a preferred embodiment, the invention provides that step 120 for building a joint arterial input function can include a step for determining, manually or automatically, an arterial input function, specific or dedicated to a station among said at least one station, the joint arterial input function being built from said specific arterial input function. As stated previously, in order to adjust a model of arterial input function closest to reality and refine the building of said function by estimating the parameters .sub.a, perfusion data may be linked to or associated with two or more stations, i.e. up to as many stations as available.

(14) A preferred such embodiment is described in connection with FIG. 6. In this example, step 120 for building a joint arterial input function C.sub.a(t) consists in adjusting said function to the measured values of contrast agent concentrations at sampling times within each station, among the stations I to IV. As specified above, said joint arterial input function can be described by an analytic function given as C.sub.a(t,.sub.a) where .sub.a is the set of parameters of the arterial input function. By way of non-limiting but preferred example, such an analytical function may correspond to a bi-exponential function such as:

(15) $C_a\left(t,{}_a\right)\{\begin{array}{cc}0& \mathrm{if}t<\\ A_1{e}^{-k_1\left(t-\right)}+A_2{e}^{-k_2\left(t-\right)}& \mathrm{if}t>\end{array}$
where .sub.a={A.sub.1, A.sub.2, k.sub.1, k.sub.2, }. Such a model is particularly suitable for the implementation of a method according to the invention, since it correctly reproduces the temporal characteristics of the arterial input function curve C.sub.a(t), that is to say, a sudden spike followed by a rapid decrease and finally a slower decrease.

(16) Alternatively or additionally, the invention also provides that step 120, for building a joint arterial input function, consists in selecting, in a database of existing arterial input functions, said functions being previously measured with a sufficient temporal resolution in a population of control patients, a joint arterial input function. The choice of such oversampled arterial input function may be carried out using the perfusion data of each station.

(17) Advantageously, when the perfusion imaging analysis system described in connection with FIGS. 1 and 2, includes output means 5, such as, by way of a non-limiting example, a human-machine interface, such as a screen or other equivalent means, to a user 6 of said system, said output means 5 cooperating with the processing unit 4, said method may comprise a subsequent step 121 to cause an output of the joint arterial input function thus built by said output means. Alternatively or additionally, such a step 121 may also cause an output of the perfusion data voxels used to build the joint arterial input function. Such a step 121 allows, in particular when the voxels have been automatically selected to build the joint arterial input function, to validate the relevance of said selected voxels, that is, that the selected voxels actually correspond to voxels of interest in the arteries.

(18) In addition, step 130 for estimating the quantity of interest, uses the built joint arterial input function: such arterial input function is valid for all stations at all times. Step 130 for estimating the quantity of interest may consist of the implementation, by the processing unit, of all known estimation techniques of hemodynamic parameters, advantageously parametric or non-parametric methods, as described above.

(19) According to the parametric approach, for a given microcirculation model, the function of residues may be described by an analytic function given as R(t,.sub.hem) where .sub.hem is the set of hemodynamic parameters. By way of non-limiting examples, a method for optimizing the parameters of the function, such that the least squares minimization, with or without constraint, or even a Bayesian estimation method, may advantageously be performed.

(20) Alternatively or additionally, according to the non-parametric approach, a convolution matrix may be built from the built joint arterial input function using a finer temporal grid than that of the acquisition protocol, so as to properly sample the different time scales present in said arterial input function. By way of non-limiting examples, the SVD (Singular Value Decomposition) method may be performed. However, the Bayesian deconvolution method is preferred in view of the particularly proven accuracy of the estimates produced.

(21) In addition, when the perfusion imaging analysis system described in connection with FIGS. 1 and 2 includes output means 5, such as, by way of a non-limiting example, a human-machine interface or any other equivalent means, for a user 6 of said system, said output means cooperating with the processing unit 4, said method may comprise a subsequent step 131 to cause an output of the estimated quantity of interest, such as, for example, hemodynamics parameters 14, in an appropriate format. By way of non-limiting examples, such a format may be a value, a color within a specific palette to express the intensity of said estimated quantity of interest, or any other equivalent means.

(22) To increase the relevance of perfusion data, the invention provides that the method may include a preliminary step 110 for pretreatment of perfusion data, said step consisting in particular in correcting artifacts or applying any other corrective filter.

(23) Computed Tomography imaging and, especially, Magnetic Resonance Imaging, like all other medical imaging techniques, is no exception to producing false images: artifacts. Artifacts are observable images that do not actually represent any anatomical reality. Quite often, it is necessary to attempt to avoid or minimize them by modifying certain acquisition or reconstruction parameters. Such artifacts may be of different kinds. In principle, by way of non-limiting examples, three corrections are generally applied to improve the quality of the perfusion data: a correction of patient movements or movements due to breathing, heartbeat and blood flow, a correction of the field of vision of the perfusion imaging device and/or an image denoising.

(24) FIG. 5b shows an example of a simplified algorithm of method 200according to the inventionto produce an estimate of a quantity of interest of an artery/tissue/vein dynamic system of a region of interest.

(25) A method 200 is arranged to produce an estimate of a quantity of interest, by way of non-limiting examples, a hemodynamic parameter or the residue function, of an artery/tissue/vein dynamic system of a region of interest. A region of interest means any region with at least one voxel. Nevertheless, a region of interest should not be restricted to a single voxel, but may include a plurality of voxels, selected manually or automatically. According to the invention, said quantity of interest may be estimated for each voxel by means of a method 100 according to the invention, as described previously, performed iteratively for each voxel by the processing means of the processing unit 4. It is thus possible, according to step 210, to estimate a quantity of interest on a plurality of voxels defining a region of interest, which may optionally extend over several stations among a plurality of stations, in order to, for example, perform the analysis of organs that are larger than the acquisition field of the perfusion imaging analysis device.

(26) In addition, said method may comprise a subsequent step 211 for causing an output of said quantity of interest, namely a hemodynamic parameter 14, estimated for the voxels of the region of interest by said output means according to an appropriate format. The output may advantageously be related to a parameter map where each voxel corresponds to a degree of intensity in relation to the estimated quantity of interest. Such step may include a substep for displaying a parameter map for each station. Such an embodiment is described in connection with FIG. 7. FIG. 7 shows five maps relating to blood flow, also known as K.sub.trans, estimated from a plurality of stations, preferably five, in accordance with the invention, the five maps corresponding to stations I to V. Such maps for the stations I to V may be advantageously used in the context of the study of myeloma, cancer affecting the cells of the spinal cord. Each station thus comprises a representation independent from one another.

(27) Alternatively or additionally, the method 200 may include a step 220 for generating a global volume from a plurality of stations. It may also include a step for causing the output, for example in the form of a consolidated map integrating or merging the maps produced for said plurality of stations. Such a step 220 may thus comprise a sub-step for joining different station maps included in the plurality of stations. Such an embodiment is described in connection with FIG. 8.

(28) FIG. 8 shows a consolidated map of a plurality of stations from parameter maps relating to estimated blood flow, said maps shown by way of non-limiting examples in connection with FIG. 7. Two adjacent maps among the different stations I to V, generally comprise an overlap region, said overlap region preferably having corresponding peripheral voxels, corresponding in pairs and forming pairs of corresponding voxels. The quantity of interest in connection with such pairs of voxels may be estimated by the processing unit 4 as the result of a linear combination between the estimates of said quantity of interest of the corresponding voxels within the two adjacent stations. Thus, any discontinuity in the consolidated map, resulting from a fusion of parameter maps specific to each station, among the stations I to V, is prevented or attenuated: observation of the presence of such discontinuity may be verified through the output means 5 of a medical imaging device, such as that described in conjunction with FIG. 1 or 2. According to the example described in connection with FIG. 8, the maps of the stations I and II, as well as III and IV, are merged to generate a consolidated map.

(29) Alternatively or additionally, the method 200 may include a step 230 for verifying the estimate of the estimated quantity of interest: such verification may be performed automatically or visually. Said verification step 230 may consist in detecting any significant discontinuity on a consolidated map. Alternatively or additionally, such verification step 230 may consist in automatically verifying that the values of estimates of said quantity of interest of the corresponding voxels are coherent, that is to say, that the values of such estimates are in a small range of values. Such verification can be performed using statistical tests, such as, by way of non-limiting examples, the Kolmogorov-Smirnov test or, preferably, the Bayesian theory. The method 200 may then include a step 231 to cause, via the output means, highlighting of the voxels, considered as little coherent in 230.

(30) Thanks to the new estimates and/or maps presented above, the invention allows providing a practitioner with a set of relevant and consistent information, which could not be available using known techniques of the state of the art. This availability is made possible by adapting the processing unit 4 according to FIGS. 1 and 2 in that the processing means perform a method of estimating a quantity of interest involving the building of an arterial input function from perfusion data linked to one or more stations. Such implementation is advantageously made possible by uploading or storing within storage means cooperating with said processing means, a computer program product. The latter in fact comprises interpretable and/or executable instructions by said processing means. The interpretation or the execution of said instructions triggers the implementation of a method 100 or 200 in accordance with the invention. The means to communicate with the outside world of the processing unit can deliver a quantity of interest, namely, the estimated parameters 14, in a format adapted to the output means capable of outputting to a user 6, while said estimated quantity of interest may be advantageously output in the form of maps, for example, as illustrated in FIGS. 7 and 8.

(31) Thanks to the invention, the data delivered are thus more numerous, consistent, reproducible and accurate. All the information available to a practitioner is thus likely to increase his confidence in determining a diagnosis and making decisions.