Method for post-processing a sequence of acquisitions to correct the magnetic field inhomogeneities of a magnetic resonance imaging apparatus

Abstract

The invention relates to a method (100) for post-processing a biomarker (MBMK) produced from a set of M volumes Z(Δωj), each of the M volumes Z(Δωj) comprising N voxels which each describe the magnitude of an experimental signal sampled at a frequency shift Δωj relative to the resonance frequency of water during an acquisition sequence by means of a magnetic resonance imaging apparatus (1). The method is arranged to be implemented by a processing unit of an imaging analysis system. Such a method (100) consists of a correction step (130, 140) by decorrelating the biomarker (MBMK) with respect to the excitation magnetic field (B1) of the imaging apparatus (1). The method (100) relates to any magnetic resonance imaging application in which a correction of magnetic field inhomogeneities is required.

Description

[0001] Method for post-processing an acquisition sequence to correct magnetic field inhomogeneities in a magnetic resonance imaging device

[0002] The invention relates to a method for post-processing a sequence of acquisitions, possibly multiple, for which a correction of inhomogeneities of the static magnetic field BO and the excitation magnetic field B1 of a magnetic resonance imaging device, better known by the acronym IRM or MRI, for "Magnetic Resonance Imaging" according to English terminology, is required. Such a post-processing method naturally finds its application in particular in the context of CEST imaging (Chemical Exchange Saturation Transfer or Chemical Exchange Saturation Transfer, according to English terminology).The application of said post-processing method according to the invention, when applied to CEST imaging, makes it possible to exploit this technique in the clinical field and thus to quantify a biomarker linked to a chemical species of interest of a human or animal organ to ultimately characterize different lesions with altered metabolic properties such as tumors, ischemic tissues or multiple sclerosis. The most widely used CEST technique in clinical research is called amide proton transfer weighted MRI (APTw MRI), because its saturation pattern generates a sufficient signal-to-noise ratio to map the amide-related signal in a three Tesla clinical MRI scanner.

[0003] Magnetic resonance imaging is based on an analysis of the response of the proton of a water molecule when it is excited in a magnetic field. This response depends on the environment of such a proton and thus makes it possible to differentiate different types of tissue. A Nuclear Magnetic Resonance imaging device 1, as illustrated by way of non-limiting example by Figures 1 and 2, is generally used. This delivers a plurality of digital image sequences 12 of one or more parts of a patient's body, by way of non-limiting examples, the brain, the heart, the lungs. Said device applies for this purpose a combination of high-frequency electromagnetic waves to the part of the body considered and measures the signal re-emitted by certain atoms, such as by way of non-limiting example, hydrogen for Nuclear Magnetic Resonance imaging.The device thus makes it possible to determine the magnetic properties and, consequently, the chemical composition of biological tissues and therefore their nature, in each elementary volume, commonly called a voxel, of the imaged volume. The Nuclear Magnetic Resonance imaging device 1 is controlled using a console 2. A user 6, for example an operator, practitioner or researcher, can thus choose commands 11 to control the device 1, from parameters or instructions 16 entered via a human-machine input interface 8 of the analysis system AS. Such a human-machine interface 8 may consist, for example, of a computer keyboard, a pointing device, a touch screen, a microphone or, more generally, any interface arranged to translate a gesture or an instruction issued by a human 6 into control or parameter data.From information 10 produced by said apparatus 1, a plurality of sequences of digital images 12 of a part of a human or animal body are obtained. We will also call such information 10 or images 12 “experimental data”.

[0004] The image sequences 12 may optionally be stored within a server 3, i.e. a computer with its own storage means, and constitute a medical file 13 of a patient. Such a file 13 may comprise images of different types, such as functional images highlighting the activity of the tissues, or anatomical images reflecting the properties of the tissues. The image sequences 12 or, more generally, the experimental data are analyzed by a processing unit 4 arranged for this purpose. Such a processing unit 4 may for example consist of one or more microprocessors or microcontrollers implementing suitable application program instructions loaded into storage means, such as a non-volatile memory, of said imaging analysis system. Said processing unit 4 comprises means for communicating with the outside world to collect the images.Said means for communicating also allow the processing unit 4 to ultimately deliver a rendering, for example graphic and / or sound, of an estimation or quantification of a biomarker developed by said processing unit 4 from the experimental data 10 and / or 12 obtained by Magnetic Resonance Imaging, to a user 6 of the imaging analysis system AS via an output human-machine interface 5. Throughout the document, the term “output human-machine interface” means any device, used alone or in combination, making it possible to output or deliver a graphic, haptic, sound or, more generally, human-perceptible representation of a reconstructed physiological signal, in this case a biomarker, to a user 6 of a Magnetic Resonance Imaging analysis system AS.Such an output human-machine interface 5 may consist, in a non-exhaustive manner, of one or more screens, loudspeakers or other suitable alternative means. Said user 6 of the analysis system AS can thus confirm or deny a diagnosis, decide on a therapeutic action that he deems appropriate, deepen research work, refine adjustment parameters of measuring equipment, etc. Optionally, this user 6 can also configure the operation of the processing unit 4 or of the output human-machine interface 5, by means of operating and / or acquisition parameters 16. For example, he can thus define display thresholds or choose the biomarkers, indicators or estimated or quantified parameters for which he wishes to have a representation. The user uses for this the input human-machine interface 8 previously mentioned or a second input interface provided for this.Advantageously, the input 8 and output 5 human-machine interfaces may constitute a single physical entity. Said input 8 and output 5 human-machine interfaces of the imaging analysis system AS may also be integrated into the acquisition console 2. There is a variant, described in connection with FIG. 2, for which an imaging system AS, as described previously, further comprises a preprocessing unit 7 for analyzing the image sequences 12, deducing experimental signals 15 therefrom and delivering the latter to the processing unit 4 which is thus relieved of this task.

[0005] Among the techniques or modalities based on magnetic resonance imaging, we distinguish Chemical Exchange Saturation Transfer imaging or CEST. Such a technique consists of applying a radiofrequency pulse according to different determined resonance frequencies œ so that chemical species of interest reach a state of saturation. Said determined resonance frequencies œ are in fact respectively associated with different chemical species of interest, such as, but not limited to, amide (NH), amine (NH2) or the hydroxyl group (OH). Their labile hydrogen atoms thus excited are exchanged with unexcited hydrogen atoms of water.Such an application of a radiofrequency pulse, at a determined resonance frequency œ other than that œ0 associated with water, continuously repeated for a determined duration, for example a few seconds, leads to an accumulation of water saturation of a chemical species of interest. A concentration of the latter can be indirectly measured by the decrease in the water signal, a phenomenon called the “CEST effect” which is analyzed in the form of a Z-spectrum (“Z-spectrum” according to English terminology). Such a decrease can be easily detected by fast magnetic resonance imaging acquisition sequences such as single shot echo-planar imaging, also known as “Single Shot Echo-Planar Imaging”, or other acquisition techniques also known as “Gradient Echo Imaging”, “Fast Spin Echo Imaging”, “Turbo Spin Echo Imaging”.Figure 3 presents an example of a Z spectrum in the form of a set of samples Z(Aœ) (or "data set" according to English terminology) according to relative frequency shifts Aœ with respect to the water frequency, such shifts Aœ being expressed in ppm. When the samples Z(Aœ) are ordered according to decreasing Aœ, as indicated in Figure 3, said Z spectrum can be described in the form of a discrete signal describing measurements, advantageously normalized by a signal measured without saturation by radiofrequency and expressed as percentages, of a continuous signal delivered by a measuring device 1 for an elementary volume of an organ. According to Figure 3, this set of samples, when the latter are ordered according to decreasing frequency shifts Aœ, constitutes a discrete signal Z substantially describing a 'V' whose minimum Z(Aœ)m is associated with the frequency shift Aœ=1 ppm.Logically, said minimum Z(Aœ)m should be associated with a zero frequency shift Aœ, if said device was perfectly adjusted and / or the static magnetic field was homogeneous.

[0006] The CEST technique improves the detection of certain metabolites in the human body, the concentration of which is insufficient to be detected by traditional magnetic resonance imaging sequences. This CEST technique thus provides valuable information for a practitioner seeking to establish a diagnosis and make a therapeutic decision in the treatment of pathologies. In the clinical field, however, such biomarkers must be produced quickly and accurately. However, according to the state of the art, the implementation of the CEST technique requires a number of samples of several volumes to correct the B0 and B1 inhomogeneities. This high number of samples leads to a total acquisition time incompatible with the requirements imposed by the clinical field in particular.

[0007] In theory, to extract the signal from a molecule of interest, it would be sufficient to apply radiofrequency saturation to three locations in the Z spectrum:

[0008] - at the resonance frequency of the molecule (for APTw imaging, 3.5 ppm of the water frequency); we can call this volume S(3.5ppm) or “Label”;

[0009] - at the frequency opposite to that of the molecule with respect to the water frequency (for APTw imaging, -3.5 ppm of the water frequency); we can call this volume S(-3.5 ppm) or “Reference”;

[0010] - at a frequency very far from that of water, to obtain a volume without CEST effects, it is necessary to normalize the first two previous volumes (for APTw imaging we could for example be at 300 ppm from the frequency of water); We can call this volume S0.

[0011] Many effects interact with saturation transfer during continuous application of radiofrequency pulses, such as direct water saturation (DS) or magnetization transfer contrast (MT). Other phenomena combine with these effects and are inherent to the implementation of a magnetic resonance imaging technique, such as magnetic field inhomogeneities. Indeed, the CEST contrast is based on the chemical exchange between labile protons and water. The MT contrast comes from the magnetization exchange between protons in water and protons in a solid or semi-solid environment. Corrections are therefore necessary to obtain a reliable CEST signal.Furthermore, although in theory to extract a signal characterizing the concentration of a molecule of interest, a water saturation only requires three spectral localizations (S(3.5ppm), S(-3.5ppm), S0 for APTw imaging) in practice such an extraction is complex due to the presence of inhomogeneities of the static magnetic field B0 of magnetic resonance imaging devices. These inhomogeneities are partially caused by the tissue properties of the patient and lead to a shift of the acquired spectral samples, as shown in Figure 3.

[0012] To overcome this problem, dense sampling is performed around the resonance frequency of water, also including the frequency of the molecule of interest and the frequency opposite to that of the molecule. The S(Aœj) measurements acquired at the different frequencies Aœj are then normalized by a measurement acquired at a frequency very far from that of water (the volume S0). With an appropriate mathematical interpolation as indicated in Figure 4, the Z spectrum can be obtained, voxel by voxel, for a range of frequency shifts of interest, where Z(Aœ) = S(Aœ) / S0. The samples of the latter, after ordering according to decreasing frequency shifts Aœ, describe a discrete signal of shape or curve substantially in "V". Said Z spectrum, is then "refocused" so that its minimum Z(Aœ)m corresponds to the frequency of water as indicated in Figure 5.The Z spectrum can thus be described, in a relative form, the water frequency being equal to a zero frequency shift Aœ, the frequencies œ being expressed in ppm in the form of a relative frequency shift Aœ, positive or negative with respect to the water frequency. Such a relative frequency shift is generally expressed in ppm. A correction of the parameter B0, as illustrated in Figure 5, can thus be applied for a plurality of voxels or elementary volumes. Maps called intrinsic MBO maps, i.e. matrix images can be calculated, then displayed, using the offset position of the minimum Z(Aœ)m of the Z spectrum for a set of elementary volumes.

[0013] Figure 6 illustrates the determination of a first non-limiting example of a biomarker produced from a spectrum Z as previously described in connection with Figures 3 to 5. Such a biomarker may alternatively correspond to the spectrum Z as such. Such a biomarker may also correspond to the difference between a first area ZA1, which is generally called “Label” according to English terminology, under the interpolated curve of the spectrum Z from the samples Z(Aœ) and a second area ZA2, which is generally called “Reference” according to English terminology, under said curve, for respective frequency shift ranges Aœ which are symmetrical with respect to the zero frequency shift. In this case, in the example of figure 6, the area ZA1 is associated with frequency shifts between +3 ppm and +4 ppm and the area ZA2 is associated with frequency shifts between -3 ppm and -4 ppm.Such a choice of ranges is usually associated with amides, which are the biomarkers of interest in APTw imaging. Indeed, the resonance frequency of amides generally corresponds on average to the frequency shift Aœ=3.5 ppm. Alternatively, if we are interested in amines whose resonance frequency generally corresponds on average to the frequency shift Aœ=2 ppm, said ranges could be chosen as being respectively from 1.5 ppm to 2.5 ppm for the ZA1 area and from -1.5 ppm to -2.5 ppm for the ZA2 area. Parametric maps associated with this biomarker can be created and possibly displayed to describe said biomarker for a plurality of elementary volumes of interest.

[0014] Figure 7 illustrates the determination of a second non-limiting example of a biomarker produced from a spectrum Z as previously described in connection with Figures 3 to 5. Unlike the previous example according to which the biomarker is obtained by subtraction of two areas ZA1 and ZA2, said biomarker can be produced by subtraction of two interpolated values ​​of samples Z1 and Z2 of the spectrum Z, said samples also being chosen so that the frequency shifts Aœ1 and Aœ2 which are respectively associated with them are symmetrical with respect to the frequency shift Aœ=0. The value Z1 can be called “Label” according to English terminology and the value Z2 can be called “Reference” according to English terminology. As in the previous example, if we are interested in amides, as shown in Figure 7, the frequency shifts Aœ1=3.5ppm and Aœ2=-3.5ppm will be preferred.Alternatively or in addition, if we are interested in amides, we can choose frequency shifts Aœ1=2ppm and Aœ2=-2ppm. To extract the signal of the biomarker of interest, the value of Z1 (Label) is subtracted from that of Z2 (Reference). This subtraction (Reference-Label) theoretically makes it possible to eliminate spurious effects such as the DS and MT effect. Such a biomarker is called “Magnetic Transfer Ratio asymmetry (MTRasym)” according to English terminology, as disclosed in the reference “Zhou, J., Payen, JF, Wilson, DA, Traystman, RJ & Van Zijl, PCM Using the amide proton signals of intracellular proteins and peptides to detect pH effects in MRI. Nature Medicine. 2003. 9(8):1085-1090”.

[0015] Another methodology to extract the signal of the biomarker of interest can be used: the inverse of the Z2 value (1 / Reference) is subtracted from the inverse of the Z1 value (1 / Label). This subtraction (1 / Label - 1 / Reference) theoretically allows to more effectively remove spurious effects such as the DS and MT effect. Such a biomarker is called "Magnetic Transfer Ratio REX (MTRREX)" according to English terminology (cf. ZAISS, Moritz, et al. Inverse Z-spectrum analysis for spillover-, MT-, and T1 -corrected steadystate pulsed CEST-MRI-application to pH-weighted MRI of acute stroke. NMR in biomedicine, 2014, 27.3: 240-252).

[0016] Such choices may be subject to predetermined settings or options entered by a user using an appropriate human-machine input interface.

[0017] Increasing the density of Z-spectrum samples leads to a higher resolution of the calculated MBO maps, but acquiring multiple Z-spectrum samples results in a significant total acquisition time in the MRI machine and is not always synonymous with optimal B0 parameter mapping results over a set of volumes of interest.

[0018] In addition to the difficulties induced by the inhomogeneities of the static magnetic field B0 of magnetic resonance imaging devices, a Z spectrum is also affected by the inhomogeneities of the excitation magnetic field B1. To overcome this problem, a correction of the parameter B1 is also generally implemented to achieve the extraction of a reliable CEST signal.

[0019] To perform B1 correction, volumes that had already been sampled at different frequencies for B0 correction must also be sampled at different B1 radiofrequency values, further increasing acquisition times in the MRI machine, as specified in the reference “WINDSCHUH, Johannes, et al. Correction of B1 -inhomogeneities for relaxation-compensated CEST imaging at 7 T. NMR in biomedicine, 2015, 28.5: 529-537”.

[0020] To overcome these drawbacks, the WASAB1 sequence has been proposed to enable a more precise determination of the resonance frequency of bulk water protons. As mentioned previously, inhomogeneities of the static magnetic field B0 and / or the excitation magnetic field B1 are unavoidable in magnetic resonance imaging. They cause distortions in the intensity of the acquired images. It is therefore necessary to implement a post-processing method for a WASAB1 sequence to increase the relevance of said intensities by precisely determining spatial information related to the distribution of the magnetic fields B0 and B1 of the magnetic resonance imaging devices. Document W02022090644A1 illustrates such a post-processing method.

[0021] WASAB1 acquisition is a magnetic resonance imaging acquisition sequence that allows simultaneous mapping of the static magnetic field B0 and the excitation magnetic field B1, based on Rabi oscillations induced by off-resonance irradiation, which can be used to correct the previously mentioned inhomogeneities. The document PAPAGEORGAKIS CHRISTOS ET AL “Fast WASABI postprocessing: Access to rapid BO and B1 correction in clinical routine for CEST MRI”, MAGNETIC RESONANCE IMAGING - XP087365586 presents a method for accelerating the estimation of said inhomogeneity maps of such magnetic fields B0 and B1 based on a WASAB1 acquisition sequence without addressing any correction to remove the impact of such inhomogeneities on the acquired data.

[0022] There are also other acquisition methods for mapping the static magnetic field B0, such as the WASSR sequence (see KIM, Mina, et al. Water saturation shift referencing (WASSR) for chemical exchange saturation transfer (CEST) experiments. Magnetic Resonance in Medicine: An Official Journal of the International Society for Magnetic Resonance in Medicine, 2009, 61.6: 1441-1450), the Dixon sequence (see ZHANG, Shu, et al. CEST-Dixon for human breast lesion characterization at 3 T: A preliminary study. Magnetic resonance in medicine, 2018, 80.3: 895-903) or the BSS sequence for mapping the excitation magnetic field (see SACOLICK, Laura I., et al. B1 mapping by Bloch-Siegert shift. Magnetic resonance in medicine, 2010, 63.5: 1315-1322).

[0023] These known techniques not only allow to obtain a better mapping of the static field B0 but also to reduce the volumes acquired for a spectrum Z. Thus, sampling around the "Label" Z1 and around "Reference" Z2 can be sufficient to correct the field B0 to the frequency shifts of interest (3.5 and -3.5 ppm for example). However, the acquired volumes remain very numerous. Generally, to avoid performing an inefficient or irrelevant interpolation to correct the impact of the inhomogeneity of the field B0, five volumes Z(Aœ) around the "Reference" and five volumes Z(Aœ) around "Label" are necessary: ​​for example, Z(-4.5ppm), Z(-4ppm), Z(-3.5ppm), Z(-3ppm), Z(-2.5ppm), Z(2.5ppm), Z(3ppm), Z(3.5ppm), Z(4ppm), Z(4.5ppm). On average, the time required to acquire a CEST volume is around fifteen seconds.It is understood that these ten volumes Z(Aœ) have already been normalized by the volume S0 (eleventh volume to be acquired), according to the formula Z(Aœ)=S(Aœ) / S0.

[0024] The numerous acquisitions carried out to correct the inhomogeneities of the magnetic fields B0 and B1 which must be implemented, elementary volume by elementary volume, lead to a prohibitively high acquisition time for an application in the clinical field. For example, the first step consisting of correcting the inhomogeneities of the static magnetic field B0, as mentioned previously, can generate an acquisition time by the magnetic resonance imaging device of between two and three minutes.If we add the time required to correct the inhomogeneities of the excitation magnetic field B1, which generally requires, per volume of interest, to iterate the ten acquisitions (i.e. five for "Label" and five for "Reference", which we assume to be acquired at a nominal value B1 equal to 2pT in the previous example) for at least five values ​​of the excitation magnetic field B1 (such as 1.2pT, 1.6pT, 2pT, 2.4pT, 2.8pT), the acquisition time rises to fifteen minutes for the most efficient systems (such as the one in the example described) to almost an hour for the least efficient.The technique set out in the document MÜLLER-LUTZ ANJA ET AL: “Comparison of B0 versus BO and B1 field inhomogeneity correction for glycosaminoglycan chemical exchange saturation transfer imaging”, Magnetic Materials In Physics, Biology and Medicine, XP06589490 or that disclosed in the document US 2012 / 019245 A1, illustrates the need to use multiple acquisitions for distinct values ​​of B1, thus penalizing the acquisition time. For this reason, most practitioners forgo the correction of the inhomogeneity of the excitation magnetic field B1.

[0025] Figure 8 visually illustrates the impact of no correction or partial correction of said inhomogeneities of the magnetic fields B0 and B1. Said figure 8 presents on the left view, a parametric map MBMK illustrating a biomarker BMK “Reference-Label” produced from a spectrum Z as previously described in connection with figures 3 to 5 and 7 for a set of elementary volumes. Said biomarker is obtained by subtraction of two interpolated values ​​of samples Z1 and Z2 of the spectrum Z, said samples also being chosen for frequency shifts Aœ1 =3.5 ppm and Aœ2=-3.5 ppm symmetrical with respect to the frequency shift Aœ=0. In the center and on the right of said figure 8, the inhomogeneity maps MBO and MrB1 are represented, respectively of the magnetic fields B0 and B1, for the same set of elementary volumes.A homogeneous static B0 field would induce a homogeneous MBO map for which the N voxels would describe a value of 0 ppm. However, we can see more particularly on the MBO map of figure 8, four zones MB0-1, MBO-2, MBO-3 and MBO-4 for which the values ​​are negative (between 0 and -0.3 ppm) and one zone MBO-5 where the voxel values ​​reach 0.3 ppm. This inhomogeneity of the static magnetic field B0 also induces on the MBMK map (left part of figure 8) four zones MBMK-1 to MBMK-4 for which the signal is very weak or non-existent. If the patient suffers from a disorder in these zones, a diagnosis could be clearly altered. Furthermore, the MrB1 map (on the right in Figure 8) clearly illustrates the inhomogeneity of the excitation magnetic field B1 (while the voxels of said MrB1 map should substantially describe a value equal to 1 if said excitation magnetic field B1 were homogeneous).However, said voxels describe, outside of a more or less central zone referenced MrB1-C, lower voxel values ​​on the periphery of said MrB1-C zone and more particularly in the left frontal zone, profoundly altering the accuracy of the MBMK map. Document US2022 / 342021 describes an effective technique for correcting inhomogeneities in the magnetic field B0 by relying mainly on a shift in the values ​​of the magnetic field B0. Said document briefly mentions, without detailing any technical basis, modalities or adaptations of application, that it would be possible to exploit such a technique for correcting inhomogeneities in the magnetic field B0 to treat inhomogeneities in the excitation magnetic field B1.However, in light of the technical teaching that can be objectively drawn from said document US2022 / 342021, it is clear that a transposition of said technical teaching proves to be ineffective in correcting the effects of an inhomogeneity of the excitation magnetic field B1, or even leads to serious biases in the CEST signal. This document US2022 / 342021 does not describe, nor claim, any example of correction of the inhomogeneities of the excitation magnetic field B1.

[0026] The invention makes it possible to address all or part of the drawbacks raised by the known or previously mentioned solutions.

[0027] Among the many advantages brought by the invention, we can mention that:

[0028] - a post-processing method according to the invention for correcting the inhomogeneities of the magnetic field B1, which only requires the map MrB1 and the map of interest to be corrected MBMK, avoiding the multiple acquisitions of other volumes necessary for the interpolations in the radiofrequency domain expressed in units of pT (to process the inhomogeneities of the excitation magnetic field B1) required by the known methods, the acquisition time being divided by at least twenty-five thanks to the implementation of the invention;

[0029] - the implementation of such a post-processing process opens up to the clinical field, in particular CEST technology, which is particularly promising for providing decision-making and diagnostic support in connection with numerous pathologies;

[0030] - the implementation of a post-processing method according to the invention finds a direct and efficient application in any magnetic resonance imaging application for which a correction of inhomogeneities of the excitation magnetic field B1 is required, such as, for example, anatomical acquisitions known under the English terms T1-weighted image (T1WI) or T2-weighted image (T2WI), the invention making it possible to prevent the use of inhomogeneity correction algorithms according to the state of the art whose implementations require an execution time of several tens of seconds or expensive hardware resources, such as the algorithm disclosed in the publication NJ Tustison, BB Avants, PA Cook, Y. Zheng, A. Egan, PA Yushkevich, and JC Gee. “N4ITK: Improved N3 Bias Correction” IEEE Transactions on Medical Imaging, 29(6):1310-1320, June 2010.

[0031] To this end, the invention firstly relates to a method for post-processing a BMK biomarker produced from a plurality of M volumes Z(Aœj), each of said M volumes Z(Aœj) comprising N voxels respectively describing the magnitude of an experimental signal sampled at a frequency shift Aœj relative to the resonance frequency of water during an acquisition sequence by a magnetic resonance imaging device, said method being implemented by a processing unit of an imaging analysis system. To open up the possibility of correcting the effect of magnetic field inhomogeneities of the imaging device on the production of biomarkers in the clinical field in particular, such a method comprises:

[0032] - a step of producing the BMK biomarker in the form of an MBMK map composed of N voxels, where each voxel i, i being between 1 and N, describes a MBMK signal value[i];

[0033] - a step of producing a map MrB1 describing the inhomogeneity of the excitation magnetic field B1 of said magnetic resonance imaging device for the M volumes, said maps each comprising N voxels; - a step of estimating the parameters of a function Fc such that MBMK[i]=Fc(MrB1[i]);

[0034] - a step of production of the corrected BMK biomarker in the form of an MBMK' map of N voxels such that MBMK'[i]=MBMK[i]-Fc(MrB1 [i])+ Fc(1 ).

[0035] According to an advantageous embodiment, the step of estimating the parameters of a function Fc such that MBMK[i]=Fc(MrB1[i]) may consist of implementing a least squares minimization on all or part of the set of N voxels of the MBMK and MrB1 maps.

[0036] When the biomarker produced is particularly affected by an inhomogeneity of the magnetic field B0, a method according to the invention may comprise, prior to the step of producing the corrected biomarker, a step of correcting the effect of the inhomogeneity of the static magnetic field B0 on ​​said biomarker produced.

[0037] To provide a representation perceptible by humans of a corrected biomarker, such a method in accordance with the invention may comprise a step of producing graphic content describing the MBMK' parametric map whose N voxels respectively encode the values ​​of the corrected biomarker, then a step of triggering a display of said graphic content by a human-machine output interface of the imaging analysis system.

[0038] According to a second object, the invention relates to an imaging analysis system comprising a processing unit, means for communicating with the outside world and storage means. Such a system is designed so that:

[0039] - the means for communicating are arranged to receive from said external world experimental data resulting from an acquisition sequence by a magnetic resonance imaging device;

[0040] - the storage means comprise program instructions, the interpretation of which by said processing unit causes the implementation of a method of a method for post-processing a biomarker according to the invention. According to a third subject, the invention relates to a computer program product comprising one or more instructions interpretable by the processing unit of such an imaging analysis system, said computer program being loadable into storage means of said system and arranged so that the interpretation of said program instructions by said processing unit causes the implementation of a method for post-processing a biomarker according to the invention.

[0041] According to a fourth object, the invention relates to a computer-readable storage medium comprising the program instructions of such a computer program product.

[0042] Other features and advantages will become more apparent upon reading the following description and examining the accompanying figures, including:

[0043] - figure 1, already described, illustrates a simplified description of a system for analyzing images obtained by Nuclear Magnetic Resonance;

[0044] - figure 2, already described, illustrates a simplified description of a variant of a system for analyzing images obtained by Nuclear Magnetic Resonance;

[0045] - figure 3, already described, presents a first example of a Z spectrum, describing the magnitude of an experimental signal sampled at different frequency shifts Am relative to the resonance frequency of water during an acquisition sequence by a magnetic resonance imaging device, said frequency shifts Aœ being ordered in decreasing order;

[0046] - Figure 4, already described, presents the same example of a Z spectrum in the form of a set of samples Z(Aœ) ordered like those described by Figure 3, superimposed on a continuous experimental signal estimated after fitting a determined model to the samples of said Z spectrum in accordance with the state of the art; - Figure 5, already described, presents the same example of a Z spectrum in the form of a set of samples Z(Aœ) ordered such as in Figure 3, superimposed on a continuous experimental signal estimated after fitting a determined model to the samples of said Z spectrum in accordance with the state of the art, after correction of the inhomogeneity of the B0 field;

[0047] - Figure 6, already described, presents an estimation of a first biomarker from samples of a Z spectrum in accordance with the state of the art after correction and adjustment of the imaging device;

[0048] - Figure 7, already described, presents an estimation of a second biomarker from samples of a Z spectrum in accordance with the state of the art after correction and adjustment of the imaging device;

[0049] - Figure 8, already described, shows the impact of inhomogeneities in static and excitation magnetic fields on the production of a biomarker;

[0050] - figure 9 illustrates a first example of correlation between the production of such a biomarker and the excitation magnetic field B1 of the imaging device;

[0051] - figure 10 illustrates the implementation of a decorrelation of said excitation magnetic field B1 on the production of such a corrected biomarker according to the invention;

[0052] - Figure 11 presents a second example of correlation between the production of such a biomarker and the excitation magnetic field B1 of the imaging device using Gaussian distributions fitted to the point cloud;

[0053] - figure 12 presents probability distribution maps respectively associated with the Gaussian distributions illustrated by figure 11 applied to a human brain;

[0054] - figure 13 presents a comparison of parametric maps describing respectively a first biomarker corrected to take into account only the inhomogeneity of the static magnetic field B0 of the imaging device on the production of said first biomarker and to take into account in addition an inhomogeneity of the excitation magnetic field B1 of said imaging device;

[0055] - Figure 14 shows a comparison of parametric maps describing a second biomarker corrected to take into account only the inhomogeneity of the static magnetic field B0 of the imaging device on the production of said second biomarker and to further take into account an inhomogeneity of the excitation magnetic field B1 of said imaging device;

[0056] - figure 15 illustrates an example of a functional algorithm of a post-processing method according to the invention.

[0057] Let us now describe in connection with Figure 15, an example of a post-processing method 100, according to the invention, of a parametric map MBMK describing a biomarker BMK produced from a plurality of M volumes Z(Aœj), each of said M volumes Z(Aœj) comprising N voxels respectively describing the magnitude of an experimental signal 10, 12 sampled at a frequency shift Aœj relative to the resonance frequency of water during an acquisition sequence by a magnetic resonance imaging device (such as the device 1 described previously in connection with Figure 1 or 2) in connection with a set of elementary volumes of interest of an organ. A CEST type acquisition, chosen to preferably illustrate a method according to the invention, induces several modules Z(Aœj). However, M can also be reduced to the value 1 in the context of anatomical acquisitions (T1WI, T2WI, etc.) or other multiphase acquisitions such as ASL acquisition (acronym for “Arterial Spin Labeling”) or functional magnetic resonance imaging (also known as “functional magnetic resonance imaging”).

[0058] Such a method 100 comprises a first step 110 of producing, for each voxel i, the values ​​of the MBMK[i] map of said BMK biomarker for a set of i voxels, calculated across M elementary volumes of interest. Such a BMK biomarker may be the “Reference-Label” biomarker produced from, for each voxel i, a spectrum Zi - such as the spectrum Z illustrated previously by figures 3 to 5 and 7 - for a set of N voxels and M elementary volumes. As a reminder, such a BMK biomarker is obtained by subtracting two interpolated values ​​of samples Z1 and Z2 from the Zi spectrum of voxel i in M ​​volumes, said samples also being chosen for frequency shifts Aœ1 =3.5 ppm and Aœ2=-3.5 ppm symmetrical with respect to the frequency shift Aœ=0. Such a BMK biomarker could further be defined as “1 / Label - 1 / Reference” to correct the data for the Spillover effect, known in CEST imaging (cf. https: / / analyticalsciencejournals.onlinelibrary.wiley.com / doi / abs / 10.1002 / nbm .3054 - or Inverse Z-spectrum analysis for spillover-, MT-, and T1-corrected steady-state pulsed CEST-MRI - application to pH-weighted MRI of acute stroke. 2014. NMR Biomed., 27: 240-252. Zaiss, M., Xu, J., Goerke, S., Khan, IS, Singer, RJ, Gore, JC, Gochberg, DF and Bachert, P.”

[0059] The biomarker could further consist simply of the Label or Reference volume. Whatever the biomarker, we will denote by BMK[i] the value of said BMK biomarker for the i ème volume of interest.

[0060] Such a method 100 in accordance with the invention further comprises a step 120 of producing values ​​of an excitation magnetic field B1 of the imaging device 1. Such a step 120 thus consists of producing the intrinsic map MrB1, i.e., for each of the N voxels i (1 <i<N) de cette dernière, les valeurs MrB1 [i] de ladite carte MrB1. Pour cela, l’invention prévoit d’appliquer toute technique connue de production d’une telle carte intrinsèque MrB1 , par exemple, la technique exposée dans le document PAPAGEORGAKIS CHRISTOS ET AL « Fast WASABI post-processing : Access to rapid B0 and B1 correction in clinical routine for CEST MRI », MAGENTIC RESONANCE IMAGING - XP087365586.When a method according to the invention consists, optionally, in correcting, in addition to the effects of an inhomogeneity of the excitation magnetic field B1, the effects induced by an inhomogeneity of the static magnetic field B0, for example by relying on a technique expressed in document US2022 / 342021 A1 or other, said step 120 may consist of producing values ​​of said static magnetic field B0 of the imaging apparatus 1. In this case, such a step 120 may also consist of producing an intrinsic map MB0, i.e., for each of the N voxels i (1 <i<N) de cette dernière, les valeurs MB0[i].

[0061] The inventors were able to observe, as indicated in figures 9 and 11, that the production of the MBMK map of the BMK biomarker whatever it may be (for example “Reference-Label”) is correlated with the values ​​of the voxels N of the intrinsic map MrB1 of the magnetic field B1 of the imaging device.

[0062] Figure 9 thus presents a scatter plot with, on the abscissa, the values ​​of the MrB1 map of the inhomogeneities of the excitation magnetic field B1 and, on the ordinate, the values ​​of the MBMK parametric map describing the BMK biomarker (“Reference-Label”). The correlation is obvious. Visually, we can easily deduce a CL line which symbolizes this correlation. The inventors therefore concluded that the values ​​of the MBMK[i] voxels of the MBMK map of the BMK biomarker can be expressed in the form of a function Fc of the voxels of said MrB1 map of inhomogeneity of the excitation magnetic field B1. If we manage to estimate the parameters of such a function Fc, then, to decorrelate in some way, said biomarker BMK from the excitation magnetic field B1, it suffices to correct each value of said biomarker by removing from it the value obtained by the application of said function Fc to said excitation magnetic field B1 for the same elementary volume of interest.Thus, we can express the corrected value of said biomarker in the form of the mathematical expression MBMK'[i]=MBMK[i]-Fc(MrB1[i])+c, c being a constant. A method 100 according to the invention thus comprises a step 130 of estimating the parameters of such a function Fc such that MBMK[i]=Fc(MrB1[i]).

[0063] As shown in Figure 15, a method 100 according to the invention can be applied to eliminate the impact of the inhomogeneity of the excitation magnetic field B1, independently, i.e. solely or subsequently, of a correction of the effects induced by an inhomogeneity of the magnetic field B0 of the imaging apparatus. Thus, such a method 100 can be applied to the raw data MBMK of the biomarker BMK (uncorrected, if said biomarker BMK does not suffer from an inhomogeneity of the magnetic field B0) or else to the data of said biomarker corrected MBMK' after an implementation of an optional step 140' of correction of said raw data MBMK to remove the effect of the inhomogeneity of the static magnetic field B0 of the imaging device, said optional step 140' being represented in a broken line in FIG. 15 to underline its optional nature.

[0064] Advantageously, the constant c will be chosen as equal to Fc(1 ) i.e. Fc(MrB1 =1 ). Alternatively, such a constant c could be calculated as resulting from the mean or median of values ​​of the biomarker BMK in an area of ​​the organ (on which the CEST acquisition is carried out) not or very little affected by the inhomogeneity of the excitation magnetic field B1. Any other technique could be implemented so as not to generate a bias in the values ​​of the biomarker MBMK' corrected by the invention. The invention provides an advantageous embodiment for segmenting the correlation between the biomarker BMK and the inhomogeneity of the excitation magnetic field B1 and thus optimizing the estimation of the parameters of the function Fc at step 130. Such an advantageous technique is detailed later in connection with figures 11 and 12.The implementation of such a segmentation technique could be used, according to an embodiment of the invention, to determine one or more areas of the organ not affected or very little affected by the inhomogeneity of the excitation magnetic field B1 in order to calculate the value of said constant c as an average, median, etc., as mentioned previously.

[0065] A method 100 in accordance with the invention then comprises a step 140 of producing the value MBMK'[i], for each voxel i, of said corrected biomarker BMK' for each of the set of N voxels such that MBMK'[i]=MBMK[i]- Fc(MrB1[i])+c, or in this case in the example linked to figures 9 and 13: MBMK'[i]=a*MrB1[i]+b. Figure 10 illustrates a new point cloud corresponding to the values ​​of the previous cloud thus corrected, said new point cloud being superimposed on said previous cloud before correction. A new horizontal DCL line now thus symbolizes the decorrelation of the biomarker BMK from the excitation magnetic field B1 by the application of step 140.

[0066] The nature of the Fc function can be variable (linear, quadratic, etc.) but it must be polynomial on an invertible monotonic transformation of the CEST signal considered. As non-limiting examples, such a Fc function can consist of an inverse function, an exponential function, or a logarithmic function.

[0067] Figure 11 illustrates the fact that an excitation magnetic field B1 can affect the different tissues of the organ considered differently, in this case brain tissues. When a tissue mask is not available, step 130 of a method 100 according to the invention may comprise a sub-step 131 consisting of an exploitation of the MBMK and MrB1 maps to segment the image reflecting the correlation between said biomarker BMK and an inhomogeneity of the magnetic field B1. As shown in Figure 11, a bivariate Gaussian mixture model (also known by the acronym GMM for the English expression “Gaussian Mixture Model”) can be exploited to adjust the point cloud formed by placing the values ​​of the excitation magnetic field B1 on a first horizontal axis and the values ​​of the biomarker BMK on a second vertical axis.To improve the determination of the GMM model, it may be advantageous to adapt the values ​​of the magnetic field B1 to the range of values ​​of the biomarker BMK, indicated B1' in Figure 11. The number of Gaussian distributions to be fitted is fixed at three in the example illustrated by Figure 11 (Gaussian distributions GD1, GD2 and GD3), but this number can vary depending on the data. Several algorithms can be used to fit the GMM model, including the expectation-maximization algorithm (also known as the "Expectation-Maximization algorithm"). Figure 12 illustrates probability maps DPMI, DPM2, DPM3, associated with each of the three Gaussian distributions GD1, GD2 and GD3 obtained from the data described in connection with Figure 11.We can easily see visually that the Gaussian distribution GD2, more compact and homogeneous than the other two GD1 and GD3, corresponds approximately to the white matter of the brain tissue as illustrated by the map DPM2. Step 130 of a method 100 according to the invention can thus comprise a sub-step for selecting the smallest identified Gaussian distribution (i.e. the one for which the product between its eigenvalues ​​is the lowest), in this case the distribution GD2, which covers enough voxels of the organ considered, in this case in Figures 11 and 12, a human brain, to avoid borderline and irrelevant cases. Such a minimum coverage of voxels by such a Gaussian distribution may be set at 30% of the voxels of the brain. Such a minimum coverage threshold of 30% cannot, however, constitute a limit for the present invention. Such a threshold could be lower than 20% or higher than the 30% mentioned.The estimation of the parameters of the function Fc, in step 130, can thus advantageously be implemented on only the voxels thus characterized by the determination of the Gaussian distribution GD2. As mentioned previously, such a determination of a GMM model could be used to identify the voxels of interest used to calculate the constant c, if this is not c=Fc(1), but results from a calculation of an average or a median.

[0068] Whatever the step 140 for producing the value MBMK'[i], for each voxel i, of said corrected biomarker BMK' for each of the set of N voxels such that MBMK'[i]=MBMK[i]-Fc(MrB1[i])+c, a method 100 according to the invention may comprise a step 150 for producing a graphic content describing the parametric map MBMK' whose voxels respectively encode the values ​​of the corrected biomarker BMK' for all or part of the set M of elementary volumes of interest then a step 160 for causing a display, in the form of an image or, more generally, an output of said graphic content associated with the map MBMK' by a human-machine output interface of the imaging analysis system AS.We can note and emphasize, through the exemplary embodiment described in connection with figure 15, that an implementation of a method 100 in accordance with the invention requires an acquisition 110 for a single determined value of the excitation magnetic field B1, unlike prior solutions which require multiple acquisitions for distinct values ​​of B1. The invention thus considerably reduces the time required for the correction of the BMK biomarker to eliminate the effect on the latter of the inhomogeneity of the excitation magnetic field B1.

[0069] Figures 13 and 14 illustrate the contribution of the invention to eliminate the effects of the inhomogeneity of the excitation magnetic field B1 after a first correction to eliminate the effects of the inhomogeneity of the static magnetic field B0. Said figure 13 presents two parametric maps MBMK' of a biomarker (in this case "Reference-Label") of a human brain and an intrinsic map MrB1 reflecting a strong inhomogeneity of the excitation magnetic field B1 of the imaging device considered. On the left, a first map MBMK' is illustrated after correction to eliminate the sole effect of the inhomogeneity of the field B0 according to any known technique. On this first map, areas referenced MBMK-1 and MBMK-2 are clearly distinguished in which the signal is weak or non-existent given the impact of the inhomogeneity of the magnetic field B1 described by the map MrB1 positioned on the right of said figure 13.In the center of Figure 13, a second MBMK' parametric map of the same BMK biomarker is shown after correction according to the invention to remove the inhomogeneity of the excitation field B1. We can see the contribution of the correction according to the invention. The MBMK-1 and MBMK-2 zones present on the MBMK' map on the left have disappeared on the MBMK' map in the center. Indeed, no zone within the latter contains pixels associated with non-existent signals. On the contrary, said corrected MBMK' map according to the invention also highlights an area of ​​interest MBMK-3 for the practitioner 6 likely to influence his diagnosis. Such an MBMK-3 zone would have gone unnoticed without the correction to remove the effect of the inhomogeneity of the field B1. Figure 14 also illustrates the contribution of the invention in a clinical context.Such a figure 14 presents two parametric maps MBMK and MBMK' of a BMK biomarker (in this case "Reference-Label") of a human brain. On the left, a first MBMK map without correction is illustrated, on which many artifacts can be clearly seen given the impact of the inhomogeneity of the excitation magnetic field B1 illustrated by the MrB1 map present on the right of said figure 14. In the center of figure 14, a parametric map MBMK' of the same BMK biomarker after correction according to the invention is represented. We can see the effect of the correction provided by the invention. The MBMK' representation no longer presents artifacts and its pixels are more homogeneous and richer in their gradation of values.

[0070] In addition to the very high precision and increased relevance of the MBMK' graphic representations made available to practitioners, the implementation of a post-processing method for a first set Z of samples Z(Aœ) of an experimental signal resulting from an acquisition sequence by a medical magnetic resonance imaging device 1 becomes perfectly usable in the clinical field, thus opening the door to said clinical field in particular to the CEST technique, to quantify one or more biomarkers linked to a chemical species of interest of a human or animal organ to ultimately characterize different lesions with altered metabolic properties. The invention cannot be reduced to this sole exploitation. It thus finds a place of choice in any magnetic resonance imaging application for which a correction of inhomogeneities of the excitation magnetic field B1 is required.

Claims

CLAIMS 1. Method (100) for post-processing a biomarker (BMK) produced from a plurality of M volumes Z(Aœj), each of said M volumes Z(Aœj) comprising N voxels respectively describing the magnitude of an experimental signal sampled at a frequency shift Aœj relative to the resonance frequency of water during an acquisition sequence by a magnetic resonance imaging device (1), said method being implemented by a processing unit (4) of an imaging analysis system (AS), characterized in that it comprises: - a step (110) of producing the biomarker (BMK) in the form of an MBMK map composed of N voxels, where each voxel i, i being between 1 and N, describes a signal value MBMK[i]; - a step (120) of producing a map MrB1 describing the inhomogeneities of the excitation magnetic field (B1) of said magnetic resonance imaging device (1) for the M volumes, said map MrB1 comprising N voxels; - a step (130) of estimating the parameters of a function Fc such that MBMK[i]=Fc(MrB1[i]); - a step (140) of producing the corrected biomarker (BMK) in the form of an MBMK' map of N voxels such that MBMK'[i]=MBMK[i]-Fc(MrB1 [i])+ Fc(1 ).

2. Method (100) according to any one of the preceding claims for which the step (130) of estimating the parameters of a function Fc such that MBMK[i]=Fc(MrB1[i]) consists of implementing a least squares minimization on all or part of the set of N voxels of the MBMK and MrB1 maps.

3. Method according to any one of the preceding claims, comprising a step (150) of producing graphic content describing the MBMK' parametric map whose N voxels respectively encode the values of the corrected biomarker.

4. Method according to the preceding claim comprising a step (160) for causing a display of said graphic content by an output human-machine interface (5) of the imaging analysis system (AS).

5. Method according to any one of the preceding claims comprising, prior to the step (140) of producing the corrected biomarker (BMK), a step (140') of correcting the effect of the inhomogeneity of the static magnetic field B0 on the biomarker (BMK) produced (110).

6. Imaging analysis system (AS) comprising a processing unit (4), means for communicating with the outside world and storage means, characterized in that: - the means for communicating are arranged to receive from said external world experimental data resulting from an acquisition sequence by a magnetic resonance imaging device (1); - the storage means comprise program instructions whose interpretation by said processing unit causes the implementation of a method (100) according to any one of the preceding claims.

7. Computer program product comprising one or more instructions interpretable by the processing unit (4) of an imaging analysis system (AS) according to the preceding claim, said computer program being loadable into storage means of said system, characterized in that the interpretation of said program instructions by said processing unit causes the implementation of a method (100) according to any one of claims 1 to 5.

8. Computer-readable storage medium comprising the program instructions of a computer program product according to the preceding claim.

Images