METHOD AND SYSTEM FOR CAPTURING AND RECONSTRUCTING IMAGES OF A HEART DURING FREE BREATHING

Description

Scope of the invention

[0001] The invention relates to the field of magnetic resonance imaging, in particular cardiac magnetic resonance imaging. More specifically, it relates to late gadolinium enhancement (LGE) magnetic resonance imaging (MRI). It is applicable to cardiac imaging, but also to angiography. The scope of the invention relates more particularly to imaging methods and devices for detecting and locating scarring in myocardial tissue. State of the art

[0002] The gold standard for evaluating regional scar formation and myocardial fibrosis is white blood late gadolinium enhancement (BL-LGE) imaging. In this imaging technique, viable myocardial signal is suppressed using inversion-recovery pulses, allowing visualization of scars with high contrast between healthy myocardial tissue and the scar tissue. However, for small scars or myocardial scars adjacent to the heart's blood chambers, the high intensity of the blood signal prevents clear visualization and precise delineation of the scars, particularly subendocardial scars. To overcome this limitation, black blood LGE (BL-LGE) imaging techniques have been developed.They allow for the simultaneous cancellation of signals from healthy myocardium and blood, thus providing high contrast both between scars and between blood and between scars and healthy myocardium.

[0003] The cancellation of the blood signal, to obtain a "black blood" contrast, is achieved by applying a radiofrequency (RF) inversion-recovery sequence to the area of ​​a patient to be imaged. This sequence consists of a 180° pulse followed by a preparatory module and a readout module. The time between the preparatory module, which occurs immediately after the emission of the radiofrequency inversion pulse, and the readout sequence is called the inversion time (IT).

[0004] We seek to acquire the signals to generate the image of the area to be imaged in order to locate and detect a scar present in the myocardial tissues.

[0005] Du document Ginami Giulia et AL : "3D Whole-heat phase sensitive inversion recovery CMR for simultaneous black-blood late gadolinium enhancement and bright-blood coronary CMR angiography", Journal of Cardiovascular Magnetic Resonance, vol 19, no 1, 27 novembre 2017, on connaît une solution en acquisition sang noir- sang blanc.

[0006] One drawback of prior art solutions, such as black-blood / white-blood acquisition, is their significant acquisition time due to the 3D acquisition process, which involves numerous calculations. Under these conditions, the patient is breathing freely. Consequently, image correction algorithms, requiring substantial computing resources, must be applied, lengthening the acquisition sequence and making real-time application difficult. Furthermore, when acquisition time is extended, the variability of respiratory responses leads to image processing artifacts that impair image clarity. These artifacts further complicate the reconstruction of clear and precise images for scar localization and detection. In addition, long MRI acquisitions are uncomfortable for the patient.A duration of 10 to 20 minutes is considered a very long time and it is difficult for the patient to remain inside the MRI without moving.

[0007] Today, these sequences require an acquisition time of over 15 minutes. The acquired images are unsatisfactory in many cases because artifacts in an image extracted from a cross-section of the 3D image lead to situations where it is impossible to distinguish the presence of a potential scar from the presence of blood located near the muscle. Indeed, in some cases, the scar is so close to the blood—it is called subendocardial—that it is difficult to determine whether it is a scar, blood, or an image artifact. A problem arises when clinical decisions, which can lead to diagnostic errors, are based on the presence or absence of a scar.

[0008] Another solution is to perform a rapid acquisition while holding one's breath, but this solution entails the need to reduce the acquisition time from a few seconds to a few minutes and often involves having to reduce the area to be imaged, for example by limiting oneself to a 3D portion.

[0009] However, a minimum number of images are necessary to reconstruct a 3D portion. One problem is that heartbeats also influence the measurement and require synchronized acquisitions. This constraint necessitates reducing the potential number of acquisitions over a given time period. Indeed, it is necessary to sequence the acquisitions based on the heart rhythm, thus spreading these acquisitions over a minimum number of beats. Acquisitions during breath-holding severely limit the acquisition capacity.

[0010] Finally, 2D acquisition methods exist, but these do not provide sufficient resolution and contrast to detect and locate scars. Multiple 2D acquisitions of the same area would be necessary to resolve certain cases, but variations in respiration and heart rate over time prevent the acquisition of 2D images of identical cross-sectional planes of the organ over time. Summary of the Invention and Advantages

[0011] According to a first aspect, the invention relates to a method for reconstructing an image of a patient's heart from a magnetic resonance imaging device generating a magnetic field comprising: A first phase of image generation comprising: ■ Acquisition of the patient's heart electrical activity; ■ Generation of a first 180° inversion radiofrequency signal to flip a longitudinal magnetization of tissues in an imaged area, said first inversion radiofrequency signal being generated between two QRS complexes of the acquired electrical activity, called the first inter-beat phase; ■ Generation of a first magnetization preparation comprising a set of pulses following the generation of the first inversion radiofrequency signal in the same first inter-beat phase; ■ Acquisition of a first 2D image by a magnetization measurement of the imaged area, said acquisition being carried out after a first predefined duration following the generation of the first magnetization preparation, said acquisition being synchronized with the heart rhythm at a first temporal marker of the same inter-beat phase;■ Generation of a second magnetization preparation comprising a set of pulses in a second interbeat phase following the first interbeat phase; ■ Acquisition of a second 2D image by a magnetization measurement of the imaged area, said acquisition being performed after a second predefined duration following the generation of the second magnetization preparation, said acquisition being synchronized with the acquisition of a heart rhythm at the same first time marker of the second interbeat phase as the first marker of the first interbeat phase; Repetition of the first phase to generate a plurality of first images and a plurality of second images for a plurality of slice planes of the heart in interbeat phases following the first and second interbeat phases, each pair of first and second images being associated with a slice plane;

[0012] Generation of a set of third 2D images for each cutting plane, each third image corresponding to a first combination made between each pair of first merged images and second merged images for each cutting plane.

[0013] This aspect of the invention relates to the case where one image per slice plane is acquired. This possibility is available for rapid acquisitions requiring approximately thirty beats to cover about fifteen slice planes with one black blood image and one white blood image per slice. However, this sequence can be improved by free-breathing acquisition by acquiring several images per slice plane, one for black blood and one for white blood.

[0014] According to another aspect, the invention relates to a method for reconstructing an image of a patient's heart from a magnetic resonance imaging device generating a magnetic field comprising: A first phase of image generation comprising: ■ Acquisition of the patient's heart electrical activity; ■ Generation of a first 180° radiofrequency inversion module to flip a longitudinal magnetization of an imaged area, said first radiofrequency inversion signal being generated between two QRS complexes of the acquired electrical activity, called the first inter-beat phase; ■ Generation of a first magnetization preparation comprising a set of pulses following the generation of the first radiofrequency inversion module in the same first inter-beat phase; ■ Acquisition of a first 2D image by a magnetization measurement of the imaged area, said acquisition being carried out after a first predefined duration following the generation of the first magnetization preparation, said acquisition being synchronized with the heart rhythm at a first temporal marker of the same inter-beat phase;■ Generation of a second magnetization preparation comprising a set of pulses in a second interbeat phase following the first interbeat phase; ■ Acquisition of a second 2D image by a magnetization measurement of the imaged area, said acquisition being carried out after a second predefined duration following the generation of the second magnetization preparation, said acquisition being synchronized with the acquisition of a cardiac rhythm at the same first time marker of the second interbeat phase as the first marker of the first interbeat phase; Repetition of the first phase to generate a plurality of first images and a plurality of second images for a plurality of slice planes of the heart in interbeat phases following the first and second interbeat phases, each pair of first and second images being associated with a slice plane;Repetition of the first phase to generate a plurality of first and second images of the same cross-sectional plane of the heart in inter-beat phases following the first and second inter-beat phases.

[0015] In other words, the process includes repeating the first phase, to generate a subset of first images and a subset of second images of each slice plane of a plurality of slice planes of the heart, in inter-beat phases following the first and second inter-beat phases.

[0016] The process also includes: Application of a non-rigid registration algorithm to each subset of first images of the same cutting plane to register said first images with each other; Application of a non-rigid registration algorithm to each subset of second images of the same cutting plane to register said second images with each other; Merging on the one hand the registered first images with each other to produce a first merged image and on the other hand the registered second images with each other to produce a second merged image; Generation of a set of third 2D images for each cutting plane, each third image corresponding to a first combination made between each pair of merged first images and merged second images for each cutting plane.

[0017] One advantage is that it allows for the production of short-duration free-breathing acquisitions that do not require the computation time of a 3D acquisition.

[0018] Advantageously, the process includes the following steps: fusion, for each of the cutting planes, of on the one hand the first images registered together to produce a first merged image and on the other hand the second images registered together to produce a second merged image; generation, for said cutting planes, of a set of third 2D images, each third image corresponding to a first combination made between the first merged image and the second merged image produced for one of the cutting planes.

[0019] Advantageously: For each cutting plane, a non-rigid registration algorithm is applied to the first images of the cutting plane to register said first images of the subset of first images with each other; for each cutting plane, a non-rigid registration algorithm is applied to the first images of the cutting plane to register said first images of the subset of first images with each other.

[0020] According to the invention, the first duration is determined so that the first images acquired are images having a first contrast allowing to display images in black blood and the second duration is determined so that the second images acquired are images having a second contrast allowing to display images in white blood.

[0021] According to one embodiment, the calibration step is followed by the following steps: Averaging of the first acquired images registered to produce a first merged image for each cutting plane; Averaging of the second acquired images registered to produce a first merged image for each cutting plane.

[0022] One advantage is to increase the signal-to-noise ratio, also known as SNR.

[0023] According to one embodiment, the first duration is the inversion time in inversion-recovery.

[0024] Advantageously, the first duration is defined so that the longitudinal magnetization of the blood and that of the healthy myocardium cancel each other out at the same instant during the acquisition of the first 2D image.

[0025] According to one embodiment, the determination of the first duration is automatically calculated from a computer-implemented process for calculating an optimal inversion time obtained from the processing of at least one image acquired from an MRI presequence.

[0026] This calculation is performed, for example, using the computer-implemented method for determining the optimal inversion time described in patent application FR2203766. The MRI presequence comprises the acquisition of a plurality of MRI images via acquisition sequences with distinct inversion times. The inversion time is calculated from these images.

[0027] In one embodiment, the process includes a step of colorizing the pixels of each first merged image having a luminance above a given threshold, the colorization step being performed prior to the first combination. In the example case, the colored images correspond to the black blood images.

[0028] One advantage is that it allows for the localization and identification of a scar near the myocardium and of blood volume(s) in a safe, i.e., unambiguous manner.

[0029] According to one embodiment, the first combination is a superimposition of the first image onto the second merged image. That is to say, a superimposition of the black-colored blood image onto the white-colored blood image.

[0030] In a particular implementation, the first image is the first merged image.

[0031] According to one embodiment, each new generation of a plurality of first images and second images of the same section plane is performed after each repetition of the first phase to generate a plurality of first images and a plurality of second images for a plurality of sections of the heart.

[0032] According to one embodiment, each generation of a new first image and a new second image in a new cross-sectional plane of the heart is carried out after all the repetitions of the first phase to generate a plurality of first images and second images of a previous cross-sectional plane of the heart.

[0033] One advantage is to acquire all the images of the same section plane in a short time interval to avoid image shift drifts in the same section plane when they are interlaced after the acquisition of different images of other section planes.

[0034] According to one embodiment, the process comprises: ■ a display of a set of first merged images for each section plane; ■ a display of a set of second merged images for each section plane; ■ a display of a set of third 2D images for each section plane; ■ each first image of a given section plane being displayed in the immediate vicinity of the second and third images of the same given section plane.

[0035] Advantageously, the first image displayed and the second image displayed for a cutting plane are respectively the first merged image and the second merged image.

[0036] According to one embodiment, the first image merged for each cutting plane corresponds to an image of a cross-section of the heart in black blood, the second image merged for each cutting plane corresponds to an image of a cross-section of the heart in white blood, the third image for each cutting plane corresponds to an image of a cross-section of the organ on which a scar, if any, is displayed in color.

[0037] According to one embodiment, the first and second magnetization preparation is a T1 weighting in T1rho, said process being carried out during the late enhancement after Gadolinium injection.

[0038] According to one embodiment, the first magnetization preparation is a T1rho type preparation, said T1rho preparation being defined by the sequence of a magnet configuration defined as follows: 90x-SLy-180y-SL-y-SLy-180-y-SL-y-90-x.

[0039] According to another example of T1rho type preparation, another magnetization configuration could be, for example, 90x-Sly-90-x.

[0040] According to one embodiment, the first magnetization preparation is a T2prep type preparation or an MTC type preparation.

[0041] According to one embodiment, the generation of a second magnetization preparation comprising a set of pulses in an inter-beat phase is followed by a filtering step of the acquired images corresponding to fatty areas of the imaged organ.

[0042] According to one embodiment, the first and second images are acquired successively in a synchronized manner with the electrical activity of the heart between two QRS complexes for several minutes during free breathing in a single examination.

[0043] According to one embodiment, the first images and second images are acquired on a set of 8 to 20 cross-sectional planes of the organ, the phase of reiterating the acquisitions of a plurality of first images and second images in the same cross-sectional plane allowing to acquire between 2 and 10 images of first images and between 2 and 10 images of second images per cross-sectional plane.

[0044] According to another aspect, the invention relates to a magnetic resonance imaging system comprising a magnetic field generator and a radio frequency device, an electrocardiograph and a processing device, a display for displaying the generated images, said system being configured to implement the method of the invention.

[0045] One advantage is that it allows the generation of images in which a radiologist can reliably locate and detect scarring of the myocardial tissues.

[0046] One advantage is the ability to display an area of ​​interest such as a scar in color.

[0047] Another advantage is that it allows for the rapid acquisition of images in free breathing, thus avoiding the need for the patient to perform the imaging in apnea.

[0048] Furthermore, the method according to the invention is simple, fast, and computationally inexpensive. It can be easily integrated into an existing MRI device without modification.

[0049] One advantage is that for each given slice of the heart, a plurality of images are collected, which makes it possible to reduce noise, or to increase the signal-to-noise ratio and reduce residual artifacts.

[0050] One advantage of using a non-rigid algorithm is that it mitigates motion artifacts related to variations in respiration. Implementing such an algorithm increases the signal-to-noise ratio.

[0051] Finally, 2D acquisition allows for faster and more robust acquisition than 3D acquisition. Brief description of the figures

[0052] Other features and advantages of the invention will become apparent from the detailed description that follows, with reference to the attached figures, which illustrate: Fig. 1 : an example of the realization of a system of the invention comprising an MRI device, an electrocardiograph, a calculation to perform the processing operations of the acquired images and a human-machine interface to view the images; Fig. 2 : a schematic representation of a portion of the sequence generated by an MRI device over two heartbeats, Fig. 3 : a schematic representation of an MRI acquisition phase in black blood and white blood performed on a plurality of heartbeats, Fig. 4: a schematic 3D representation of a heart illustrating the different cross-sectional planes and the images acquired in each cross-sectional plane, Fig. 5 : a flowchart of the steps of the process according to an embodiment of the invention. Invention description MRI system

[0053] There figure 4 schematically represents an imaging system S 1 comprising a magnetic resonance MRI 1 device according to the invention including a set of excitation and measurement equipment, a computer K 1 for processing the acquired images and an HMI 1 display for displaying the generated images.

[0054] As is known, the excitation and measurement equipment set includes an MRI imaging device comprising a static magnetic field generator GEN_B 0 including a main polarizing magnet, a gradient generator GEN_GRAD, and a radio frequency device DISPO_RF including a radio frequency signal generator and a radio frequency antenna.

[0055] The GEN_B 0 static magnetic field generator includes a magnet intended to generate a substantially uniform polarized static magnetic field in a polarization zone (usually a tunnel) intended to understand the area to be imaged of the patient, for example the heart.

[0056] The GEN_GRAD gradient generator comprises three gradient coils or solenoids arranged and configured to vary the magnetic field strength in the polarization zone along orthogonal axes, conventionally denoted x, y, and z, fixed relative to the polarization zone. The choice of currents circulating in these coils allows selection, from several possibilities, of a thickness and a slice plane in which the magnetization of the area to be imaged of the patient within the polarization zone will be measured.

[0057] The GEN_RF radiofrequency device comprises coils or solenoids and is capable of generating MRI acquisition sequences including sequences for magnetizing the area to be imaged and sequences for reading signals from the area to be imaged. Each acquisition sequence includes at least one radiofrequency pulse with predetermined and controllable frequency, shape, duration, phase, and amplitude.

[0058] The preparatory module, also called the preparation or preparation phase, is configured to excite, that is, to modify the direction of magnetization of the tissues in the area to be imaged, i.e., to modulate their magnetization. It is denoted TAprep, TBprep, or T1rho in the rest of the description.

[0059] The readout module is configured to measure the magnetization of the area to be imaged resulting from the preparation sequence. The readout module is also called the acquisition module, readout phase, or readout window.

[0060] In other words, the images are acquired by magnetization measurements of the area to be imaged.

[0061] The set advantageously includes an ELC 1 electrocardiograph intended to acquire an ECG 1 electrocardiogram of the patient.

[0062] The imaging system includes a K1 processing unit configured to perform calculations, including image registration and averaging. The processing unit can also optionally colorize the images.

[0063] The S1 imaging system further includes a display to show two-dimensional images in greyscale and / or colour of the area to be imaged from signals measured by the RF magnetic field generator and processed by the processing device.

[0064] In one embodiment, the reconstructed image is a grayscale and color raster image. It comprises a set of pixels, each characterized by an intensity I that can take on a set of M values ​​(M being a finite integer greater than 1) corresponding to M grayscale levels ranging from 0 to M-1. For example, this value can take 256 values ​​between 0 and 255. This value can be higher in other embodiments. In one embodiment, the images can be partially colored for a selection of pixels having a luminance above a given threshold.

[0065] Representations of the images generated by the processing device K 1 are intended to be displayed on a screen of the human-machine interface HMI1 of the imaging system S 1. Acquisition sequence

[0066] There figure 2This represents an example of a portion of a black-blood / white-blood MRI acquisition sequence SEQ 1 of a myocardial image, along with the ECG 1 electrical activity shown over two consecutive heartbeats. An ECG 1 acquired by the ECL 1 electrocardiograph during two heartbeats and the SEQ 1 acquisition sequence applied to the area to be imaged are superimposed to better represent the temporally related characteristics of each of these curves. The ECG acquisition is a step denoted ACQ 0. The lower part of the figure 2 represents the variation of the longitudinal magnetization Mz of the tissues in the area to be imaged as a function of time t as well as two images IM 1 and IM 2 reconstructed from signals measured during the sequence.

[0067] Preferably, the invention allows for the synchronization of MRI excitation and acquisition sequences with the electrocardiogram (ECG). This synchronization makes it possible, on the one hand, to perform measurements at times when the heart is most stable and, on the other hand, to take measurements at equivalent times between two identical positions of the heart in order to compare images of the same slice. One way to synchronize these excitations and acquisitions is to trigger the measurement device on the QRS complex.

[0068] In the following description, the first image IM 1 will refer to an image acquired in black blood and the second image IM 2 will refer to an image acquired in white blood. For example, throughout the sequence, starting with the first beat during which image acquisitions are performed, odd-numbered beats are used to acquire black blood images and even-numbered beats are used to acquire white blood images. This example will be described in more detail later. Other configurations are possible for carrying out the method of the invention, for example, associating odd-numbered beats with white blood image acquisition and even-numbered beats with black blood acquisition.

[0069] The invention also applies in cases where white blood images are acquired before black blood images. The invention also applies in cases where the black and white blood images are not acquired immediately over two consecutive beats, but, for example, separated by one or more beats without measurement. However, one advantage of acquiring these images over two consecutive beats is to minimize acquisition time and thus minimize the effects of respiration and therefore image artifacts over a two-beat timescale.

[0070] For MRI acquisition in black and white blood, a gadolinium-based contrast agent is advantageously injected intravenously into the patient 10 to 15 minutes before the acquisition sequences are applied to obtain images with maximum contrast between scars and healthy tissues and blood. In the heart, contrast is rapidly cleared from healthy myocardium, which is poor in interstitial tissue, but accumulates for a prolonged period in myocardial scars. The effect of gadolinium is to shorten the T1 relaxation time of the tissues where it accumulates. The relaxation of the magnetization in scars following a magnetization pulse is therefore faster than that in blood and healthy myocardium.

[0071] Initially, we aim to generate images exhibiting a contrast in black blood. In such an image, the intensity of the pixels corresponding to blood and healthy muscle is zero (black pixels) or virtually zero because it is acquired when the longitudinal magnetization of blood is zero.

[0072] In order to generate such an IM 1 image, the DISPO_RF device generates an inversion-recovery acquisition sequence.

[0073] This acquisition sequence includes a 180° reversal pulse, denoted RFI on the figure 2 , which flips the longitudinal magnetization of the tissues in the imaged area in the opposite direction, that is, it reverses the longitudinal magnetization. On the figure 2We observe that the magnetization of the area to be imaged changes from Mz to -Mz under the effect of the reversal pulse Rfi. Due to longitudinal relaxation, the longitudinal magnetization of the different tissues present in the area to be imaged increases, returning to its initial value, passing through zero. Naturally, the rates of magnetization reversal of the different tissues are different.

[0074] As is well known, the preparation sequence then includes a preparatory module, for example an adiabatic preparation in T1rho, noted T1rho or TAprep on the figure 2 of TSL duration, designated in Anglo-Saxon terminology as "Spin Lock Time".

[0075] Other preparations, such as a T2prep preparation or a MT magnetization transfer preparation, can be used.

[0076] It is noted in this application that the preparation can sometimes be defined as the T1rho preparation or sometimes as the T1rho preparation preceded by the 180° RFI inversion sequence. Since the second image, IM2, is acquired without a 180° inversion pulse, but only with a T1rho preparation, the preparation in this case is limited to the T1rho preparation.

[0077] The preparation sequence (Rfi(180°), T1rho) is configured so that the longitudinal magnetization of the blood and that of the healthy myocardium cancel each other out at the same instant TI, as shown by the position of the NP point on the A(blood) and A(Musc) curves. The inversion time TI corresponds to a time measurement between the emission of the Rf signal at 180° and the ACQ 1 acquisition module. The instant TI corresponds to the duration denoted D 1 in the application. Indeed, the invention also applies to cases in which the magnetization measurement would not necessarily be calculated at the inversion time, but possibly at another time depending on the desired result.

[0078] We note interchangeably an acquisition window or a reading window for each beat, they are noted ACQ 1 for reading windows relating to black blood acquisition and ACQ 2 for reading windows relating to white blood acquisition.

[0079] It is noted that on the figure 2 At this same instant TI, the longitudinal magnetization of the scars A(Cica) is clearly greater than zero; indeed, the point PP is above zero. By acquiring the signals from the area to be imaged at this instant TI, we obtain an image exhibiting very high contrast between the pixels corresponding to blood and healthy myocardium, which are black, and the scars, which are generally white.

[0080] The inversion-recovery sequence then includes an ACQ 1 readout sequence comprising a 90° pulse and a readout gradient to read the transverse magnetization of the area to be imaged.

[0081] The inversion time (TI) is the time between the 180° pulse and the ACQ1 acquisition window. The ACQ1 acquisition window time marker is imposed upstream in the ECG, generally during diastole. It is chosen as the instant when the muscle and blood signals cancel each other out, that is, when the longitudinal magnetizations of the blood and myocardium cancel each other out, in order to generate the image with the best contrast. To obtain the best contrast between myocardial scars and healthy blood and myocardium, the goal is to initiate the inversion Rf signal at an earlier time to induce signal cancellation at a duration TI.

[0082] The sequence is followed by a white blood cell acquisition phase performed at the beat following the black blood cell acquisition. During this phase, a second IM2 image is acquired. The preparation phase is noted as TBprep on the figure 2According to one embodiment, the TBprep preparation could be different from the TAprep preparation used to image the heart area in dark blood; however, it is preferable that the TAprep dark blood and TBprep white blood preparations be identical. The invention therefore relates to other preparations; however, in the following description, the white blood acquisition phase is described using the same white blood preparation as the black blood preparation, for example, a T1rho preparation.

[0083] The time interval D2 separating the temporal marker from the acquisition window AQ2, which is fixed for each beat, and the temporal marker corresponding to the TBprep preparation in white blood can be adjusted to optimize the white blood contrast in order to best image the contours of the muscle against the blood.

[0084] The magnetization gradient measurement is configured for an ACQ 2 acquisition of the white blood image at the same inter-beat marker as the black blood image. Thus, on the figure 2 , the ACQ 2 reading phase is located at a duration of the previous QRS complex identical to the duration between the QRS of the beat preceding the ACQ 1 reading phase.

[0085] When reading the magnetizations at this same instant, it is possible to obtain an image in which the pixels of the blood areas are represented in white, i.e., with high luminance, and in which the pixels of the myocardial tissue are slightly less luminous than those of the blood. Thus, it is possible to obtain the 2D topology of the slice, allowing visualization of the blood and myocardial areas relative to each other.

[0086] It is understood that by combining the IM 1 and IM 2 images it will be possible to locate and detect the scar present if necessary.

[0087] To this end, the invention makes it possible to carry out successive measurements of the sequence of these two images IM 1 in black blood and IM 2 in white blood on a plurality of sections of the heart to study the entirety of the heart and taking into account several images per section in order to make the analysis of the images more robust.

[0088] There figure 3 represents an example of a sequence of these phases, denoted Phi, with i = 1 to N, over a plurality of consecutive beat pairs. The phases PHi each comprise, over two beats: First beat: a GEN 1 generation of an RF inversion signal, a generated preparation phase noted GEN 2 and a readout phase noted ACQ 1; Second beat: the generation of a GEN 3 preparation phase and a readout phase noted ACQ 2.

[0089] There figure 4 represents a 3D view of a core comprising a plurality of K cross-sectional planes denoted PC k, with k = 1 to K. figure 4 allows the representation of a plurality of section planes, each comprising a plurality of images acquired using the method of the invention. The images of section plane PC k are represented; the others are not. figure 4 . The images are respectively noted IM 1 k< (i) for the ith image acquired from the PC k slice, with i = 1 to N.

[0090] In one embodiment, an initial acquisition of the heart image allows for the evaluation of the number of slice planes and therefore the number of acquisition repetitions. If the heart is considered small, between 7 and 9 slice planes can be defined; if the heart is larger, between 14 and 20 slices can be defined. If the heart is of average size, the number of slices can be defined between 10 and 13 slices.

[0091] This estimation can be carried out at the beginning of the examination, for example during a presequence.

[0092] In another example, the heart size is automatically determined during the initial acquisitions of the invention's method, during the first few beats. This size can then be automatically used to dynamically calculate and acquire images based on a given number of slices.

[0093] Based on an example of a complete 3-minute sequence with a pulse rate of 60 beats per minute, considering a configuration of 15 slice planes for the organ, there would be a total of 3 x 60 = 180 images, including 90 black blood images and 90 white blood images. Such a configuration allowed the acquisition of 6 images {IM 1 k< (1), IM 1 k< (2), IM 1 k< (3), IM 1 k< (4), IM 1 k< (5), IM 1 k(6)} per slice plane in black blood and 6 images acquired {IM 2 k< (1), IM 2 k< (2), IM 2 k< (3), IM 2 k< (4), IM 2 k< (5), IM 2 k< (6)} per slice plane in black blood.

[0094] One advantage of the invention's method is that it can be performed during free breathing. To this end, it is necessary that the images acquired in the same slice be processed so that they are not altered by the effects of free breathing. The respiratory rate is inherently different from the heart rate. Even though correlations exist between these two rhythms, it is possible for respiration to accelerate while the heart rate remains stable, or vice versa.

[0095] The problem is that during the second ACQ 2 reading, breathing may have significantly altered the position of the heart, which means that images acquired at another beat of the same slice may be offset from an image acquired previously in the same slice.

[0096] In one embodiment, the images from each slice acquired in black blood are registered using a non-rigid image registration algorithm. The images acquired in white blood for each slice are registered identically to the method chosen for black blood. It is possible to choose a different algorithm for processing the black and white blood images, but it is preferable to choose the same algorithm for ease of implementation. The registration significantly improves the image that will be merged from multiple images acquired in the same slice, because the contrast is better and respiration artifacts have been compensated for thanks to the non-rigid algorithm.

[0097] As a first example, a non-rigid algorithm is one based on the mutual information method between images, which relies on statistical relationships. The function to be optimized can be implemented using a statistical similarity criterion. One advantage of this method is that the matching of homologous attributes of images in the same slice is independent of their geometric position.

[0098] In a second example, a non-rigid algorithm based on a transformation model is implemented. The transformation model allows the determination of functions that minimize the difference between two images. This difference can be represented as a geometric error to be minimized. Various approaches can be used, such as those based on extracting geometric primitives or shape descriptors like salient points, shape singularities, or contours from each image. A parametric or non-parametric approach can be employed.

[0099] Depending on an example of optimization of a transformation model or a similarity criterion, the least squares method can be used.

[0100] Other optimization methods, such as gradient descent, can be implemented. However, this latter method is more specifically applied to image intensities and is not optimal within the scope of the invention, since the aim is to optimize the sharpness and contrast of the merged image. Nevertheless, the invention includes this embodiment.

[0101] Registration can be performed by choosing a reference image and determining a transformation function for the other images of the same cross-section with respect to this image. Each image is then registered by optimizing a transformation to obtain the reference image according to a geometric criterion, starting from the image in question.

[0102] When a plurality of images are aligned with each other in a cutting plane, it is possible to perform operations aimed at merging these images in order to produce a single merged image per cutting plane.

[0103] According to one embodiment, an averaging step of the images of the same slice is carried out in order to reduce noise and increase the signal-to-noise ratio.

[0104] Averaging has the advantage of preserving image detail, as it increases the signal-to-noise ratio (SNR). This technique smooths noise to reduce residual image artifacts. Furthermore, averaging improves the bit depth of the digital image beyond what is possible with a single image.

[0105] One advantage of averaging images from the same slice is that it reduces the maximum deviation. The noise amplitude decreases as the square root of the number of images used; that is, with only 4 images, the noise amplitude can be reduced by a factor of 2. For example, in a 2-minute free-breathing acquisition, it is possible to collect 4 to 5 images per slice, resulting in good noise reduction performance.

[0106] Each set of images acquired on the one hand in black blood and on the other hand in white blood are respectively merged so as to produce a single black blood image IM 1F k< and a single white blood image IM 2F k< per section plane PC k.

[0107] In other words, a single fused black blood image IM 1F k< and a fused white blood image IM 2F k< are produced per section plane PC k as seen in figure 5 .

[0108] According to one embodiment, the method of the invention then comprises a step of colorizing the brightest pixels of the black blood image. The brightest pixels correspond in particular to the scar, given that the signals from the muscle and the blood cancel each other out and are represented by low-intensity pixels, i.e., black. Colorizing the scar makes it possible to identify and represent its outline.

[0109] One advantage of colorization is that it allows for clear identification of the scar and the use of this representation in combination with the white blood image, which depicts the muscle contours. This combination thus allows for the identification and localization of the scar, particularly in relation to the muscle and blood near the scar.

[0110] The COMB1 combination of black blood images IM 1F k< and white blood images IM 2F k< can be performed for each section plane PC k. Thus, the same combination operation allows for a plurality of combined black blood - white blood images to produce a single image IM 3 k< per section plane PC k.

[0111] The COMB1 image combination can be achieved in several ways. In one example, the images are superimposed. The superimposition can involve adjusting the opacity of the black blood image to properly visualize the contours of the white blood image. The opacity of the black blood images can be set to a value between 40% and 90%, for example, 75%. In another example, scar pixels with a luminance above a given threshold are extracted and integrated into the white blood image.

[0112] Other image combination methods can be used to produce a single image from the black blood image and the white blood image by cross-section.

[0113] In other words, for each PC k section plane, a unique combined image IM 3 k< is produced from the fused black blood image IM 1F k< and the fused white blood image IM 2F k< .

[0114] One advantage of the method of the invention is that it generates multiple reconstructed images for each slice plane from a single black blood image and a single white blood image. Depending on the size of the heart, between 8 and 20 slice planes can be selected. The method of the invention allows a user, particularly a radiologist, to view a reconstructed image for each slice plane. Each image makes it possible to display the presence or absence of a scar by precisely locating it in relation to the anatomy of the heart, the myocardium, and the blood volumes surrounding the imaged area.

Claims

1. Method for reconstructing an image of a patient's heart from a magnetic resonance imaging device (MRI1) generating a magnetic field (B0) comprising: - A first phase (PH1) of image generation including: ▪ Acquisition (ACQ0) of the electrical activity (ECG1) of the patient's heart; ▪ Generation (GEN1) of a first 180° inversion radio frequency signal (Rfi) to inverse a longitudinal magnetization of tissues of an imaged area, said first inversion radio frequency signal (Rfi) being generated between two QRS complexes of the acquired electrical activity (ECG1), so-called first inter-beat phase (Tc1); ▪ Generation (GEN2) of a first magnetization preparation (TAprep) including a set of pulses ({pi}) following the generation of the first inversion radio frequency signal (Rfi) in the same first inter-beat phase (Tc1); ▪ Acquisition (ACQ1) of a first 2D image (IM1) by a magnetization measurement of the imaged area, said acquisition (ACQ1) being performed after a first predefined duration (D1) following the generation of the first magnetization preparation (Tprep), said acquisition (ACQ1) being synchronized with the heart rhythm (ECG1) at a first time marker (M1) of the same inter-beat phase (Tc1); ▪ Generation (GEN3) of a second magnetization preparation including a set of pulses ({pi}) in a second inter-beat phase (Tc2) following the first inter-beat phase (Tc1); ▪ Acquisition (ACQ2) of a second 2D image (IM2) by a magnetization measurement of the imaged area, said acquisition (ACQ2) being performed after a second predefined duration (D2) following the generation of the second magnetization preparation (TBprep), said acquisition (ACQ2) being synchronized with the acquisition of a heart rhythm (ECG1) at the same first time marker (M1) of the second inter-beat phase (Tc2) as the first marker (M1) of the first inter-beat phase (Tc1); - Repetition of the first phase (PH1), to generate a subset of first images and a subset of second images of each section plane (PCk) of a plurality of section planes of the heart, in inter-beat phases (Tcik) following the first and second inter-beat phases (Tc1, Tc2); - Application of a non-rigid registration algorithm (ALG1) to each subset of first images (IM1k(i)) of one of the section planes (PCk) to register said first images (IM1k(i)) of the subset of first images together; - Application of a non-rigid registration algorithm (ALG1) to each subset of second images (IM2k(i)) of one of the section planes (PCk) to register said second images (IM2k(i)) of the subset of second images together; - Merging, for each of the section planes, on the one hand the first images registered together to produce a first merged image (IM1Fk) and on the other hand the second images registered together to produce a second merged image (IM2Fk); - Generation, for said section planes, of a set of third 2D images, each third image (IM3k) corresponding to a first combination (COMB1) made between the first merged image and the second merged image produced for one of the section planes, the first duration (D1) being determined such that the first acquired images are images having a first contrast making it possible to display black blood images and the second duration (D2) being determined such that the second acquired images are images having a second contrast making it possible to display bright blood images.

2. Method according to claim 1, wherein the registration step is followed by the steps of: - Averaging the first registered images to produce a first merged image (IM1Fk) for each section plane (PCk); - Averaging the second registered images to produce a first merged image (IM2Fk) for each section plane (PCk).

3. Method according to any one of claims 1 to 2, wherein the first duration (D1) is defined in such a way that the longitudinal magnetization of blood and that of healthy myocardium are nulled at the same instant TI of the acquisition (ACQ1) of the first 2D image.

4. Method according to any one of claims 1 to 3, wherein the determination of the first duration (D1) is automatically calculated from a computer-implemented method of an optimal inversion time performed from a processing of at least one image acquired from an MRI pre-sequence.

5. Method according to any one of claims 1 to 4, comprising a step of coloring the pixels of each first merged image (IM1Fk) having a luminance greater than a given threshold, the coloring step being performed prior to the first combination (COMB1).

6. Method according to any one of claims 1 to 5, wherein the first combination (COMB1) is an overlay of the first image (IM1Fk) on the second merged image (IM2Fk).

7. Method according to any one of claims 1 to 6, wherein: ▪ each generation of a new first image and a new second image in a new heart section plane is performed after: ▪ all the repetitions of the first phase (PH1) to generate a plurality of first images ({IM1k(i)}i[1-N]) and second images ({IM2k(i)}i[1-N]) of a preceding section plane (PCk) of the heart.

8. Method according to any of claims 1 to 7, comprising: ▪ a display of a set of first merged images (IM1Fk) for each section plane (PCk); ▪ a display of a set of second merged images (IM2Fk) for each section plane (PCk); ▪ a display of a set of third 2D images for each section plane (PCk); ▪ each first image (IM1Fk) of a given section plane (PCk) being displayed in the immediate vicinity of the second image and the third image (IM3Fk) of the same given section plane (PCk).

9. Method according to any one of claims 1 to 8, wherein the first merged image (IM1Fk)for each section plane (PCk) corresponds to a black blood image of a section of the heart, the second merged image (IM2Fk) for each section plane (PCk) corresponds to a bright blood image of a section of the heart, the third image (IM3Fk) for each section plane (PCk) corresponding to an image of a section of the organ on which a scar, if present, is displayed in color.

10. Method for reconstructing an image according to any one of claims 1 to 9, wherein the first and second magnetization preparation (TAprep, TBprep) are T1rho weightings.

11. Method for reconstructing an image according to any one of claims 1 to 10, wherein the first magnetization preparation and / or the second magnetization preparation (TAprep, TBprep) is a preparation of the T2prep type or a magnetization transfer preparation (MT).

12. Method for reconstructing an image according to any one of claims 1 to 11, wherein the generation (GEN3) of a second magnetization preparation (TBprep) including a set of pulses ({pi}) in an inter-beat phase (TcN) is followed by a step of filtering the acquired images corresponding to fatty areas of the imaged organ.

13. Method for reconstructing an image according to any one of claims 1 to 12, wherein the first images (IM1) and the second images (IM2) are acquired successively in a manner synchronized with the electrical activity of the heart between two QRS complexes over a plurality of minutes in free breathing during a single examination.

14. Method for reconstructing an image according to any one of claims 1 to 13, wherein the first images and the second images are acquired on a set of 8 to 20 section planes of the organ, the phase of reiteration of the acquisitions of a plurality of first images ({M1k(i)}i[1-N]) and second images ({IM2k(i)}i[1-N]) in the same section plane (PCk) making it possible to acquire between 2 and 10 images of first images ({IM1k(i)}i[1-N]) and between 2 and 10 images of second images ({IM2k(i)}i[1-N]) per section plane.

15. Magnetic resonance imaging system (S1) comprising a magnetic field generator (GEN_B0, GEN_GRAD) and a radio frequency device (DISPO_RF), an electrocardiograph (ELC1) and a processing device (K1), a display (IHM1) for displaying the generated images, said system being configured to implement the method of any one of claims 1 to 14.