阅读说明：本技术 磁性粒子成像系统的校准方法 (Calibration method for magnetic particle imaging system ) 是由 肯·巴里斯·托 塞尔哈特·伊尔贝伊 阿尔珀·贡戈 侯赛因·埃姆雷·古文 图勒加·丘库尔 埃 于 2018-05-11 设计创作，主要内容包括：本发明公开了一种校准由磁场发生器和测量装置组成的磁性粒子成像系统的方法,其提出了一种编码校准场景,该编码校准场景包含分布在其体积内的多个纳米粒子样品,该编码校准场景大于该场景所在的视场。在视场上,向一个或多个方向上线性移动和/或在一个或多个轴上旋转该场景,并且进一步地,本发明也公开了一种用于移动编码校准场景的机械系统。(The invention discloses a method for calibrating a magnetic particle imaging system consisting of a magnetic field generator and a measuring device, and provides a coding calibration scene, wherein the coding calibration scene comprises a plurality of nanoparticle samples distributed in the volume of the coding calibration scene, and the coding calibration scene is larger than the field of view in which the scene is positioned. The scene is moved linearly in one or more directions and/or rotated in one or more axes over the field of view, and further, a mechanical system for moving the encoded calibration scene is also disclosed.)

1. Method of calibrating a magnetic particle imaging system for performing magnetic particle imaging of a field of view, the method comprising the steps of:

by linearly moving the calibration scene in one or more directions and/or rotating the calibration scene about one or more axes by a mechanical system,

scanning a field-free region in the field of view and acquiring calibration measurement data at a plurality of locations of the calibration scene,

in the data acquisition process, a system matrix is reconstructed by using the measured data and the position information of the calibration scene and adopting a compression sensing method.

2. Method of calibrating a magnetic particle imaging system according to claim 1, comprising the steps of: reconstructing the system matrix using an optimization problem represented by the following inequality:

where P is the nanoparticle density distribution within the field of view at each measurement location and D is a matrix related to the sparse transformation of A, the system matrix; a. the_pIs a measurement matrix that is transformed into fourier space for each measurement location of the encoded calibration scene; epsilon_pRepresenting a constant related to the error caused by the system noise.

3. Method of calibrating a magnetic particle imaging system according to claim 1, wherein the calibration scene is continuously moved or rotated.

4. The method of calibrating a magnetic particle imaging system according to claim 1, wherein the calibration scenario comprises a plurality of nanoparticle samples.

5. Method of calibrating a magnetic particle imaging system according to claim 1, wherein the nanoparticles in the calibration scenario are randomly or pseudo-randomly distributed.

6. Calibration method of a magnetic particle imaging system as claimed in claim 1 characterized in that the nanoparticles in the calibration scenario are distributed continuously to be filled and emptied from both ends.

7. The method of calibrating a magnetic particle imaging system of claim 1, wherein the position of the calibration scene is continuously monitored using a tracking device to measure the position of the calibration scene during data acquisition.

8. Calibration device for a magnetic particle imaging system, comprising:

a calibration scene having a distributed sample of nanoparticles in an internal volume thereof, the calibration scene being larger than a field of view of the magnetic particle imaging system,

a mechanical system that linearly moves the calibration scene in one or more directions and/or rotates the calibration scene about one or more axes.

9. Calibration apparatus for a magnetic particle imaging system according to claim 8, wherein the external geometry of the calibration scene is a rectangular prism, a cylinder, a sphere or any arbitrary shape.

10. Calibration apparatus for a magnetic particle imaging system according to claim 8, wherein one or more reflectors are attached to the calibration scene to track its movement.

11. Calibration apparatus for a magnetic particle imaging system according to claim 8, wherein said calibration scene comprises a single tube taking an arbitrary path traversing said calibration scene for filling or emptying nanoparticles from said single tube with or from nanoparticles.

12. Calibration apparatus for a magnetic particle imaging system according to claim 8, wherein the calibration scenario comprises a plurality of tubes traversing the calibration scenario to fill or empty the plurality of tubes with or from nanoparticles.

13. Calibration apparatus for a magnetic particle imaging system according to claim 8, wherein the calibration scenario is a hollow structure with one or more openings for filling or emptying nanoparticles from the hollow structure with nanoparticles.

14. Calibration apparatus for a magnetic particle imaging system according to claim 8, wherein its position is tracked by tracking means.

15. Calibration device for a magnetic particle imaging system according to claim 8, wherein the mechanical system comprises a control unit communicating with the magnetic particle imaging system for performing a calibration method.

Technical Field

The present disclosure relates to a method of calibrating a magnetic particle imaging system by proposing an encoded calibration scenario comprising a plurality of nanoparticle samples distributed within its volume, the encoded calibration scenario being larger than a field of view of the scenario that moves linearly in one or more directions and/or rotates in one or more axes, and further to a mechanical system for moving the encoded calibration scenario.

Prior Art

Magnetic nanoparticles can be used for various purposes in medicine, such as angiography, stem cell tracking, cancer cell and targeted drug imaging. Magnetic nanoparticles can be non-invasively imaged using a Magnetic Particle Imaging (MPI) method. In the magnetic particle imaging method, two different methods are used as criteria for image reconstruction.

The first method is the system calibration method described in US8355771B2, in which a small sample of nanoparticles is mechanically scanned at the required system resolution step in the field of view to obtain the calibration data [1] of the system. An image is generated using this calibration data (also referred to as the system matrix). In standard system calibration methods, the calibration measurements will last a long time, since the sample nanoparticles must be mechanically scanned and measured at each grid point within the field of view. The mechanical scan time from one point to another and the time to acquire measurement data is approximately 1.3 seconds [2 ]. For a small field of view with 30x 30x 30 grid points, the calibration time will last 9.75 hours. In clinical practice, a large imaging volume calibration may last for months. The system needs to be calibrated frequently, since it is known that nanoparticles differ in characteristics from batch to batch, and are also affected by the imaging sequence. Thus, for large field of view systems, standard system calibration methods cannot be practically employed. In addition, since the nanoparticles to be scanned must be less than or equal to the voxel size of the image, the number of nanoparticles in the sample scanned is limited and the signal-to-noise ratio is small. One way to increase the signal-to-noise ratio is to collect multiple data at the same location and average them. Thus, the mechanical movement cannot be done continuously, the scanner stops at each grid point, and moves to the next point after sufficient measurements are made to reach the desired signal-to-noise level. This limits the speed of the calibration measurement.

Recently, a calibration method has been proposed in US patent application US20150221103A, in which a sample of nanoparticles is scanned at random positions, the number of which is much smaller than the total number of voxels in the field of view. This is possible according to compressed sensing theory [3] because the system matrix is sparse in some transform domains (discrete fourier, cosine or chebyshev). It has been shown that this method can reduce the number of scan points by 80-90%. In addition to making measurements for all voxels (N) in the field of view, system calibration can be accomplished by making fewer measurements at random M (< N) voxel locations using compressive sensing techniques. Since it is impossible to calculate analytically how small the M value is, the M/N ratio should be selected according to the image quality. Experimental images were obtained in the above references. Although an image quality of 0.1M/N is acceptable, the image quality is significantly degraded for lower M/N ratios. The calibration time can be reduced by a factor of 10 using this method, but still requires a very long calibration time due to the need to mechanically scan the sample, e.g. a measurement area of 200x200x 200 points will require more than 10 days to measure.

The second reconstruction method is the X-space method used in EP3143929a 1. In this method, there is no calibration step. The image is generated using a signal equation model for Magnetic Particle Imaging (MPI). By using the MPI signal equation, image reconstruction can be done in the time domain. In this method, deviations from the MPI ideal hardware are not taken into account and the resolution is lower than in the system calibration method.

A.von Gladdis et al [2] discloses an electronic calibration method for accelerating the calibration process. The nanoparticle sample was placed in a separate calibration unit that could generate a uniform magnetic field in any direction, simulating the magnetic field to which the nanoparticle sample was exposed in the MPI system. Although this method provides faster calibration than the standard method, it requires the use of a separate calibration unit. The magnetic field distribution of an MPI system must be measured separately in the field of view; and the measurement values of the calibration unit have to be related to the measurement values of the MPI system. As with standard system calibration measurements, the advantages of electronic calibration are limited because the magnetic field distribution measurement of MPI systems requires a mechanical scan of every voxel within the field of view.

Object of the Invention

In the present invention, a large calibration device for calibration of MPI systems is proposed, which is referred to as an encoding calibration scenario. The encoded calibration scenario includes nanoparticle samples at multiple locations. The encoded calibration scene moves linearly in one or more directions and/or rotates about one or more axes. During the above-described movement, calibration measurement data is acquired at certain positions of the encoded calibration scene. These measurement data are used to generate a system matrix using a compressive sensing method. The advantages that distinguish this method from other available methods are listed below:

according to the prior art, in US20150221103a1, a number M of voxels, randomly or pseudo-randomly selected from the prime number total (N), are scanned mechanically one by one in a single nanoparticle sample for MPI system calibration. In the present invention, a calibration device is proposed, which comprises a plurality of nanoparticle samples and is larger than the field of view of the imaging system at least in one direction. Thus, it receives a much higher level of signal than it receives from a single nanoparticle sample. This allows measurements to be made during continuous movement of the calibration scene, greatly speeding up the calibration process. Furthermore, as nanoparticle samples are measured at different locations simultaneously in one measurement, the information content per measurement increases. Thus, fewer measurements can be used to form the system calibration matrix. This provides an advantage for systems with a large field of view.

In the method proposed by A von Gladdis et al [2], the characterization of the nanoparticles was obtained with a separate calibration unit. Therefore, it is necessary to measure the magnetic field in the field of view of the MPI system. In the present invention, all factor effects (magnetic field inhomogeneity, nanoparticle response) are taken into account in a single calibration scan.

Brief description of the drawings

Fig. 1 shows a cross section of a bore of a magnetic particle imaging apparatus, a non-uniform primary magnetic field and a uniform secondary magnetic field with two zones, and a field of view.

Figure 2 shows the entire field of view, which is hypothetically divided into small voxels, and the calibration setup using a sample containing nanoparticles.

Fig. 3 illustrates an encoded calibration scenario in which a plurality of nanoparticle samples are randomly or pseudo-randomly distributed within their volume.

Fig. 4 shows a comparison of the standard compressive sensing method with the proposed method of the present invention using a simulation model for the same noise level. The proposed method exhibits better image quality with a smaller number of measurements (M).

Fig. 5 and 6 show the nanoparticle positions at 0 and 45 degrees angles, respectively, for a spherical calibration scenario.

Fig. 7 shows a spherical calibration scenario that rotates on one axis and slides on the other axis.

Fig. 8 shows a calibration stand and a rotation mechanism that rotates around different rotation axes.

Fig. 9 and 10 show a spherical calibration scenario and an external mechanism for linear and rotational movement of the spherical calibration scenario in top and side views, respectively.

Fig. 11 shows a calibration scenario comprising a nanoparticle chamber connected to each other by a thin channel and one or more points for filling or draining nanoparticles.

Fig. 12 shows a spherical scenario with a rod-like nanoparticle sample.

Fig. 13 shows a calibration scenario designed as a long rectangular prism linearly moved on a sliding belt.

Fig. 14 and 15 show a cylindrical calibration scene and external mechanisms for linear and rotational movement of the calibration scene in top and side views, respectively.

Fig. 16 shows a cylindrical calibration scenario with a cylindrical cavity for the nanoparticle sample.

Fig. 17 shows a cylindrical calibration scenario with tubules that are complex curves in three dimensions and have inputs and outputs.

Parts reference

1. MPI system

2. Primary magnetic field

3. First region of the primary magnetic field

4. Second region of the primary magnetic field

5. Secondary magnetic field

6. Field of view

7. Voxel

8. Magnetic nanoparticle samples

9. Mechanical scanner

10. Encoding calibration scenarios

11. Spherical calibration scenario

12. Center of rotation

13. Mechanism for translating and rotating calibration scene around one axis

14. Mechanism for translating and rotating calibration scene around two axes

15. Auxiliary mechanical system for translational and rotational calibration scenarios

16. Sliding rail

17. Rotating shaft

18. Thin channel

19. Opening of the container

20. Rod-like nanoparticle samples

21. Rectangular prism calibration scenario

22. Sliding belt

23. Optical reflector

24. Laser tracker

25. Cylindrical calibration scenario

26. Columnar cavity

27. Input for filling magnetic nanoparticles in a calibration scenario

28. Output for discharging magnetic nanoparticles in calibration scenarios

29. Thin tube

Detailed Description

In an MPI system (1) consisting of a magnetic field generator and a measuring device as shown in fig. 1, the distribution of magnetic nanoparticles is imaged using a non-uniform primary magnetic field (2) with two zones [4 ]. The first (3) of the two zones has a very low magnetic field strength and is called Field Free Region (FFR). The magnetic nanoparticles in the FFR can be magnetized in the direction of an external secondary magnetic field (5). In the second region (4), the magnetic field strength is high and the magnetic nanoparticles in this region are in a state of saturation. Thus, the response of the magnetic nanoparticles to the secondary magnetic field (5) is small. The secondary magnetic field (5) is applied as a time-varying magnetic field to the entire field of view (6). The time dependent magnetization of the magnetic nanoparticles in the FFR is measured by a receiving coil. The amplitude of the measured signal is proportional to the number of nanoparticles in the FFR. The FFR is scanned electronically or mechanically throughout the field of view (6) to obtain a distribution of nanoparticles in the field of view (6). Since magnetic nanoparticles have a non-linear magnetization curve, the signal received from the particles in the FFR contains harmonics of the emitted signal frequency. The characteristics of the received signal depend on the characteristics of the nanoparticle (size, shape, material, etc.) and the nanoparticle environment (viscosity, temperature) and the magnetic field characteristics of the imaging system. In MPI, the best image quality is achieved with an image reconstruction method based on a systematic calibration method that takes into account the influence of all these factors [5 ].

In a system-calibrated image reconstruction method, it is first assumed that the entire field of view (6) is divided into small voxels (7). A sample (8) filled with magnetic nanoparticles having a voxel (7) size is used to form a system matrix. To this end, a sample (8) containing nanoparticles is scanned to each voxel location by a mechanical scanner (9). A secondary magnetic field signal is applied and the nanoparticle signal received by the receiving coil is stored in a digital storage unit, e.g. a hard disk. In practice, measurement data is acquired multiple times at the same voxel point, and the signal-to-noise ratio is increased by averaging the measurement data. The signals measured from the individual voxels are converted to the frequency domain using fourier transforms, forming the columns of the system matrix (a). The entire system matrix is generated by taking measurements at all voxel positions. This process is called a calibration step.

For imaging, measurement data is acquired by scanning the FFR inside the object, and an image is reconstructed using the measurement data and the system matrix. For this purpose, the system of linear equations Ax ═ b is solved. In this system of equations, a is the system matrix, b is the measurement vector taken from the object, and x is the nanoparticle distribution inside the object. The main drawback of the system matrix calibration method is the long duration (about 1.3 seconds per voxel, multiplied by the number of voxels) [2 ]. In addition, since the sample size of the nanoparticles is very small, the signal level is low, and it is necessary to increase the signal-to-noise ratio by performing multiple measurements. This prevents continuous mechanical scanning, resulting in an extended measurement period.

The present invention proposes to solve the problems of the prior art using a coded calibration scenario (10). An encoded calibration scenario may be defined as a device comprising a plurality of nanoparticle samples distributed within its volume. The method has the following advantages: the signal level increases in proportion to the number of particles used in the calibration scan, and the conditions for compressive sensing problems also increase [6 ]. As a result, calibration may be performed using a fewer number of measurements using a compressive sensing algorithm (e.g., a greedy reconstruction algorithm, approximate messaging, optimization-based techniques, etc.) [3 ].

According to the compressive sensing theory, the correlation between calibration scenarios should be minimized. Thus, the nanoparticles may be randomly or pseudo-randomly distributed in each calibration scenario.

The method is realized as follows: the number M of calibration scenarios to be measured is predetermined. For this purpose, a simulation model of the imaging system can be used, or a plurality of calibration scenarios can be generated during the generated system test of the imaging system. New scenes are measured until the image quality is of a sufficient level from a practical point of view. Measurement data for M encoded calibration scenarios are collected and recorded. Once these measurements are completed, the system matrix a is reconstructed using the following optimization problem:

where P is the nanoparticle density matrix of the measured encoded calibration scene and D is the transform matrix associated with a sparse transform (e.g., discrete fourier transform, discrete chebyshev transform, discrete cosine transform, or any other transform that can represent a vector with fewer vectors relative to the original domain); a. the_pIs a measurement matrix that is transformed into fourier space for each measurement location; epsilon_pRepresenting a constant associated with errors caused by system noise. Different algorithms in the literature can be used to solve the above optimization problem (e.g. Fast Iterative Shrink Threshold Algorithm (FISTA), Alternating Direction Multiplier Method (ADMM) [7]]). Furthermore, adding a similar regularization function or using an unconstrained form will not alter the above-described benefits of the present invention.

Using the simulation model shown in fig. 4, this method was compared to a standard compressive sensing method at the same noise level. An object of N-3200 pixels is imaged separately by a standard compressive sensing calibration method using M-2560 calibration points and M-320 encoded calibration scenes. The image quality obtained by the standard compression sensing method is poor, and a high-quality image can be obtained by encoding a calibration scene.

In one embodiment, the random points represented by P may be selected from a domain that can be quickly transformed, such as an Hadamard matrix, in order to shorten the solution time for the problem given in the inequality. In this case, the P matrix may be expressed as a masked unitary transformation (masked unitary transformation). It has previously been shown that the optimization problem can be solved efficiently in cases involving masking unitary transform spaces [8 ]. In this way, the problem of solution time can be further reduced.

In actual operation, the time to switch between encoding calibration scenarios will be much longer than the measurement time of a single encoding calibration scenario. Thus, the total calibration duration will be determined by the total number of encoding calibration scenarios used and the time required to change (replace) the encoding calibration scenarios. To alleviate this problem, the present invention proposes a calibration scenario that is larger than the field of view in at least one direction. Rather than changing the calibration scene individually, the present invention moves the scene linearly in one or more directions and/or rotates the scene at one or more center points. Calibration measurements are made at certain locations during the continuous movement. The distribution of nanoparticles in the imaging field of view varies as a function of time. Thus, at different times, different portions of the calibration scene may appear in the field of view. In a preferred embodiment, the measurements are made during a continuous movement of the calibration scenario. This is possible when the signal-to-noise ratio is high due to the large number of nanoparticles used in the calibration scenario. Therefore, there is no need to repeat the measurement and take the average. In this way, the measurement time can be greatly shortened. As a result, the system can be frequently calibrated to obtain high image quality.

The position of the nanoparticle sample in the calibration scenario must be known accurately. The calibration scene may be generated using a high precision production method and/or may be measured using a high resolution imaging method such as X-ray imaging after the calibration scene is generated.

The calibration scenario may move linearly and/or rotationally. In an example embodiment, a spherical calibration scene is rotated about one axis and measurements are taken at K degree intervals. The position of the nanoparticle sample (8) in the calibration scenario varies as a function of the angle of rotation. For example, the nanoparticle positions of the spherical calibration scene (11) at 0 and 45 degrees are given in fig. 5 and 6, respectively. At each rotation angle, a new position of the nanoparticle in the field-of-view grid is calculated, as well as the nanoparticle density at each grid point in the new position. The error in this calculation depends on the accuracy of the rotation measurement of the rotating mechanism. If the accuracy is not sufficient, a highly sensitive position tracker (e.g., a laser tracker or similar purpose device) can be used to accurately measure the new position. In order to obtain a system matrix with high accuracy, the process can be repeated at a plurality (L) of different rotation centers (12) to increase the amount of measurement data. Therefore, the total number of measurements is M ═ (360/K) × L. Once these measurements are taken, the system matrix can be reconstructed by solving the optimization problem given in the inequality above. In the inequality, P is a matrix containing the nanoparticle density distribution in the field of view at each measurement location.

The calibration scene may be moved and/or rotated using a mechanism that translates and rotates the calibration scene about one axis (13). The mechanism (13) required for the rotational calibration of the scene can be designed as an integrated unit or as an external unit of the MPI system (1). An exemplary embodiment is shown in fig. 7. Here, the spherical calibration scene (11) rotates on one axis and slides on the other axis. In this way, calibration scenarios can be measured at different centers of rotation relative to the center of the field of view, and the diversity of calibration scenario measurements is increased. The linear sliding movement and the rotational movement can be continuously performed during the calibration process, compared to the step movement, thereby reducing the calibration time.

As shown in fig. 8, the mechanism for translating and rotating the calibration scene about the two axes (14) can also be designed to rotate about different axes of rotation. In this case, the conditional processing of the P-matrix autocorrelation can be improved, which helps in the solution of the optimization problem. In the implementation of the invention, the calibration scenario and the rotation mechanism can also be designed as external units according to the mechanical requirements of the MPI system, such an embodiment being illustrated in fig. 9 and 10. In fig. 9, a spherical calibration scenario (11) is shown in a top view, which is moved linearly by means of a reel system on a sliding rail (16) and rotationally about a rotational axis (17). Fig. 10 shows a side view of this calibration system. An auxiliary mechanical system (15) for translating and rotating the calibration scenario comprises the necessary equipment (motors, encoders, movement transfer elements and computer control interface) needed to perform the linear and rotational movements of the calibration scenario. In a preferred embodiment, the mechanical system comprises a control unit which communicates with the MPI system (1) to perform a calibration procedure using the calibration scenario. To this end, the control unit electronically receives the desired position of the calibration scene from the MPI system, moves the calibration scene to the desired position, and outputs position information of the calibration scene obtained from encoders in the mechanical system and/or tracking device that measure the position of the calibration scene.

The calibration scenario should allow for rapid filling (and emptying) of different nanoparticles. In one embodiment, the three-dimensional encoded calibration scene may be formed of a plurality of mechanically separable layers such that the nanoparticle sample may be altered. In another embodiment, a single layer calibration scenario may be used for two-dimensional calibration. It can be mechanically scanned in three dimensions to calibrate the three-dimensional field of view. Fig. 11 shows another embodiment. The calibration scenario comprises nanoparticle chambers interconnected by thin channels (18), and openings (19) for filling or discharging magnetic nanoparticles inside the calibration scenario. The calibration scenario is a hollow structure with one or more openings for filling or emptying nanoparticles from the hollow structure.

The nanoparticle samples present in the calibration scene are not necessarily placed in a single voxel (10) in a matched manner. The scene may include samples of nanoparticles of different sizes and shapes. For example, the nanoparticle sample can have any shape, such as spherical, elliptical, or rectangular prism, and cover a plurality of voxels. In one embodiment, a rod-like nanoparticle sample (20) as shown in fig. 12 is provided, taking into account the spherical scenario. The wand can be easily removed and inserted into the scene. The calibration scene may be generated in any shape, e.g. spherical, cylindrical, cubical, rectangular prism.

In a further embodiment, shown in fig. 13, the calibration scene (21) designed as a long rectangular prism is moved linearly only on the slide belt (22). Calibration measurements are taken at certain locations in the field of view. Fig. 13 also shows an optical reflector (23) and a laser tracker (24) to ensure accurate position measurement during movement. One or more reflectors may be connected to the calibration scene to track its movement.

In the embodiment shown in fig. 14 and 15, a cylindrical calibration scenario (25) is employed. Since the calibration scenario is relatively voluminous, the calibration can be performed with a smaller number of rotations than required for the calibration scenario presented in fig. 9 and 10. However, such a calibration scenario requires a larger opening, which may be suitable for an open-hole MPI system.

Fig. 16 shows an embodiment comprising a cylindrical cavity (26) for a nanoparticle sample, which can be rapidly filled and emptied.

Fig. 17 shows an embodiment comprising a tubule (29) in the form of a complex curve in three dimensions with a single input (27) for filling and an output (28) for expelling magnetic nanoparticles within a calibration scenario. The calibration scene may include a single or multiple tubes traversing any path of the calibration scene to fill or empty the tubes with or from nanoparticles.

Reference to the literature

[1] Weizencker J, Gleich B, Rahmer J, Dahnke H, borbert J (2009). Three-dimensional real-time in vivo magnetic particle imaging, Phys Med biol.2009; 54: L1-L10.

[2] Von gladis, m.graeser, p.szwargulski, t.knopp and t.m.buzug. mixed system calibration for multidimensional magnetic particle imaging, phys.med.biol., vol.62, No.9, pp.3392,2017.

[3] Compressed Sensing Theory and Applications (Compressed Sensing Theory and Applications), ed.by y.c.eldar, g.kutyniok, Cambridge University Press, New York,2012.

[4] Gleich and j.weizencker, (tomography using nonlinear response of magnetic particles) Tomographic imaging using the nonlinear response of magnetic particles, 435(7046) 1217-1217,2005.doi:10.1038/nature03808.

[5] T.knopp, j.rahmer, t.f.sattel, s.biederer, j.weizencker, b.gleich, j.borgert, and t.m.buzug. Weighted iterative reconstruction of magnetic particle imaging (Weighted iterative reconstruction for magnetic particle imaging), phys.med.biol., vol.55, No.6, pp.1577-1589, 2010 doi:10.1088/0031-9155/55/6/003.

[6] G.r.ace, d.j.brady, l.carin, h.arguelo, and d.s.kit, "compressed Coded Aperture Spectral Imaging," IEEE Signal Processing megazine, vol.31, No.1, pp.105-115,2014.

[7] Ilbey et al, "Comparison of System matrix-based and projection-based reconstruction for free-line Magnetic Particle Imaging" (systematic of system-matrix-based and projection-based reconstruction) "International Journal on Magnetic Particle Imaging, vol.3, No.1,2017.

[8]H.E Güven,A.and M.Cetin, "enhanced Lagrangian Method for Complex-Valued Compressed SAR Imaging" IEEE Trans. Complex. Imaging,2(3): 235- > 250,2016.